Dan Bulwinkle

Real Artists Ship

Wed, 01 Apr 2026 00:00:00 +0000

Version 0.96

Noah Wyle in Pirates of Silicon Valley, the only Steve Jobs movie worth watching

A couple years ago I was in Healdsburg, CA with a mission to find some art to decorate my house. There are several galleries in town, and I visited almost all of them. I chatted with the owner of the Harris gallery, which does have some nice art.

I kept my mind open when browsing, though I was also hoping someone had done a painting of a nice spot on a hiking trail, the Russian River Overlook. Unfortunately I couldn’t find any such painting, even online, and if I did find something it was a charicature of abstraction¹².

Russian River Overlook

Much of the artwork I saw wasn’t very good. That’s been a universal truth when browsing art for sale whether it be at a coffee shop or a gallery.

What is good art? In humanities class in high school our teacher told us to flip to the end of our text, Fleming’s Arts & Ideas, and write “What is art?” with the hope that at the end of the course we’d have a better idea. My definition was that if one person thinks it is art, it’s art. In Edmund Burke’s On Taste, he talks about taste being global and objective, and if your exposure is limited, how can you be a good judge? And, Ani Difranco sings, art is why I get up in the morning, my definition ends there.

One day, maybe 8 years ago, I was walking along 21st street a couple blocks from Dolores Park in San Francisco, and I saw a steel artwork fence above a garage. I took a photo and sent it to my uncle since it had all of the characteristics of his work. “I did that piece before you were born,” he replied. It was as if I had stumbled upon a Ghiberti or a Jenny Holzer.

Growing up on the East Coast, I hadn’t met my uncle until I visited a little more than 15 years ago. When I arrived at his studio on Hannah Street in Emeryville I was taken aback. During the 2008 downturn he had purchased two neighboring properties and they were both filled with steel sculptures. That isn’t to detract from the studio itself, which is an enormous, decorated steel hangar featured in the 2003 movie Hulk³. When I went inside there were no fewer surpises. For example, he opened a drawer and there were hundreds of original and print works.

I’ve had coffee with him a few times, and he’s very down to earth (as you can see in the video above). He was “discovered” in the early 90s, meaning galleries sought to display his work for sale. His work was featured in Basic Instinct, Microsoft’s art gallery, and he’s donated works to the Oakland museum. His website, itself an expression, makes one’s mind boggle at the extent of the works he’s produced and the detail of those works. This isn’t Jackson Pollock splashing paint on a canvas, he’s gracefully cutting steel⁴ with a hand torch! He has to source all of his supplies, organize it, cut it, weld it, clean up. How many pieces can you complete in a day? He hangs out and socializes at places like Jump N Java in Oakland and Roy’s of Amboy. Not to mention dealing with all other aspects of living. Idling must not be in his vocabulary.

People will always criticize art. I mentioned the Mona Lisa to a gallery owner who went out of her way to tell me how she thought it was awful. I don’t disagree that it is overhyped. At the Louvre, I scoffed at the Mona Lisa crowd, and turned to my right to see, hanging in solitude, Madonna of the Rocks⁵. In Healdsburg I criticized some gallery works as low effort, grade school level. I think few have the talent to produce anything like the Mona Lisa or a hand-crafted steel sculpture.

It seems even those who qualify as mediocre simply do their trade as a hobby. Hobbies are great; they often yield happiness. Whether visual artists, pianists, chefs, athletes, homebuilders, or software developers. They aren’t serious. They don’t challenge themselves. They lack passion. On the other hand, real artists ship. Artists love making art. That’s all they do, produce. It’s difficult to avoid greatness when you deliberately take an interest.

Coincidentally, today is Apple’s 50th birthday. Steve Wozniak has been announced as Apple’s incoming CEO effective this August. Congratulations, Steve!

That is, bad. ↩︎
I did find one I liked “View to the Sea” but when I emailed them they said, “Who are you?” and I sent a picture of my empty wall and they didn’t reply. I was going to give in and get a Sonoma landscape from someone whose work seemed a bit manufactured, but I couldn’t purchase that either because they were “offline” or unavailable. ↩︎
It seems they removed it in the streaming version since he sued them for displaying his work without consent. He used the proceeds from the settlement to buy tickets to see Hulk for the neighborhood kids. ↩︎
Even though he is known for steel, his passion is clay tile. ↩︎
Arguably way better from an art history perspective. ↩︎

Two Lessons in Education: Structure and Pace

Wed, 21 Jan 2026 00:00:00 +0000

Version 0.91

Two Computer Science Departments

In high school, I participated in a summer program at a university that is consistently tied for the best computer science department in the world. One of the courses I took was Introduction to Computer Science focusing on the C programming language, but at the end of the course there was an option to take the final exam for the version of the course CS majors take. The deal was, if you took that test you had to accept your grade. A handful of us studied for this while the rest of the students were enjoying a closing ceremony event. I earned an A.

The college I ultimately attended did not have a highly ranked computer science department. I thought I could transfer the credit. They declined. I asked an administrator directly: “no.” Finally, I had a meeting with the dean, and he told me to just take their Introduction to Computer Science course. I begrudgingly agreed.

He was right. It is one of the few schools teaching mainly in Lisp, and they teach it well. You’d learn recursion, tail recursion, and how to write code that had no state.¹ Brilliant. Despite this school not being competitive it was an incredible education in computer architecture, programming languages, compilers, formal verification, and research.² The top schools have a lot of smart students and professors learning and teaching in python³, but I’d take this education any day.

Two Introduction to Java Courses

The next lesson was during my tenure as an undergraduate assistant helping with an introduction to Java course. I did this for two semesters. For the first semester, the professor had students build a graphical game. These are not computer science students; there’s a high likelihood that this was their first time coding. Neither before then, nor since then, have I seen office hours packed with as many students as during that first week. It was hot, sweaty, and students were overflowing from the room. The professor had assigned them a task of drawing the game board or some unbelievable task of that nature.

The following semester had a different professor teaching introduction to Java. He started out with the basics, hello world, and receiving input from the console and manipulating it. Near the end graphics were introduced. Office hours were a free study period for me that semester.

There’s a bit of bias since some students dropped the first semester and enrolled in the second semester, but overwhelmingly the first group of students were better off. This makes sense. A psychological study known as the Yerkes-Dodson Law showed that there is a sweet spot for stress. Too much stress is bad. Too little stress is unproductive. You don’t benefit unless you have a little fire under you.⁴

My first job out of college was programming in Lisp. I really liked that job, but was disappointed that they used so many (for ()) loops. ↩︎
I had my name on a research paper or two. ↩︎
In any computer science program you are going to learn more than one language. We also did Java, C, 68000. ↩︎
If you find yourself procrastinating on some project, artificially add some stressors to force yourself to achieve your goals. ↩︎

Alpha Books

Mon, 15 Dec 2025 00:00:00 +0000

Version 0.96

Several years ago Chris Beiser asked people on Twitter what books no one else would recommend that give you “alpha.” I spent the time sifting through the Twitter thread, collecting the book recommendation into a spreadsheet. Later, the thread resurfaced, and I put them in CSV on Github at the request of others.

To the best of my recollection, the criteria were:

a) it could not be a mainstream book or one that many people would recommend, and

b) it had to be stunning enough to leave you with a strong intellectual advantage

In manually sorting through the Twitter thread I did my best to exclude books that were dismissed as being mainstream and include books that (paradoxically?) were affirmed by others.

I haven’t read them all, though I intend on reading as many as possible. Most of the ones I’ve read, I’d say were alpha indeed. Instead of recommending poorly written¹ books that may be insightful, I’ll recommend a few that were both alpha and well-written. Not a book report, just a few comments.

Down and Out in Paris and London by George Orwell. Incredible account of being without money in Paris. I read this along with Germinal by Zola which was roughly based on mining strikes in northern France 10 years before it was written. Both accounts of going hungry are eye opening.

The Ascent of Money by Niall Ferguson. I read this along with The Deficit Myth and Debt: The First 5,000 Years. Debt is a non-alpha book. Even though Ferguson and Graeber were both affiliated with the London School of Economics, Ferguson had a much more clear picture of how barter led to debt, how bonds (un)successfully funded wars, and he had interesting insights on how land is valuable and why South America struggled. The first 3rd of Debt could have been summed up in maybe 20 pages, but anthropology is Graeber’s specialty so of course it was detailed.

A Primate’s Memoir by Robert Sapolsky. I don’t know if it is the Hetch Hetchy, but somehow writers affiliated with Stanford are excellent. This is a sober account of baboon (and human!) behavior in Kenya. One of the things I appreciated about this book is how I was able to become more grounded with Uganda, Kenya, Tanzania, Sudan, Ethiopia, and Somalia. I’d like to see an alpha book thread that focuses on countries or regions so I can become better acquainted.

Alpha Book List

Title	Author	Summary
Thoughtful Interaction Design	Jonas Löwgren & Erik Stolterman	A foundational book on designing interactive systems with care and reflection. It argues that design is an act of inquiry, ethics, and aesthetics as much as engineering.
Down and Out in Paris and London	George Orwell	A semi-autobiographical account of Orwell’s time living in poverty. It exposes the harsh realities of class, labor, and dignity in early 20th-century Europe.
The Structure of Scientific Revolutions	Thomas S. Kuhn	Kuhn’s classic on how science progresses through paradigm shifts rather than linear accumulation of knowledge. It reshaped our understanding of scientific change and the sociology of knowledge.
If on a Winter’s Night a Traveler	Italo Calvino	A postmodern novel that turns reading into an adventure. Each chapter begins a new story, exploring the act of storytelling itself with wit and playfulness.
MITI and the Japanese Miracle	Chalmers Johnson	A deep study of Japan’s Ministry of International Trade and Industry and its role in guiding postwar industrial policy. Johnson shows how state-led capitalism fueled Japan’s economic rise.
How Asia Works	Joe Studwell	Examines why some East Asian economies succeeded while others stagnated. Studwell highlights the importance of land reform, industrial policy, and financial control in economic miracles.
Darwin Machines	Gary Cziko	Explores how Darwinian principles of variation and selection extend beyond biology. Cziko applies evolutionary thinking to learning, culture, and cognition.
Prometheus Rising	Robert Anton Wilson	A blend of psychology, mysticism, and social theory. Wilson presents a model of human consciousness inspired by Timothy Leary’s eight-circuit brain theory.
Quantum Psychology	Robert Anton Wilson	Explores how quantum theory and relativism can inform human perception and behavior. Wilson uses humor and exercises to challenge readers’ reality tunnels.
Conjectures and Refutations	Karl Popper	A cornerstone of philosophy of science, presenting Popper’s idea that scientific knowledge advances through falsification. He emphasizes open criticism and the tentative nature of all theories.
The Quark and the Jaguar	Murray Gell-Mann	Nobel laureate Gell-Mann explores the relationship between simplicity and complexity in nature. The book connects fundamental physics to the emergence of life and consciousness.
Ideology and Economic Reform under Deng Xiaoping	Ronald Coase & Ning Wang	Explains how pragmatic reforms transformed China’s economy after Mao. It focuses on ideology’s adaptive role in legitimizing market experimentation within socialism.
Against Method	Paul Feyerabend	A provocative critique of scientific rationalism. Feyerabend argues that there is no single scientific method and that progress often comes from breaking rules.
Changing Minds	Howard Gardner	Gardner, known for multiple intelligences theory, explores how and why people change their beliefs. The book offers frameworks for persuasion and education in complex societies.
A Primate’s Memoir	Robert Sapolsky	A biologist’s witty and moving account of years spent studying baboons in Africa. Sapolsky blends field science with reflections on stress, society, and human nature.
The Origin of Consciousness in the Breakdown of the Bicameral Mind	Julian Jaynes	Jaynes proposes that early humans experienced thought as external voices before developing introspective consciousness. A bold and controversial theory of mind’s evolution.
Life Itself	Robert Rosen	A theoretical biologist’s attempt to define what distinguishes living systems from machines. Rosen challenges reductionist science and presents a relational view of life.
A Confession	Leo Tolstoy	A spiritual autobiography chronicling Tolstoy’s crisis of meaning and search for faith. It captures his turn from literary fame to moral and religious inquiry.
Maya: A Novel	Jostein Gaarder	A philosophical novel blending love story and reflection on evolution and consciousness. Gaarder weaves existential questions into a poetic narrative.
African Folktales	Paul Radin	A collection of traditional African stories highlighting moral lessons, wit, and cultural heritage. Radin’s anthropological framing preserves the oral tradition’s richness.
Tales of the Southeastern Indians	John R. Swanton	An important compilation of Native American myths and legends from the Southeastern U.S. It captures creation stories, trickster tales, and tribal wisdom.
Japanese Tales	Royall Tyler (Translator)	A vivid collection of medieval Japanese stories mixing humor, ghosts, and morality. Tyler’s translation brings classical folklore to modern readers with elegance.
Protocol: How Control Exists After Decentralization	Alexander R. Galloway	An exploration of power in digital networks. Galloway argues that protocol—technical standards—functions as a new form of control in decentralized systems.
Breakdown of Will	George Ainslie	A psychological study of self-control and procrastination. Ainslie presents “picoeconomics,” showing how short-term choices compete within the self.
The Man Without Qualities	Robert Musil	A monumental modernist novel set in pre–World War I Vienna. Musil dissects reason, emotion, and the search for meaning in a crumbling world.
Capitalism and Desire	Todd McGowan	A psychoanalytic critique of capitalism’s logic of endless desire. McGowan uses Lacanian theory to explain why consumer society thrives on dissatisfaction.
The Trap	Sir James Goldsmith	A critique of globalization and unchecked market forces. Goldsmith warns that free trade and growth ideology threaten society’s social fabric and environment.
The Ascent of Money	Niall Ferguson	A sweeping history of finance from ancient money to modern markets. Ferguson shows how financial innovation shaped human progress and power.
Essence of Decision	Graham Allison	A classic study of decision-making in crises, centered on the Cuban Missile Crisis. Allison introduces three models—rational actor, organizational, and bureaucratic—to explain political behavior.
To Live and Think Like Pigs	Gilles Châtelet	A fiery philosophical critique of neoliberalism and mediocrity. Châtelet rails against a society that trades imagination for comfort and conformity.
The Tale of Genji	Murasaki Shikibu	Often called the world’s first novel, this Heian-era masterpiece explores love, beauty, and impermanence at the imperial court. Its psychological depth remains timeless.

For example, books which are written in Harvard-ese or books that have a seemingly unedited information dump. ↩︎

The Pool of CEOs is Bigger Than You Think

Mon, 01 Dec 2025 00:00:00 +0000

Version 0.95

This evening, Travis Kalanik put out a call for tips to find the right “re-founding” CEO for a company he purchasedcalled eXtremeGammon. People were quick to point out how this was just like January of 2010 when Kalanik solicited tips on a CEO for Uber, and Ryan Graves said he had a tip: himself. Graves went on to be Uber’s early CEO growing the company before voluntarily stepping aside and Kalanik taking the helm.

Recently there has been talk of Tim Cook tapping a longtime Apple hardware exec as a successor. Cook himself was COO when Steve Jobs bestowed the honor of leading the company. Most of the big tech companies have had successors who have come from inside the company. This makes sense because you want someone who rubs shoulders with the existing CEO and has a good understanding of the culture. Outsiders are risky as we’ve seen when Steve Jobs selected John Sculley from Pepsi to run Apple back in the 80s.

But, as I’ve said on this blog before, the job of the CEO isn’t just to grow revenue. Cash flow is an extremely important job, of course. I’d argue that more important is vision.¹ Few people have it, and certainly if your job is running hardware engineering or managing manufacturing operations, one wonders why suddenly vision is in your cards.

I’ll suggest something wild. If you are looking for a successor to a strong company, do hire internally; however, hire people based not only on merit but potential for vision. Hire them pretty much exclusively for the purpose of succession two or three years down the line when the current CEO is planning to retire or exit. There exists a whole world of people, and to believe that the right successor is in a small pool already in the company is myopic. Those with vision may be like the Swiss patent clerk or young prodigy in India just waiting to be discovered.

Obviously in the context of a cash cow business. If you have no revenue vision could be delusion. ↩︎

No Longer an Amazon.com Customer

Sun, 23 Nov 2025 00:00:00 +0000

Version 0.95

A few weeks ago I eliminated my Amazon account. It is the last of the MAMANA (formerly FANG) services to go.

I realized that I’m not only an early adopter of technology but an early adopter of banning companies with bad policy. The problem is, most people have no idea about these policies and even fewer care. You’d probably have to hand-hold and explain how you are tracked by iOS apps even if you have the “Don’t Track Me, Bro” button toggled to elicit a reaction of some kind. Even then, people will point to futility: “cameras are everywhere” or “I gave up (on privacy|on life).”

What I can’t quite understand is why companies, or the faces of them, are no longer diplomatic. If you behave eccentrically, should you expect people to keep buying your cars? Yesterday it was just a car, and now it may be an extreme political symbol? GTFO. If you support some controversial position like mass surveillance, when it goes wrong, like Salt Typhoon but worse, how might you respond to shareholders as customers distance themselves? It’s ludicrous. Build an excellent product. Sell, sell, sell. People love it and buy more, more, more. People tell their friends, and they buy more, more, more. It’s that simple. Don’t complicate it with bad policy.

After I left Amazon, I noticed a couple of things. Virtually anything I could get from Amazon I could get elsewhere. eBay works very well. While reading Woolf’s To the Lighthouse¹ she mentions (presumably) Walter Scott, so I looked up Ivanhoe which I hadn’t read.

On Amazon, Ivanhoe was $6.51 and on eBay it was $4.97. I’d also have to spend an additional $30 on Amazon to get free shipping. This eBay seller had no minimum and free shipping to boot. eBay would not only save me more than 20%, but obnoxious cognitive overload trying to decide what else I “needed.”

Being without Amazon isn’t perfect. I needed a Zigbee controller for my electronic window film. On eBay, it was over $20. On Alibaba under $6, but coming from China.² Amazon seems to warehouse Chinese goods so you can get it fast. It was only $10 there.

As more people are educated, they may see they’ve been operating like zombies in BigCo’s ecosystem. Deleting my Amazon account gave me pause. But if you have self-respect you’ll live deliberately.

Definitely a love-hate relationship with this book. I want to know if Janeane Garofalo spent her youth perusing Woolf. ↩︎
Not only the wait, but dealing with tariffs. Tariffs have been easy; I’ll do it only if I really want something. ↩︎

Building a High Performance Home

Sat, 08 Nov 2025 00:00:00 +0000

Version 0.88

Double stud walls provide a thermal break between studs.

In 2018, I set out to build a high-performance home, unfeasible in the Bay Area, so a couple years later I moved to the East Coast to find land.

Finding a House

The problem with living in desirable places is it can be difficult to acquire land. I demanded walkability, further narrowing the market. If the acquisition target is a tear-down, it will be as appealingto the local builders as it is to you. In the Boston area, I spent a considerable amount of time writing letters to people with extra land or who mightbe close to selling their, sorry to say, tear-down¹. It was looking bleak in 2021 and 2022, so I moved down the coast when I found a house that was inthe middle of everything, walkable, and had parks, restaurants, and coffee shops.

The first step was to meet with the code official to see if what I wanted to do possible. They speculated it was, but that I should have a surveyingcontingency. Fortunately at that time the market looked like it was on the way down so the seller agreed. It was possible to build the rectangle Iwanted so the sale proceeded.

Finding an Architect and Builder

A lot of the legwork for this house started ahead of time. Once I got serious about acquisition I already had a list of resources to get the process going. The problem is that without land you can’t plan. I discovered on the DOE website a list of zero energy home builders. From that I contacted two and one was interested. Common advice is to get an architect reference from a builder or a builder reference from an architect you like. The architects that the builder recommended were good, but I didn’t have 6 months to sit around waiting for plans. I read that 4-8 weeks isn’t impossible, so that was my goal. Around this time the builder gave me a gift: Pretty Good House, a book about building homes that approach passive house standards. I contacted one of the authors, Emily Mottram, and asked if she could be the architect. She couldn’t, but she was able to consult, which was fantastic. One of the builder’s recommendations also agreed to consult. To solve the architect problem, I hired a drafter.

Before I hired a drafter, I had the builder vet him. This is where problems started. The builder’s team was the owner of the company, the project manager, and coordinator among some otherpeople who I didn’t interact with. The PM would actually be on site working on the project, and the owner wouldn’t have as much of a role in day-to-day activity, apparently. The owner calledthe drafter and had a conversation. The drafter was a former builder and he turned out to be great; however, the owner and the PM had diverging requirements. I later learned that the PMwanted very detailed building plans. So I went on a gig site and tried to find someone to build out the plans. I found one or two candidates who gave me plans they had designed, and the PMagreed that it was the level of detail they were looking for. Months had already gone by at this point. I told the builder I wanted to startMarch 1, and they said there’s no reason it couldn’t happen. Only there was a reason: we had no contract. I asked for a contract but they wouldn’t give me one until the plans had been secured. Since it wasgetting into April, I decided to look for another builder. To be fair, March wasn’t possible because the town had some red tape I had to get through to proceed, but we could have started mid-April ifthere had been a contract.

The plans had been finalized with Emily’s input on performance. The other consulting architect added valuable aesthetic detail like lighting. For HVAC, I hired Corbett Lunsford, the YouTuber,to design the Manual J. A Manual J is a calculation of heating and cooling requirements based on insulation and heat gain from windows, people, and appliances. I made a list of builders andemailed them asking if they could build a high performance house based on the attached plans. Some didn’t respond, some said it wasn’t for them, and I had two that were interested. One of themwasn’t interested in working remotely (I was still in Boston), so that left one builder. During the phone call with the owner I mentioned Corbett and how he worked on the Manual J. The builderresponded, “I think I know what you mean.” Now, I was in a position of desperation as I had been through much more than you care to read in order to finally be able to build a house. One of the firstlessons learned is never, ever be desperate. If a builder doesn’t know what a Manual J is, run². In fact, you should be very knowledgeable. There’s a ton to learn in construction. Watch MattRisinger, Stud Pack, Corbett. Do not join a cult, but take an intellectual approach. Emily wasn’t familiar with a particular WRB (weather barrier that is attached to the plywood before siding goes on), butshe advised against using Risinger’s favorite Zip System because it has a permeability factor of 1 which means water vapor can’t escape and therefore the cellulose insulation could harbor mold. There’s nuances,but not knowing Manual J after being in construction for 25 years is a huge red flag. I hired them.

Excavation and Foundation

It took the builder around a month to put together a contract. When I was at the code official’s office he told me I could be my own builder. In retrospect I would have done this if I realized just how much moreefficient I’d be without this builder. I could have lit duffle bags full of cash on fire and still have come out ahead. It seemed to be a win that the builder had construction scheduled just a couple months out.The site work required engineered plans. When they started work, I got an email from the code official saying they failed a site requirement. Even I noticed this on the plan, and I asked the code official if itwas a red flag and he told me that some construction workers just aren’t used to it. Nevertheless, the fact that they ignored engineered plans gave me pause. The code official ended up being less helpful thanI imagined, but critical in some important regards. After excavation he requested the builder to have a professional engineer evaluate the soil and write a recommendation. I was already required to have the soil testedand knew where the water table was and the makeup of the soil at various levels, but requiring a civil engineer to write a proposal meant I was less likely to end up with a Millennium Tower situation.

One mistake that I made was using Superior Walls, which are prefabricated dense concrete (waterproof) that are also highly insulated. My plans originally called for insulated concrete forms. These are forms made out of EPS foam into which concrete is poured as opposed to steel forms which get removedand reused. The builder often uses Superior Walls so they recommended it and said it would save a substantial amount of money. I wanted to have an attached greenhouse eventually, and the basement egress window wouldhave exited into the greenhouse. The code official said that wasn’t ok. The Superior Walls had already been cast at that point. I had to pay a reasonable fee for the window to be closed in and a new one to be cut, thoughI wasn’t told about this until much later.

Framing

Superior Walls was probably a poor choice for many reasons. Because of their nature, you have to build the first floor deck before you pour the concrete, and when you pour the concrete there have to be 2x4s around the perimeter.Emily was dismissive of the idea that we insulate the foundation, which I wanted to do with EPS foam. The reason being that the 2x4 perimeter would yield a path to the earth temperature which can be in the mid-60s (18-19C). Inoticed a buckle in the deck, told the builder, and the framers fixed it. This was the beginning of my question: who’s the builder?

After the slab was poured the framing continued. My plans called for double stud walls about 10" thick. There are two advantages of double stud walls: a lot of insulation and elimination of thermal bridging. Most walls today are constructed with 2x6 studs and have an insulating factor in the 20s. Studs have an insulating value of about R-2, so they provide no thermal break.

One thing nearly all houses do not have is a seal between the foundation and the rim joist. I spent $500 on Fentrim tape made by Swiss company Siga. The framers went around the entire base and sealed the gap. This is a must if you want an air tight house. Matt Risinger also adds a layer of acoustic caulk between the foundation and sill plate to prevent air and bugs.

For the roof structure, I used 12" raised-heel trusses. That’s because normal trusses prevent insulation at the joint, while raising the joint 12" allows for that much insulation making a totally insulated box. I was away when those went up, but I was happy to learn that the framers put the WRB on properly: in order to have an air tight house you need to have continuity from the exterior WRB across the ceiling line, connecting the WRB to a permeable plastic. Steve Baczek, the New England architect, shows it on drawings by taking his pen, Big Red, and making a complete connection from one point all around the house and back to that point without lifting the pen. If you lift the pen, there’s a leak in the air barrier.

Windows

Passive houses tend to have few windows placed strategically to take advantage of southern sun for heat gain in the winter. In order to find suitable windows I checked the Passive House Institute US website for PHIUS certified windows. Zola has some high end European style windows and they were on the list. They had a minimum order amount and it was beyond what I’d consider paying. Pella was also on the list. I knew the name from Risinger’s channel but didn’t realize they had PHIUS standard windows. A nice thing about Pella is you can estimate window cost using their online tool, and it yielded a reasonable sum.

Unfortunately you still have to deal with sales, though this was one of the better sales people I dealt with. When planning a house, if you are all in, it is impossible to be excellent with every detail. I assumed because I was purchasing triple pane PHIUS certified windows they’d be essentially soundproof. After the house was done I asked what the STC rating was. STC is typically between 26 for basic windows and 37 at the more sound proof end. Zola, who I inquired about before Pella, has windows upwards of 40, but they won’t engage unless you have an order of over $50k. Pella told me their STC was 29, pathetic.

Pella is not recommended. I emailed the CEO about the lack of upsell. I would have bought a thicker outer pane which is what’s required for a better STC rating. I think 37 is probably the minimum.

Pella couldn’t accommodate a triple pane basement egress, so I decided to find a tilt turn. It was double pane, but made in Poland.³ I very much like tilt turns and also told Pella they should adopt them. The CEO made no remark so I don’t think we will see anything promising from Pella.

Exterior Doors

Risinger demonstrated a triple locking door to his garage, so I decided to install Thermatru triple locking doors. Triple locks have a mechanism connecting latches at the top and bottom of the door. When engaged, the two latches push the door tighter against the rubber seal so the door is more airtight. Thermatru doors are also insulated rather than a solid slab.

I don’t recall whether the builder or framer initially installed the doors. While waiting for the trades to come in, it rained hard many days in the autumn and early winter. There’d be puddles of water on the inside of the house, even after the roof was up. When I brought this up the builder and code official said it was normal until the siding went up. I’ve watched a lot of videos and I’ve never seen this after the WRB is up.⁴ I started looking at it in greater detail.

The house has a rainscreen for two reasons. One for its intended purpose to allow rain that gets behind the siding to escape and dry out. The other is because the double stud wall builds up moisture in the winter and needs to dry out via WRB vapor permeability. The builder initially wanted to use hydrogap which has 1mm raised bumps allowing moisture to escape. I saw some multi-million dollar homes being built with this, and what I noticed is it was wrinkled in several places, which leads me to believe the moisture would get trapped. But also it didn’t solve the second problem of allowing vapor to escape. We used furring strips. For some reason, the builder and framer put 8" wide furring strips around the windows and doors before they put the flashing tape on. This meant that water was getting behind the furring strips and into the windows and doors. Just look up a YouTube video. You pay for inexperience with time.

So they had to put on enough tape to cover the wide planks around the doors and windows. It looked ridiculous and wasted a lot of tape. Even after that there was still water coming in the door. Since I haven’t studied exterior door installation in depth I’m still not sure what the issue was, but I’m also not convinced it was totally solved. They had one of the trades cut a drip pan for under the door. That didn’t work so I asked them to just buy a prefab drip pan so any water that collected under the door would be ushered outside.

HVAC

HVAC is the heart of a high performance home. It’s why I hired Corbett who is an HVAC nerd. Like the builder, the HVAC company was accommodating. They fulfilled all requests. When I did the initial walkthrough I naively thought the two main guys were going to install it, but it was some other younger group of people. I told the two main guys, for example, that I wanted the exhaust of the ERV in one location and the intake in another, but it was backwards when installed.

At rough-in they only installed the main ducts and pathways for return, but they also delivered the indoor air handler which ended up sitting a basement that would flood periodically. One mistake I made, maybe, was informing the HVAC installer that one of the ERV ducts was dented (due to a knot in the wood) and they snapped at me over email saying their installers wouldn’t do that (it was before Thanksgiving so maybe influenced by other factors). I should have just told the builder, who seems to frequently blame any problem on “the homeowner” to save face. I wonder if I were my own builder how would I handle it, but I realize that some people are just that way. “It was an act of God that bent the duct!” Incidentally I sent photos, and it should have been clear that it was due to a knot, which I don’t blame anyone for because knots are a pain. And I ran it by Corbett before asking. One truth about ducts is that to get good airflow you want to have as few turns as possible, no more than 1 or maybe 2.

The installer was supposed to come back before the flooring was installed, but of course the builder failed to schedule it. At the time the floor was installed only the heat pump was operating, not the rest of the HVAC equipment which I had called for. I asked about the whole home dehumifidier and got “Oh, yeah.” They seemed to install it as an afterthought even though it was in the contract. Same with the HEPA filter. For that, they had to replace a whole section of the return duct.

Plumbing

I ordered the plumbing fixtures myself, and before the plumber arrived I was told to order “the bathroom fixtures.” Managing this project through software would have been advantageous many times, and this is a perfect instance of why. I ordered the toilets and tubs and all of them arrived.

The plumbers did a walk-through like HVAC, and I indicated I wanted to have copper pipes coming from the water line into the kitchen, which wasn’t far. Otherwise the whole house would be PEX, which is standard today. I also explained we were putting in a water softener, whole home water purifier (like a large Brita filter), and a fun reverse osmosis setup. I looked into reverse osmosis systems and the old school systems have tanks. There are newer systems that produce instant water at pressure needed for appliances like refrigerators. My plan was to have the reverse osmosis system feed into the refrigerator line and pot filler line.

The plumber came back and told me he asked his boss and said the copper line would deteriorate due to the reverse osmosis system. I explained that the reverseosmosis system would have a mineral replenishing device attached so the pH level would be sufficient. But, I also reasoned that even though PEX doesn’t use plasticizers, even if plastic particles were to get in the system the reverse osmosis system would take care of it, and also reverse osmosis itself is full of plastic components (you can either have PFAS or microplastics, I said).

For a hot water heater I installed a heat pump electric hybrid. For an outdoor water spigot I went with Aquor which has a quick-connect attachment to the hose so you can easily connect and disconnect it.

The toilets and shower fixtures were mostly known brands, but certain things like the tub filler and the kitchen faucet were not because the cost difference was so great.

Back to rough-in. I went over one day to see the tubs in the bathrooms, though uninstalled. It turns out that the plumbers had planned to install their own tubs. This ended up being a big issue for various logistical reasons, but in the end it was resolved. How much time is wasted and how many resources are used due to flubs like this?

Everything worked out with plumbing, except there is a knocking noise at one of the drain pipes behind the wall when hot water runs through it. The inexpensive fixtures worked out, and the reverse osmosis system works well.

Electrical

Even though the builder had difficulty going outside their usual scope and learning what it meant to build this kind of house, the subcontractors I would say were solid. Maybe 70% or more of them were excellent. The electrician was accommodating and they didn’t have to be, so I appreciated working with them.

I came across a YouTube video about whole home DC lighting. Normally lighting systems run thick AC wires and convert at the ballast for DC LED lighting. ATX LED had a whole home system that converted AC to DC at one location and ran low voltage wiring throughout the house. Additionally, they had USB-C and USB-A outlets which is something I’ve wanted for a long time (you ever notice how many wall warts there are?). This was super exciting, but it depended on the electrician, and luckily they agreed to do it.

Matt Risinger, based in Austin, had a video where viewers asked why he has Romex for his lighting instead of low voltage. There are times when I could take a hint but failed to do so. I asked Murray, founder of ATXLED, why Risinger wasn’t a client and his response was something about money for exposure. But if Risinger wanted the product, why wouldn’t he just install it and not advertise it? Murray and I worked on designing the low voltage system based on the plans. In addition to the lights I had close to a dozen low voltage outlets and also spent extra on an automation system.

The rough in for the electrical went slowly. It ended up taking months longer, and I don’t know if it had to do with the builder trying to figure out the basement water issue. In retrospect, it should have been straight forward but in the end the electrician told me there was a lot of confusion and a lot of calls to Murray. The low voltage system is about 6W per light with 100W per hookup. You connect a 5 wire low voltage wire to the port and run it to a switch. Each switch also has a daisy-chain port so you can connect the low voltage to another switch (each switch has a microcontroller). Two of the wires are power, one ground, and two are data.

Ready for books.

Interlude: A Firing?

The day before Christmas I went into the house to install an outlet so that the solar installer could get the system operating before the end of the year and we could run the sump pump using solar. I went into the basement and it was completely filled with sewer gas. No one could work in that environment. I set up the outlet box upstairs and then ran down a few times holding my breath to install it. The builder had duct taped over the sewer pipe and someone partially ripped the duct tape off.

I went to the mountains later that week and decided I should see if the original builder would be interested in taking over. The current contract didn’t look like it was written by a high priced lawyer and had language like, “if we can’t agree, we will amicably part ways.” The original builder agreed to be on site in the new year to look at the project.

We toured the house, and I sent an approximate remainder of the budget. I hadn’t heard back for a couple weeks, then the project lead was on vacation in the tropics, then their work shop burnt down via the next door business. If things had gone more smoothly and they said they could do it for the amount quoted, I would have fired the current builder. One holdup was that the water in the basement hadn’t been addressed yet. In retrospect, I should have gathered estimates from sub-contractors myself, paid the original builder an advisor fee, and fired the current builder.

Insulation

Early March, they finally got the rough-in inspection. This inspection is for everything visible including framing and plumbing so insulation could go forward. One anti-feature of the builder is he was never there for inspection. He always doled it out to an underling. But, the inspector told me this house was well-built and the same couldn’t be said for many of the houses he reviewed. Just from driving by new construction I believed him. You see houses with janky framing, OSB instead of plywood, or wrinkled Hydrogap instead of fully adhered WRB – and these are $3 million houses whose unsuspecting future owners merely see the Dekton or flush baseboards.

Mid-March the insulators were coming, so I took videos and photos of everything on each level inside the walls and ceiling. This proved useful multiple times both before and after the house was finished. As the insulators arrived they started spray foaming. I was confused, because dense-pack cellulose does not require spray foam. Fortunately, the builder got themselves a stop-work order posted on the door threatening arrest, and when I told the insulators they took off. This gave me time to talk with the builder and code official.

Next, the insulator came back and started adding the cellulose. Only, they were dumping it in the wall cavity like Randy Quaid hosing excrement into the storm sewer in National Lampoons Christmas. I sort of lost it, because this was one of the most critical parts of the house. Had I waited 5 minutes the builder would have been there and I could have channeled my frustration at him. Instead, imagine telling a professional they are doing it wrong. (They didn’t like it.) I ripped off the mesh keeping the cellulose in place on one of the “completed” cavities and it fell out of the wall. Dense-pack doesn’t do that. Plus, there was a baseball sized gap. A poor job, indeed.

Despite showing the builder YouTube videos on how to do it, the insulator ended up packing the walls better but they just didn’t apply the right technique. Later on, there was more drama with the attic insulation.

Since I couldn’t trust the insulators, I went into the attic and marked the height which represented R-60 with a sharpie on a handful of trusses. Sure enough, after they insulated the level was so low in some places that it didn’t even meet code. The builder went up and looked toward the middle of the house and after several minutes said, “Yeah, we’ll have to get them back here. We’ll take care of it.” The second time they came back, apparently they didn’t tell the builder. One day the builder looks up in the attic and says with a very poor performance, “Oh. Oh! It looks like the were here already.” I measured and it was off by inches of insulation. That can have a real cost over years of heating and cooling.

I told the builder he should have someone from the company check the work of the insulator in real time. I said this in front of one of the employees who volunteered. The third round was scheduled for a Friday morning. The employee kept asking the builder about it who gave shady responses, as if he wanted the employee to quit talking about it. The insulator didn’t come that day. Instead, they came some day I wasn’t available. I asked if he verified the level of insulation (within some margin is fine), and the response I got wasn’t, “It’s within reason,” but “They blew in everything they had!”

At some point around this time I learned from Stud Pack that there is a Greenguard Gold spray foam insulation. Had I known that, I would have gone with it. It’s kind of difficult to fake spray foam. One major reason the Pretty Good House people like cellulose is that it is recycled and therefore “green;” however, it is not green if the installer commits fraud and you end up spending more on energy in the coming years.

Siding

One thing I found odd was that both builders I considered asked about siding and insulation as if it would break the bank. The only viable options in my mind were Hardie and SmartSide.

You may notice that high end homes tend not to have seams, at least in the front of the house. This is because they design the house to accommodate the max length of the siding so they do not need more than one plank on the horizontal. This mattered to me a lot originally and so I chose SmartSide which has a slightly longer plank. When the design came in, I decided it wasn’t worth it over the function of performance home principles and went back to Hardie. SmartSide is easier to cut but it requires a gap between planks because it is engineered wood. Hardie doesn’t require the same gap and is the best for fire resistance.

The WRB has 150 days of exposure to sunlight allowed with no guidance after that. The siding went up right around the deadline, except for one side of the house. The builder should have known that the WRB would expire and at least put a tarp over it. I pointed out that the one side was no longer guaranteed to do its job so I asked that they remove it and replace it. The WRB is extremely sticky and doesn’t want to come off. The company rep on site started to remove it, but when I came back later I noticed he just put another layer on top of the old layer! You can’t do that because the cellulose insulation needs to breath. I’m still not sure if the rep did it on his own or not, but I asked the builder to remove both layers and install a fresh layer of WRB. It took multiple guys several hours but they eventually did it. They told me the rep was fired not just for performance on this project but for other issues as well.

Greenhouse

One feature I wasn’t sure about installing right away was a greenhouse. BC Greenhouse had been selling greenhouse kits for decades so I got a quote and it was reasonable for the size. It was engineered for 120 mile an hour winds, a formidable structure. It was also accessible directly from the house via vestibule. The builder flipflopped on building it but eventually subcontracted it out. The instructions weren’t great and there were missing parts but they were able to sort it out. Definitely glad I did this at the time or it may never have been built.

Drywall, Doors, Flooring

The drywall was a disaster. In general it looks fine, but it wasn’t finished anywhere near flush with the wall, something I’ve never seen before. I asked the electrician about it who said they’re used to it while installing box extenders to make up for the gaps between the outlet boxes and drywall. The backsplash had to be recalculated. And the cabinets wouldn’t fit in the kitchen because the estimate of the thickness of the drywall was double.

One day the code official stopped by to check the drywall. I watched him tell the company rep on site (the same guy who was fired following the WRB incident) that the drywall was missing some screws but otherwise it looked good. The rep said “OK” and continued his unrelated work. He never told the drywall installers anything.

I noticed the finishers were not installing corner bead around the windows. This gives the modern drywall return a nice clean look. While it wasn’t in the contract the builder and I discussed it, and I told him we talked about it pointing to the spot where we had the conversation. They were planning to just put caulk on trimmed drywall which would have looked bad, so when I insisted on the corner bead they charged me extra for it. This was one of the few disputes, so I looked into how to handle it. When you have a dispute like this, you can propose meeting them half way. I went to their office to discuss it and they agreed to cut the cost in half. Another builder might point out that it wasn’t in the contract, so definitely have such things spelled out in the contract.

When attending open houses I could spot solid wood floor work and shoddy work. I vowed to make sure the wood floor was well done. There were multiple factors at this point, particularly lack of time and also the estimate from the builder for the floor was incorrect. I found a floor installer myself who I didn’t vet and this was a bad idea, though a worse idea was hiring this builder in the first place. The builder and installer met and we decided on a date for installation. The installer was clear that the wood needed to be conditioned a week ahead of time. Of course the builder did not have the HVAC ready in time so it had to be pushed back, which of course upset the installer as they’d lose work.

It was rescheduled and the wood was delivered; however, only the heat pump was running and the whole home dehumidifier hadn’t been installed yet. So the wood was sitting in sort-of conditioned space for a week. Further, the planks are wider than usual, and that means the tolerance was low and they should have been sitting for slightly longer than usual. I didn’t learn this until later. At installation the floor looked incredible, but with a stable dehumidification system the planks shrunk. Legal consultation says: use your builder’s risk insurance, but because of deadlines that wasn’t feasible. In retrospect, since the installer was actually a middleman I should have just paid him a finder’s fee and hired the installation company directly. But also this wouldn’t have happened if the flooring had been installed on time in spring rather than the sweaty humid summer.

A couple of fun stories regarding tile. The builder didn’t have the water turned on so the tile installers found some stagnant dirty water and used that to mix their cement. For the shower, I wanted a curbless so I had them install a drain in the center of the shower. For some reason the builder moved the drain to the wall, which doesn’t make any sense because either you’d need a curb for the slope or the wood floor wouldn’t be level with the tile floor. This was discovered and reversed so it sloped in the opposite direction. I assumed the tile installer would install the drain correctly for this setup (maybe I was tired at this point) but of course they didn’t. I have a temporary solution but eventually that will need to be redone.

Lastly, the doors. I’ve never seen a door with a halo before. I don’t know if the builder installed the interior doors themselves but I can only assume since they installed the exterior doors. Every single door has some kind of issue from a strange crunch when closing it, divergent gaps along the hinges, halos. Of course they used the thinnest door possible to save money. Had I researched interior doors I may have my luxury builder’s license by now. One other thing is that I paid extra for solid wood doors. One day I noticed one of the doors was made from MDF. I looked around and there were about 6 of them. “I don’t know how that happened” the builder says. Keep an eye on your builder and subs, especially for things that you would never know otherwise. Once the doors had the hardware installed it’d be more difficult to notice.

Cabinets and Finishes

The cabinets started to go in right before I traveled for several days. You can buy decent cabinets online, so I had pre-built cabinets shipped. They arrived about a month and a half early sitting in a humid garage. When the cabinet installer came he told me “this is my wheelhouse” which got me excited. As he started installing the cabinets he realized they wouldn’t be centered and wouldn’t fit in the space, partly due to the fact that the drywall was over 1" total thickness between the two walls. The builder told him to wait until he arrived, which was hours later. A ridiculous waste of money – there was an auxiliary cabinet setup that the installer could have worked on. Days later I walked in on one of the builder’s employees installing the auxiliary cabinets, without finish screws! Since I was away I don’t know who ended up installing the cabinets but it largely worked out. If you have sparse cabinets you can DIY, but if you have a complicated cabinet setup you probably want to have a reputable pro design and build it.

The electrician came through and tried to get the DC lighting system working. Because of the builder’s delays it had been months since they worked with the system so they probably forgot how it is supposed to work. Murray from ATX LED has a “How to Wire” guide and it should have been clear, but I was later told that they lost money because it was too complicated. I installed a lot of low voltage wire for art lights, bookshelf lighting, and future projects.

I learned that paint, even if greenguard gold, can have PFAS. I opted for a latex paint that supposedly had no offgassing nor PFAS. One day the painter came back to do touch-up paint and finish an area that hadn’t been painted. He told me something like, “This [paint] is better. That other stuff doesn’t have volume.” I can only take that to mean he used a different kind of paint for that area. I think the builder not only lies to the subs, but also to his own employees, putting words in my mouth. The lying throughout the project from various people made me sad for humanity.

Deck, Driveway, Final Grading,

The deck, driveway, and final grading were all ridden with issues of course. The builder installed the deck themselves, and I’ve never seen such a horrendous job. If they had watched just one YouTube video on installing composite decks on the ride over, the deck might have been a “B” but instead I don’t even know how to grade it, I’ve never seen something so bad. They didn’t put spacers between the boards, nor filler in the screw holes, they used the wrong size post, and they cut the composite material as if it were lumber, which isn’t how you do it.

For final grading I saw the excavator trucking out what looked like native soil and trucking in rocky soil. I asked him about it and of course I received a likely untruthful answer. I realized what was going on: the builder was worried about water penetration and wanted to make the soil as impervious as possible. Only now having a layer of rocksoil instead of topsoil makes any landscaping more difficult in addition to preventing even growth.

Technology

Even though I’ve developed bluetooth on NRF52 and iOS, I hadn’t kept up with low energy systems. In Mountain View in 2015 there was an event sponsored by Intel that reviewed the contenders for the space. I recalled some of them now that I was researching what to use: Zigbee, Thread, Z-Wave. Choosing just one seemed to have advantages like stronger network and less potential for interference. Most products appeared to support either Z-Wave or Zigbee. Zigbee was an open protocol design so I decided on that.

During the doldrums of construction I bought some essential items and tested them out with a Zigbee controller to make sure they worked. I almost made a mistake buying a battery-powered deadbolt. The deadbolt would have hit the triple locking mechanism, so I returned it. Instead, I was inspired by someone who made a custom lock for his home office. I bought a fingerprint reader (with future plans for facial recognition) and a solenoid lock I planned to control with an ESP microcontroller with a built in Zigbee radio. I installed all of the wiring but realized due to the triple locking mechanics the dead-bolt would have to be permanent, which means I need to move the solenoid (easy) and develop a way – perhaps grounding – of unlocking from the inside.

When trying to plan for tech, I watched several IoT YouTuber videos. One thing I couldn’t grasp was how they had so many devices. I now know that this simply sneaks up on you. Interior shades, exterior shades, window film, art lights, can lights, exterior lights, air gradient, RATGDO, etc.

For physical connections, I put USB-C (from ATXLED) and CAT-6a in many of the spaces. The “security” cameras, including the doorbell, have CAT-6a connections. I say “security” because realistically home thieves wear all black and are in and out within seconds, so cameras are useless, despite what the aspiring despots at Amazon may claim. Instead, cameras are useful for just being lazy and not having to go to the window to look outside. The USB-C has proven quite useful for charging my laptop or phone anywhere, including the greenhouse.

Without having a homelab before, much research was needed. I determined a used Dell Optiplex i7 newer generation with 32GB RAM and 1TB SSD was sufficient for a server. Zigbee operates via a Conbee II USB. Proxmox as a VM manager was new to me, as was pfSense for routing. Home Assistant seems like unnecesarily complex source code, but as an application it works well. For NVR software, I’ve tried a few but none are great.

I bought a refurbished 6-zone network audio amplifier for the ceiling speakers, and luckily it worked with Home Assistant because otherwise I’m not sure how it was supposed to operate.

ATXLED sold me a home automation package. Unfortunately, I haven’t been able to use it because it requires connecting to some random server which is against the policy of the home (higher paygrades and such). Murray told me the root password for the controller which is just a Raspberry Pi with a hat, but haven’t had the time to figure out how to get it to work with Home Assistant. I had to modify some of the vanity lighting in the bathrooms by ripping out the transformer or installing a step-down DC-to-DC converter to work with the 48V source. And as for my custom LED systems, I ran a lot of 18/2 from some of the shade terminals to power art lights and cabinet lights. Despite 72 feet of LED strips in my bookcase, I hardly use it. More interesting is a the lighting I installed for a post-occupancy custom closet cabinetry install. The light switch was covered by the new cabinetry, so I installed a door sensor that turns on the cabinet lighting when it is dark via Home Assistant. The few door sensors I have installed have had their moments of general utility.

Internet inspiration. Reality: clearcoat maple cabinets, black trim, and LED.

Originally I wanted either a solar roof or solar shingles. I reached out to Tesla multiple times for solar roof and powerwall and they didn’t get back to me. Certainteed shingles seemed interesting, but I ended up going with regular panels. The solar installer recommended Enphase, but I learned it conflicted with the house policy; specifically, Enphase requires cellular data to operate. No bueno. The installer introduced me to Sol-Ark which makes a hybrid inverter, a perfect solution for me. The hybrid inverter lets you tie together solar, battery, and grid connections seemlessly. I bought a 14kWh battery and the installer had an electrician wire it up. Unfortunately, they installed it incorrectly. After a few power outages I finally was able to solve the issue.⁵ I can technically add more panels or even a wind turbine in the future.

Conclusion and Future

After the house was built I was asked what my favorite feature was. “The automatic shades and not hearing the sump pump,” I’d reply.⁶ The automatic shades are still my favorite feature. Every evening just before sundown, they operate and it makes me smile.

There were few issues after move in. There was a noise around the floor above the water heater, like an expansion. The plumber came out and couldn’t figure out what was going on. Also a drain pipe made similar noises.

The HVAC was probably the biggest issue. High performance homes are designed for retaining heat. Little did I realize how difficult it would be to cool it properly. Like a more decadent passive house on the east coast, sleeping at night can be a problem. Generally if outdoor temperature is above 32F/0C, I don’t need to run the heater.⁷ But even with such low outdoor temperatures I may have to cool the 2nd floor with the heat pump because it is 70F/21C in the bedroom.

One of my favorite concepts in computer science is the idea you can run any program with a single loop. The HVAC system has several components that each have their own motor. Instead, it should be one big air loop.⁸ There is an avant-garde home called an earthship which has a huge tunnel of air, so if your earthship is too hot or too cold you open the vent and your home becomes mild. The creator of earthships says your home will be 68F/20C year-round – as it should be! Having an electronic damper to allow your home’s air travel through an underground pipe network means, depending on where you live, you could condition your air to within a margin and then heat or cool it from there. Lastly, I think there should be more dampers and vents. Air flow should be precisely engineered in a simulation. The current state of the art is quite nascent, imo.

There are some obvious lessons here. One, if you hire a builder make sure they are elite. Two, make sure the contract spells out the details as much as possible. There is a case in contract law from the early 20th century where Reading pipe was specified in the contract and the builder used a different pipe. The client demanded it be replaced. The client lost the court case. I learned that contracts are one of the weakest areas of law. Because the project went beyond the contract timeline by 2 months, I asked an attorney if I could recover the market value for renting it out and was told, “Did you have plans to rent it out?” The only thing I could recover were known costs like the amount of extra rent I had to pay. I decided to wait until the last minute to drop that on the builder, otherwise they might have inflated some other cost for the last payout. Matt Risinger, an elite builder, refuses custom contracts. It doesn’t hurt to try to specify details and consequences of failing to adhere to those details. While pipe may have been irrelevant decades ago, there are a lot of reasons today that subsitutes are not identical. OSB and plywood are different for both health and air barrier reasons. One solution to this is taking a cost-plus approach. You pay for all of the materials and the builder takes a percentage of the project.

If I could have done it all over again, I’d have been my own builder and hired the first builder as a consultant. I’ve never seen such incompetence in the tech industry, so perhaps that’s why I was so naive. For any future projects I’ll hire an elite team. Fortunately, I can take my time for any future projects and complete due diligence.

One realization I had when the house was near finished was living quarters don’t matter that much. Sure, excellent air quality and having different spaces to retreat is nice, and there is a positive psychological effect to reading between a piano and a built in bookcase. But I didn’t mind my college-grade rental. I wanted to sell my newly finished home and move to the Bay Area. I found some homes on the peninsula for just over a million which needed some serious renovation. The plan would have been to seal any obvious penetrations with tape and run aerobarrier. In 2018 and 2020, wildfire smoke lingered for over a week, so air sealing a home there is a must.⁹

At move in I had no interest in construction. A month later I was checking up on Stud Pack. Somehow I came across a project by an architect in Belgium who built his energy efficient home inside a large greenhouse. This fascinated me because greenhouses give you a microclimate which you can influence. I’d like to one day build that home.

Recommendations

Samsung All-in-One Heatpump Washer Dryer 🔥🔥🔥
Graywind Blinds 🔥🔥🔥
Plumbing Deals 🔥🔥🔥
Hunski Hardwoods 🔥🔥🔥
Sol-Ark 🔥🔥
Nu-Aqua 🔥🔥
Used Axis Cameras from eBay 🔥🔥

Anti-Recommendations

Pella 👎🏻
Ferguson 👎🏻
Broan ERV 👎🏻
Tesla 👎🏻
Cellulose Insulation 👎🏻
Fast-Growing-Trees 👎🏻

Postscript

There are comments on the Internet regarding this post like “Not to be mean” or something implying some sort of weakness on my part. On the contrary. I got what I wanted. It is flawed, it isn’t perfect, but on the whole it is great. Most of these commenters do not read the whole post, and who would, unless you are planning to build a house. A theme here is, “who is the builder?” the answer to which is me. Had I not been an integral part of the process this house would indeed be less than great. Every builder makes mistakes. Risinger’s cabinet door was drilled wrong, and even worse someone put a hole in one of his drain pipes. Also, Sam Altman negotiated several million dollars off his newly constructed home only to learn it was built so poorly that it produced mold. I know for a fact I don’t have such problems. Given the conditions under which I was operating it turned out just fine. I learned enough to minimize the mistakes next time (yeah, next time, no quitting over here). I have complete control over my environment going forward, I plan to only hire top quality people whether it be construction or any other industry, and I know exactly what to expect. Like Risinger, I’ll still make mistakes but fewer and different ones.

To illustrate the extent of my desire: I created a database of places I wanted and created a website that promised to help people get the most value of their home. Thiswasn’t sinister as I planned to genuinely consult with them. In anycase, I never fully executed on this but one of the houses ended up in the hands of a builder and anotherperson as a result of letter-writing responded a year later. ↩︎
Run until your muscles burn and veins pump battery acid. Then run some more. ↩︎
It was a family business and the person helping me when I picked it up was speaking Polish. ↩︎
There was an instance on Stud Pack. It turned out there was an issue with the installation. ↩︎
They installed an Automatic Transfer Switch, which Sol-Ark does not recommend and YouTube setups I’ve seen only use a manual transfer switch. ↩︎
I didn’t get into this, but the sump pump was sitting on concrete directly and it would vibrate so loudly that you could hear it clearly on the first floor. I bought some rubber feet and asked them to install it, which solved the problem. Who is the builder? ↩︎
This is true if it is sunny. If it is cloudy or night, indoor temperature might only be mid-60s (17-18C). ↩︎
An exception may be HEPA. ↩︎
The Bay Area has excellent air quality and only a few instances of horrible unbreathable air. You can drive down to the Central Coast or past Reno to find fresh air (which is what I did back then) without doing this, but it is nice to have control over your air quality. And, maybe wildfires will increase, maybe tech will save the day. We don’t know. ↩︎

Why Self-Driving Cars Still Don't Exist

Wed, 24 Sep 2025 00:00:00 +0000

Version 0.96

Stanley, winner of the DARPA Grand Challenge. Steve Jurvetson / Wikipedia

I had planned to write a version of this in 2018 specifically addressing Tesla, butI think it is important to include the now-autonomous car industry as well.

Since I haven’t heard it said explicitly, I will go over 3 specific reasons why autonomous carsaren’t and can’t be autonomous, and why companies like Tesla are in worse shape than those like Waymo.

I am absolutely an optimist. I’m frequently an early adopter of great technology.¹ With robotics I’mno less optimistic; the thing to remember about robotics is that there is a line between engineering andfantasy.

You can engineer a robot to clean debris from a pool. The reason this is the case is because the taskis clearly defined and the environment is clearly defined. For autonomous cars, the task is nearly (with some edge cases)perfectly defined, but the environment is not a grid with traffic lights and stop signs, but a very complex world withseveral unpredictable factors including weather, humans, animals, road conditions. To tackle this and believe it is merelya difficult engineering problem is as foolish as the predictions of the 1956 Dartmouth conference.

Lidar, Cameras, and Sensor Fusion

The trigger for this blog post was a clip that Musk re-xed²where the former CEO of defunct self-driving company Cruise claimed Tesla was right to chiefly utilize cameras and stay clear of Lidar sensors. According to him, Waymo is heading in that direction. Over a long enough time horizon he believes autonomous cars will all be camera-based.

Robotics is not a computer science efficiency problem. The reason for an array of sensor types is not to train a camera system and it is not to otherwise bootstrap a self-driving car to use fewer sensors. The real world is very noisy, and different sensors work well for different use cases at differenttimes. Critically, autonomous robots in high stakes environments like roadways require redundant sensors.

I’ve heard several experts³ shy away from that reality saying things like “it’s complementary, not redundant.” It may be true that the primary use case is complementary, but if a rock smashes into the middle of the windshield (where the main stereo camera system is housed) while in the left lane of a busy a highway, I hope the autonomous car I’m in has Lidar so it can safely pull over. If you’ve driven through Vermont or Tahoe in the winter in a Tesla⁴, you get messages asking to clean some camera or other. The engineers thought about the “de-icing problem” but failed to put little windshield wipers by each camera to clear salt residue. Even a radar sensor could mean the difference between a safe halt or a multi-car accident.

Tesla removed radar sensors claiming that when the car approached a bridge, the radar sensor would give an inaccurate reading and cause problems. I’ve owned an ICE car that was decked out with a radar sensor for following the car in front of it without intervention on the highway. That worked out rather well. If anything, no matter the company, every autonomous vehicle should start out with a fusion of sensors. In order to remove sensors, it should be proven that the sensors are not necessary. For example, maybe you don’t need 3 lidar sensors, you only need one.

Musk’s quip that “we don’t have lasers shooting out of our eyes” is telling: robots are not humans and as long as they are using electronic components they will need to be treated as machines. If they are in an uncontrolled environment they are going to need redundant hardware.⁵

Software Verification

Autonomous cars are stochastic systems. A neural network, end-to-end or not, cannot be deterministic. You can run camera input through a Gaussian function for preprocessing and some variant in stimulus, like the way the sun shines or reflects off a color, will cause the result to be different.

If you have driven in a Tesla, you’ve seen the objects on the screen transform into different objects, sometimes ghost objects that aren’t there.

Tesla classifies its own car is a truck. Nov 2023

Tesla’s console software is complex and buggy. A few weeks ago the Tesla audio app was stuck in a weird state where the display didn’t reflect what I was hearing on the speakers. After trying many ways to overcome the situation, I had hold both steering wheel buttons down to restart the computer. You don’t want to do that if the autonomous driving software falls into some local minimum! Which makes me wonder: why doesn’t Tesla verify their console software? It doesn’t give me confidence that there are bugs like that in the console, and I’m surprised that other people haven’t run into bugs and made the analogy: what if there is a bug in the autonomous software, which, by the way, already isn’t possible to apply software verification due to its non-deterministic nature.

Sentience

In order to have complete self-driving, without any human intervention what-so-ever on the roads as they exist today, you needsentience.⁶ Or, to put it another way, if self driving cars truly exist, you’ll have a lot more types ofrobotics completely autonomously operating.

A good illustration of the degree of cognitive ability here is in the movie Short Circuitwhen Johnny 5 lands on a grasshopper and asks Stephanie to re-assemble the grasshopper. It quickly learns the value of life. It’s alsowhat Detective Spooner reflects on in iRobot: don’t calculate the odds of survival, save the girl. An LLM, despite largely being a reflection on society’s social mores, is not sufficient because it lacks authenticity. Another way to say that, is that as long as OpenAI has a disclaimer below their input box, there’s a chance the resulting decision will be garbage. The cars need to be an incarnation of Takumi otherwise there will be no trust.

Today, every single autonomous car company has humans in the loop ready to take over. The extent likely varies, and in the case of Cruise we saw that there are some limitations to this attempt to solve the last 5% of autonomy. It is very unlikely that – until almost every car is autonomous – this will change in the near future. There must be a fundamental breakthrough that is on the order of a sentient human.

About 10 years ago in Mountain View I was riding my bike through a round-about. A Google car (as it were) had entered, so I thought I’d wait for it to make its way through before I went. I guess most bikers claim right of way, because the car came to a halt. The two testers had to hit a button to restart the car. So I went off to the side, again, to wait for it to continue, and I rode in circles. The car halted once again. The testers were laughing and hit the start button again, and I went on my way. It was an important lesson: self-driving cars have no way to deal with people in the real world. In the recent LA unrest Waymo cars were abused. If a mob descends on a car in New York, the driver will get scared and attempt to flee; would a self-driving car do that today? Could you issue a voice command “my safety is at stake, be more aggressive” and would it obey?

Rodney Brooks has been criticizing the hype of self-driving cars for years. There are few people in AI who seemgrounded (no pun intended) to the nature of intelligence (LeCunn, Mitchell, Brockman are a few others). Brookshas been at it since nearly the beginning, starting out at Stanford’s SAIL and engaging with robots of varioustypes at MIT for decades. To boot, he is one of the few people who has actually created a massively successful consumerrobotics company, and that, too, was decades ago. It’s worth reading Brooks’s latest prediction scoresheet.

What’s the solution?

In 2020 I considered emailing the board⁷ of Comma AI, a company which makes a dashboard device that connects to a car’sCAN bus – an adhoc self-driving technology. It seems this wasn’t yet working as anticipated, and egos clashed betweenthe heads of Tesla and Comma AI regarding who would achieve self-driving first. (Why the chutzpah? Prove it.)

My suggestion was going to be to steer the company in a direction where the customer is not drivers, but insurance companies.It could be a system where the more people who joined, the steeper your insurancediscount would be. Why? Because if everyone had a form of self-driving technology, there’d be an enormousreduction in accidents. There would still be accidents, but of the type where two cars collide head on or side-swipe,the number would approach zero.

Rodney Brooks pioneered the subsumption architecture, which is essentially levels for robot control and was adapted to hierarchical controlsystems in Stanley, which won the DARPA grand challenge, and other autonomous vehicles. At a basic level a car should avoid hitting anything around it and communicate with other cars. If such a system were implemented in vehicles, you could imagine a worst case scenario of spring-like behavior as people drove, but the best case would mirror an ideal computer simulation of traffic on the roadway since one person braking for no particular reason wouldn’t cause a domino effect of cars braking.

I’m not sure why every company seems super focused on full self driving, rather than a robust network of partially self-driving cars. Society would be better off. After over 10 years of this, people still buy into the hype despite daily fatal collisions with cars, bicycles, and pedestrians. It’s unfathomable that leaders of these companies are flexing but, imho, have proven nothing. If your cybertruck is safe, it shouldn’t be flipping over a retaining wall in Oakland under any circumstance. I hope either I’m wrong and full self driving cars become standard in short order, or that something changes enabling every car with a CAN bus to at least avoid serious accidents.

I bought an iPhone the day it came out when at the time severalcritics wrote that it was NGMI. Without test driving one, I bought a 4 digit VIN Model S in 2013 becauseof the New York Times article from that February with a photo of a Model S being towed. ↩︎
Original Stripe interview ↩︎
I mean, there really is no such thing as a self-driving expert because self-driving cars do not exist. ↩︎
Still unfortunately the best EV. I feel like I’m waving $50,000 in the air and dozens of automakers shrug indifference. ↩︎
Tesla does have redundant AI chips! ↩︎
This is a cognitive science term. It doesn’t mean “senses” but rather a higher order cognitive function. ↩︎
I could have networked my way or cold emailed, and I decided it just wasn’t worth it, though I wish I did anyway. ↩︎

Trader Joe's Does Not Have Surveillance Cameras

Wed, 17 Sep 2025 00:00:00 +0000

Version 0.91

Just Looking to Buy Some Low Voltage Wire

In February of last year, I went to Home Depot to get some low voltage wire. For some reason, it is secured behind a locked fence. There was no one in the electrical area, so I walked a few aisles down and found someone in paint. Another person joined, and we had a very DL conversation about how I almost just reached behind the fence and grabbed the wire I wanted. The young man said, “I wouldn’t… they have some wild systems.” The young woman smiled and thought about it then said, “Yeah don’t do it.” This could have been a conversation at a bar or a social event, not like speaking with a company representative. These are the type of people who wouldn’t object under normal circumstances.

Back as far as 2003, Home Depot had a sophisticated surveillance system, including the ability to email photos to the police. It’s not hard to imagine how it may have become even more sophisticated in the last 20 years.

Home Depot Surveillance Command Center via CNN

In March, I went back to Home Depot for some low voltage wire.¹ Again, there was no one in electrical. I walked 5 aisles or so in one direction including paint and not a soul was around, then 5 aisles in the other direction and no associate to help this customer. So, I reached behind the fence and pulled out the low voltage wire I wanted and headed to checkout.

Thief. (Or maybe just a guy trying to buy low voltage wire.) via ABC

Another feature of Home Depot is they have maybe one or two clerks at checkout. I used self checkout which was the only option. My change was $5 and the machine said to talk with an associate. The associate had to go over to another cash register and take out $5 to give me change. I’m confused at this point: why would the machine not say something like “cash unavailable” or alert an associate that it was out of cash? I’ve been using self checkout for longer than I’d like to admit, and I’ve not once had this issue. Also, the machine had neither $5 bills nor $1 bills? It simply doesn’t make sense.

As I walked out to my car and by the police car that was idling perpendicular to mine, I thought about the experience I just had. What were those two associates so concerned about that they told me not to reach into the cage? People do that all the time, no? Or maybe it isn’t normal behavior and you might just be stealing something.

Later that week I needed yet more wire, so I decided to go to the local hardware store. As you walk around, associates ask you if you need anything, which I guess acts as both customer service and a deterrent, since there are no security cameras inside. I asked for some 18/2 and the associate went over and measured out the amount I wanted and cut it. Honesty policy about the length when I checked out.

Trader Joe's. Look! No cameras. Harrison Keely via Wikipedia

Trader Joe’s also does not have security cameras on the exterior nor the interior of their stores. Any stores, as far as I can tell. It’s a magical experience. I park, shop, pay, and leave pretty content. No harassment to use a points card or download an app for double the points! No cameras, no license plate readers. Instead of playing some complex game, Trader Joe’s offers simple products at low prices and the human touch.

What to do?

I don’t want this to come off as anti-technology. If Home Depot had an automated greeter that said, “Can I help you?” as I walk in, I’d say, “yes I need some low voltage wire.” It could tell me “Walk 10 aisles to the right and turn left.” An associate could meet me at electrical, and I could get my wire and be on my way.

Best. When people thought different.

Is there another way Home Depot could be more inviting instead of treating me like a potential thief? When I was really young there was a store called Best. I don’t recall much, but I remember when my brother bought a digital piano² it came down on a conveyor belt. Maybe Home Depot could do something like that. Have a showroom model for most products and after purchasing, the product would come down the conveyor belt. It would reduce the surface area of product theft.

"I'd be better off with one good dog." - Security Guard in Manhattan Project

People who are building technology for maximal surveillance are building in the wrong direction. They should be building clever processes that actually help 99.997% of people rather than keeping a watch for the minuscule number of bad guys. For the record, I don’t care if there are closed circuit cameras everywhere, as long as they are just dumb boxes that require work³ to review for some serious incident. In any case, for now there is zero chance I’ll visit a Home Depot with their unwelcoming array of surveillance tech. Instead I’ll be supporting anyone offering excellent user experience.

Postscript

I suspect some may not have come to terms with my writing. It isn’t that Home Depot (and others, like Wegman’s) has a frightening dystopian system, but rather the opposite. It has a badly cobbled together set of software that doesn’t work, and certainly doesn’t improve my experience. There’s an absurd amount of bad software in the world (even Google and Apple are guilty). Trader Joe’s works well because it doesn’t even bother with software.

I have a lot of low voltage wire in my house. Blog post soon. ↩︎
Pretty sick synthesizers, a lot of fun! ↩︎
Applying AI after retrieving the relevant footage might be all right. ↩︎

VPNs should use Public-Key Cryptography instead of long codes

Fri, 09 May 2025 00:00:00 +0000

Version 0.9

A few years ago there was a lot of hype around Mullvad VPN, particularly on Hacker News. Whenever I saw an article postedwhere the comments subsequently extolled the virtue of the service, I had to check profiles to gauge the integrity of theposter.

VPN services have a spotty history due to internal policies, technical oversights, and shady corporate structures. They are tasked with allowing people to masquerade as if they were in another location and among a large crowd of other users, so you will see them competing on “no logging,” security audits, and various methods of payment.¹

When security researchers have something to say online, I pay attention. Like anything else, you cannot take what they say at face value but strongly consider itas truth, especially when you had no basis for whatever the topic may be. Long ago I used algo until wireguard became a favorite tunneling method of that crowd.Below is a list of esteemed security researchers and the VPN service each recommends.

Name of Security Researcher	VPN Recommendation

table 1.1: list of security researcher vpn recommendations

I wouldn’t say I bought into the hype. In the absence of any security researcher recommending it, I decided to give it a try. Signup involves receiving a long code which you useto generate up to 5 profiles. You can use their client or install the wireguard (etc.) profile yourself. I did both methods and managed my profiles from time to time using my long code.I had thought I settled on just two profiles: one for mobile, and one for desktop, deleting a couple while playing around with it. To my surprise, I logged in and saw a total of4 profiles. I deleted two I didn’t recognize, made sure it was the case, and continued testing it out. A week or two later I logged in and saw a total of 3 profiles.

What was going on here? Was there a bug generating profiles? Was someone else’s profile being generated under my account? The account, after all, is just a string. If I had somemalware on my computer you’d think credit card information stored in the browser is more valuable than a $5 VPN service. There is little information about the nature of the company:how many employees do they have, what process do they have in place, is this access code encrypted using bcrypt with sufficient rounds or plaintext in their database? You can imaginea number of scenarios if this access code is in plaintext. It’d be a huge security risk. Even if it were encrypted, at some point is isn’t, which makes it a de facto risk.

I’m a fan of Problem-Solution blog posts. Instead of sounding like I’m complaining, I’ll present it as “this isn’t quite right” and “how it could be better.”²It seems obvious that instead of some long string that’s assigned to you, they issue you a private key. This way, you can simply generate a new private key on your own system andsign over the corresponding public key with your old private key so the company knows you’ve authorized a new key. Anyone remotely technical could do this. For those non-technical, the company’s app,presumably open source and audited, could do it.

For some reason they all accept Bitcoin and Monero but nothing else? Bitcoin isn’t any more anonymous than Ethereum (read: they’re not). ↩︎
There’s a ton that’s obviously wrong with the world that could be vastly improved and therefore there are companies and nonprofits just itching to be formed. ↩︎

The Best CEOs Respond

Fri, 18 Oct 2024 00:00:00 +0000

Version 0.97

The CEO of Mom's Organic asking for feedback

The CEO of a company is responsible for 3 things: the vision of the company, the talent, and cash flow. The latter two are largely affected by the vision. Without a strong vision the company may deteriorate into an unrecognizable asset on the decline.

There are different degrees of visionaries. Most CEOs just aren’t visionaries and are lucky a competitor doesn’t come along and destroy their margin. Many people tend to think of Steve Jobs when they think of the quintessentialvisionary. It’s eye opening when you recognize he spent 12 years at NeXT, and he was 50 when the iPhone started development. He had over 30 years working alongside other brilliant people before developing the product that would catapult Apple into the multi-trillion dollar company it is today. Steve responded to customers:

akhurstiwriting.blogspot.com / thecleverest.com

I’ve sent a lot of emails to a variety of CEOs. I try to keep it short. A few years ago I sent Tim Cook a list of items, including one that prevented me from buying things due to the nature of how Apple handles credit cards¹. Not only was there zero response, but I believe that issue persists today².

There are a handful of CEOs who do respond. Elliot Noss at Tucows responded both times I emailed. Dane Jasper at Sonic responded. Pasquale Romano at ChargePoint called me. I got a lengthy email response from someone on Jeff Bezos’s team. Amazon has dubious business practices, but the thing that they continue to get right (generally) is customer happiness.

A natural argument is that CEOs are too busy to respond. One of the first high profile emails I sent was to Douglas Hofstadter and within a day or two I received a thoughtful email³. Do you think he has more important things to do than send a student a thoughtful letter?

The very idea that a CEO is too busy to respond to an email contradicts the fact that they should be spendinga good amount of time revising and extending the vision of the company. That doesn’t happen in a vacuum. Some people can see all of the cracks in the world, things that need to be fixed, solutions to fix them; there isn’t a single CEO who wouldn’t benefit from feedback from smart employees and customers.

CEOs should – at the very least – acknowledge their loyal customer. A simple “noted!” would do. They could forward the email to their executive team⁴. The customer thinks: “Wow, they took action.” Not all, but some of these customers will amplifythe message to their peers; a formidable word of mouth.

It’s not because Apple failed to implement most of the items on my list⁵ that I’m no longer interested in being an Apple customer. It’s not because Elon Musk didn’t take action on the terrible user interface update for the radio controls⁶ that I’m no longer interested in being a Tesla customer⁷. It’s that there is little evidence that these CEOs have a vision and care about the company⁸. They are riding on a cash cow that will slowly deflate, but it’s the kind of situation that they won’t realize it until it’s too late. CEOs should be focused on the vision of the company. Talent and cash flow will follow.

This is convoluted and a testament to the complexity of Apple’s code base. But importantly Apple lost some amount of revenue but it is unclear how much, because how many other people have this issue? ↩︎
People point to the massive growth in market capitalization as evidence the CEO is doing an excellent job, but in my opinion some CEOs just get on the flywheel at the right time. ↩︎
A bit off topic: I never emailed Paul Graham, but I did approach him at Startup School, and not only did he engage with me, but I was the one to end the conversation. While Michael Seibel is a responder of emails, Paul Graham should probably run Y Combinator. ↩︎
In terms of productivity, they could arrange their email client such that it is a simple click to send the email into a folder which has a cron job waiting to ferry all of the emails to that team. ↩︎
Apple did fix the keyboard issue. I guess you have to create a particular mob for them to really listen. ↩︎
Tesla used to have a matrix for radio stations. Now it is an array which you have to scroll over, and after 5 seconds it scrolls back to the beginning for some reason?? Talent follows vision. ↩︎
In the early Model S days, Tesla had an exclusive club of owners in a private forum on their site. Owners were encouragedto give feedback to management and even had a spreadsheet of top requests (it took a year for the backup camera to show guidinglines). ↩︎
If your vision is to create wild long-shot products and you don’t address problems with your current best-sellers, that’s a problem. Only if you have good sense that the innovation (e.g. not Vision Pro and not pseudo-robotic cars⁹) will replace the existing product is it remotely ok. ↩︎
I’ve been meaning to write a blog post about these non-robotic cars since 2018. Maybe next week! ↩︎

Autocomplete Selections Should Confirm with User

Mon, 29 Jan 2024 00:00:00 +0000

One UX issue that I run into more often than I’d like: when you type in a text box and a list or drop down appears and you tap on it immediately, only to learn that between the milliseconds that your brain executed the tap and your fingers touched the screen a new item replaced your intended target.

Sometimes this is a more benign environment where you are searching for something and after you type a few letters what you want appears, but you managed to type another letter and the results are changed due to network delay.

I don’t want to stray from the topic; however, I must point out that I’ve seen this behavior pervasively in advertising. I first noticed it back in the early 2010s when using Meebo probably just before Google acquired it. You’d go to type a message in response to one someone sent and, surprise!, the spot where you were about to tap to focus the input box became an ad, pushing the input box higher in the interface. It seemed too well timed to be just a coincidence. Today I see this exact kind of behavior with Amazon, though it seems they are starting to transition to just making every result an ad so no need to trick you into clicking. Sadly I’ve even seen this with DuckDuckGo, eroding my trust in that company¹. What I can’t fathom is how the engineers behind these platforms let it happen. They must know what they are doing. Or, if you are a fraud investigator and search on Amazon and accidentally click one of these ads do you say to yourself, “Haha! I was fooled again!” The advertisers meanwhile don’t seem to care either, because revenue > cost of marketing?

Transitioning back to the crux of my post: there’s no technical reason for this! If you are planning to asynchronously return some kind of result in addition to your server-side result, it’s obvious that you can put a placeholder in your template. The contents of the placeholder can be up to your imagination: “Our valued sponsors have a result just for you.” It may be more effective to humanize the results than just treating the user as a sucker. “Got you, fool! Now pay me $8.”

Another way to deal with this, in the case of results in apps with search/dropdown combos without monetization incentives, is to calculate the time it takes between the last result populated and the newly invalidated list. In fact, use the same UX exploit that seemed to be happening with Meebo but display a dialog asking the user if they meant the previous result, current result, or something else. That this would only happen in circumstances where there was incongruity between the user’s input and the lagged result means it would be infrequent. Empirically it doesn’t happen often, but when it does it can be obnoxious and cause more grief than a simple confirmation dialog box.

Version 0.95

In fact, this just happened to me while using DDG yesterday. ↩︎

Automate Hotels

Thu, 13 Apr 2023 00:00:00 +0000

A Hotel Lobby¹

Version 0.95

The Problem

I’ll quickly review some issues with the hotel experience, though I imagine they are evident to most people.

An obvious issue is check-in. It’s unclear why this process is so time-consuming. I know there areapps that allow you to check in and use your phone as a key (which sometimes fails to workaccording to various hotel reviews), so let’s simply consider a hotel without thiscapability. You have to say your name, they sometimes ask for more info, ID, credit card, andthe process can take 10-20 minutes. That’s a lot of time for something that doesn’t exactly add to the experience.

Dustin Curtis has probably stayed at over a thousand hotels.I was surprised to learn he uses hotels.com and sometimesHotelTonight². Discovery, it seems, is still a problem. At the time Dustin wrote this hotels.comhadn’t updated its interface in 20 years. It wasn’t broken. I don’t know what they didbut since the update a few years ago there have been various issues like the map does not display all thehotels and filters do not produce correct results. Hipmunk was pretty solid early on andsomehow that fizzled out. Hotels Tonight works great for narrow use cases. Most of the rest of the websites are half-baked.

Hotels are commodities. Not long ago I stayed at a hotel in Colorado and loved it,so I booked the same brand in Indiana and it was like crossing the railroad tracks.For that reason I’m not going to be loyal to a brand.

If you’ve ever been to New York, hotels typically partner with a garage. Oftenthis isn’t a problem, but it can be a bottleneck and I’ve definitely waitedquite some time for the valet to return with the car.

There are quite a few other issues with hotels like new ones tend to be poorly built and I have yet to see a hotel luxuryor not that has iron clad blackout blinds, but those issues are beyond the scope of “automate hotels.”

The Solution

Automate all the things!

Of course it isn’t. Hospitality is about making people feel human and enjoying it.
— dustin curtis (@dcurtis) January 14, 2020

Well, ok, automate some of the things. New York has a parking garage called AutoMotion. You approach the garage and drive your car onto a lift. After you get out you put in a credit card and get a receipt with which to retrieve your car. The car is then whisked into the garage using a robotic system. The price is market, but no tipping and a fairly predictable amount of time for retrieving your car. There are only a handful of these, and they seemed to stagnate not long after their debut more than 10 years ago. It works and should be the norm in any new construction not only in New York but everywhere it any city.

Checking into a hotel should require nothing more than flashing your ID. What reason is there not to prepare your key cardahead of time? Why so much typing at the computer? Indeed, there are some hotels with key card kiosks. Misunderstandings like forgetting to inform the hotel you have a pet shouldn’t be more than a small percentage of cases. I see no reason why it should take any longer than 2 minutes to check in.

There are quite a few other things that could be done to modernize the hotel experience. Why is it that you get assigned a random room? Every room should be set at market rate. Few hotels have decent soundproofing in the cavity between the floor and the ceiling, and if you are on vacation the last thing you want is to hear the person upstairs on a business trip getting ready at 6am. On that note, hotels should allow patrons to submit their plans to an automated system that helps themidentify where other people like them are going to be. Do you want to be in a room next to someone smoking on the balcony?

The point is, hotels could do better and enable a predictable experience. Automation can help quite a bit here, but hotelexecutives need to secure the fundamentals. It seems like low hanging fruit if someone wants to start a hotel.³

Photo by Oswald Elsaboath on Unsplash ↩︎
I’m fairly certain this was a tweet, likely deleted ↩︎
Dustin Curtis wanted to do this – I wish he would! ↩︎

Government Should Incentivize High Performance Home Builders

Sat, 12 Nov 2022 00:00:00 +0000

Passive House in Brooklyn. Photo by Sam McAfee.

Version 0.92

In the US, new homes built today vary by the standards set by local governments. Standards may include specificeffective R-values for walls of the building envelope and usually a larger effective R-value for the roof. This,however, is not ideal because leaks can occur through cracks, the concrete slab, and thermal bridging.

Thermal bridging occurs when you have a low R-value item like lumber that directly connects the interior with the exterior of the envelope. There are techniques to eliminate thermal bridging like using raised trusses for the roof or putting insulation on the exterior of the house. Concrete slabs can be insulated using structural foam, and you can create an air-tight house by tapingand caulking all cracks and holes including those created by the trades that work on the interior¹.

The issue is that builders have almost no incentive to take these extra steps. In some markets, it can distinguish a home, but most people apparently are happy with basic new construction. Further, homes are being built with more square footage. In some markets 3000-4000 square foot new homes just 10 years ago was standard, and today it is more like 5000-6000 square feet. Finishing an attic and a basement costs less since those structures are there anyway, so this helps margins. But do newer generations want such a large footprint? Is it necessary?

People informed about health and environmental effects of natural gas are looking away from gas stoves and toward induction stoves. The topic of the environment has gone from recycling cans to actively tracking one’s carbon production. It’s unclear to what extent, though. If a builder created two homes next to each other, one 5200 square feet with standard construction practices, while the other was 4000 square feet using high performance building techniques and both were offered at the same price, which would be coveted by the 30-something home buyer?

It’s hard to know the answer because building a speculative high performance home is risky in most markets. San Francisco and Boston have not only high land costs, but building a home that isn’t high performance can be anywhere from $350-$450 a square foot. High performance homes that aim for passive house or net zero energy usage can add 10% or more to the cost. Only if builders start seeing high custom home demand for net zero features will we start to see more speculative homes being built this way.

But that’s not enough because there are people who don’t have the wealth to build a custom home nor buy a new home. It’s also hard to recognize the home as an investment in future energy costs. Utility costs are on the rise, and not by a trivial amount or an amount keeping pace with inflation, but as much as 50%². Even if carbon footprint isn’t on someone’s mind, rising energy costs should be.

Normally I opt for market solutions to most problems. People seem to think government can play a dominant role in ceasing or reversing climate change, but that ship sailed 20 or 30 years ago. Climate change needs a beast of a solution, or a lot of little solutions that represent such a force of nature to reverse carbon emissions. Market forces at work here include companies that brag about negative emissions. I suspect we’ll see some unimaginable advances in the coming years thanks to the expectation that corporations not only claim zero emissions, but actively invest in carbon dioxide removal.

What can government do, then? Make high performance homes a standard. I don’t mean forcing builders to do it, but incentivizing them to do it and then high performance homes will be expected by the market. The same applies to retrofitting existing homes with net zero elements³.

The system could work like this: if the average new home is closing in on $500,000, then it might cost $50,000 to make it high performance: triple pane windows, extra thick insulation, insulated slab, and an air change per hour rate of 1.0 or less. The builder gets reimbursed for these costs plus a 10-20% fee. The government could amortize the cost over the course of 50 years. In this example it might be $100 a month, which is about the average cost of heating alone in the US⁴. Having a high performance home means heating and cooling require very little energy since the home prevents air from escaping (or entering) except using an energy recovery ventilator⁵ to keep the air fresh with minimal energy loss. Ancilliary benefits include comfortable atmosphere and allergen-reduced air. Overall, this would be a win for the homeowner, the builder, and society at large (government).

How the system is implemented is open to interpretation. That partly depends on the philosophy of the nature of government sovereign currency. It seems since 2010 there has been more adoption of Modern Monetary Theory as the idea of “current generation paying interest, future generation pays off debt” seems untenable. In terms of inflation, I don’t think there is a material supply issue but instead a trade worker constraint. For the repayment, maybe there is a third party that manages the implmentation, a bond system, or traditional IRS tax-oriented system. The claim of 10% is also for illustrative purposes: it could be less, especially as standards increase and environmentally minded buyers come into the market. There are also some builders who have their eye on both budget and high performance⁶⁷.

There is even a service that sprays a type of sealant into the air while the house is pressurized to seal any little cracks that went unnoticed. ↩︎
https://news.yahoo.com/advocates-call-utility-lower-proposed-012531440.html ↩︎
For example, replacing old siding can mean a chance to put batt insulation on the exterior. ↩︎
https://www.consumeraffairs.com/news/average-heating-bills-could-rise-172-this-winter-092322.html ↩︎
https://en.wikipedia.org/wiki/Energy_recovery_ventilation ↩︎
https://www.youtube.com/watch?v=hwpoo6T3oIA ↩︎
https://www.youtube.com/watch?v=fjwjeoLi6X0 ↩︎

Ask VCs for Referrals Anyway

Sun, 29 May 2022 00:00:00 +0000

Version 1.0

Some years ago I saw a well-known¹ investor on Twitter say, in effect, thatif he decided not to invest in you, don’t ask for referrals.

From an investor’s perspective, this often makes perfect sense. Great investorsplace their money with people they believe will succeed, and perhaps therewere so many red flags the investor couldn’t possibly recommend the entrepreneurto anyone.

If the investor is talking to the entrepreneur in the first place, however, itprobably means they passed some kind of initial screen: they aren’t completelyout there. There are countless stories where investors declined entrepreneursearly on like multiple decacorn company builders Dorsey and Musk.

Don Valentine did not initially invest in Apple². Instead, he referredMike Markkula to Steve Jobs. Markkula invested first, then Valentine investedwith Sequoia a year later. Can you imagine if Don Valentine simply brushed them off and said no?I may not be talking about Don Valentine right now, for one.

Most venture capital firms have an investment thesis that is procured to theirLPs. Angel and seed investors have their own verticals and criteria forinvesting. The narrow scope of these theses precludes at least some power law wins.After all, I know of no thesis that is, “Invest only in companies that will go public.”

“I have traction on my college social network” “We are a B2B outfit.”

“I have an idea for an electric car.” “Oh, we only invest in software.”

A great investor will try to help in some way. They may know someone in theirnetwork who has a completely different perspective. If that person invests theycan turn around and reason with the first investor, or the first investor canstill have a way in to the next round. It’s also just good business.

I don’t quite recall who it was. ↩︎
https://en.wikipedia.org/wiki/Mike_Markkula#Apple ↩︎

Phone is Still the Best Way to Order Pizza

Thu, 27 Jan 2022 00:00:00 +0000

Version 0.9

When ordering pizza for pickup, there is no easier way to do it than usinga phone. On average, it takes no more than 20 seconds:

Dial number
Say: “Large Pizza for Pickup”
Hear: “OK. 15-20 minutes.”

There are many, many websites and apps that partner with local pizza shops.Often these apps require you to sign up or add excess information. Why isone required to put in first name, last name, phone, email, and address whenphone number and first name is all that’s required when calling?

What’s worse is that almost all of these apps require you to “sign up.” This isone of those situations where signing up is irrelevant, just like with mostmobile apps¹. In theory I should be able to autocomplete first name andphone number (maybe credit card²) and voilà my pizza is in queue. Signing upfor yet-another-food-service-app is a turn off.

That technology is causing more friction than less isn’tsurprising. It seems recently that a lot of technology does this. An obviousexample is captcha. We wouldn’t have hilarious coffee drinking robotcommercials without it!

If only the solution were so simple as to build an app that only required firstname and phone number (and dumped that info or just store it in localstorage).The biggest issue of course is getting the menu items into the system. Maybethat’s not even necessary. Why not just have an image of a menu and a box whereyou type the order. “Extra large pizza, half green peppers and onions, halfextra cheese.” Already this system feels like it would in the worst case matcha phone call, but probably cut that time in half – especially if utilizinglocalstorage. Maybe one day technology can play a role in improving userexperience rather than destroying it.

Mobile apps can create unique device identifiers, and on iOS they canalso leverage iCloud to store information when moving to new devices. ↩︎
This shouldn’t be necessary, but if for some reason people are submittingfake pizza requests I guess it could be required. People still use cash;something I’ve been doing recently as a mechanism for budgeting. ↩︎

New Blogging Paradigm: Drafts

Fri, 21 Jan 2022 00:00:00 +0000

One of the challenges for blogging consistently is finding the time to do itand make good posts. Posts like “Who is Satoshi” or “YC Bookmarklet” weresimple enough to knock out, so I did.

For some time I’ve wanted to write about in depth topics like self-drivingcars. This is what I’ll post next; and while I try to keep a positive vibeon this blog, that next post will be rather critical.

In order to accomplish this, instead of spending a lot of time crafting a blogpost, I will post drafts and revise them. Each post will start at 0.1 and maybeone day it’ll reach 1.0.

Who Is Satoshi Nakamoto?

Thu, 01 Apr 2021 00:00:00 +0000

The mystery surrounding the identity of Satoshi Nakamoto, creator of Bitcoin,has been entertaining and somewhat frustrating to observe. When that reporterfound a random guy with the name Nakamoto and turned his life upside down, it wasprobably peak “we have no idea who this is” and yet it is still frustrating to seefrom time to time bright people say, “Aha! It is x!” with weak evidence.

Satoshi wanted to stay anonymous, and for good reason. If Bitcoin worked,everyone would be after him.¹ Personally I didn’t adopt Bitcoin soonerout of fear of alternative money laws in the US, so imagine creating analternative money! Then, as it grows more successful, the early miningactivity attributed to Satoshi is worth enough to be competitive with thewealthiest individuals in the world. The wealthiest people have “security teams.”Satoshi, I guarantee, does not have nor want a security team.

A few years ago I myself was curious so I looked at the genesis of Bitcoin.Based on that story I chose one person and compared that person’s code withthe code that Satoshi Nakamoto initially distributed. Next I checked thatperson’s writing with Satoshi’s writing. The code and writing were verysimilar. The two people’s characters were similar. What was different? Well,if you were going to try to conceal your identity you’d want good opsec butyou’d also want to be good at counterintelligence. Since Satoshi has fooledalmost everyone²³, I’d say he did those things well.

I won’t disclose who I think Satoshi is directly. What’s the point ofunveiling who it is? If the person wants others to know directly that he isSatoshi then he will do so. In fact, I think one day he may. Here I will onlymake three claims: Satoshi is one person, he is still alive, and he identifiesas “he.” I was wrong in my estimation of the identity of Startup L Jackson⁴so certainly I could be wrong about Satoshi. But for fun here is a hash of aunique sentence stating his identity:

ae5822f1add18cf25046267de7ede6fea25a512ce06b3d51dbafbcf3e2d7ed7b/512256

Yes, I’m making a claim on gender here. ↩︎
Amusingly, Gwern touches on it in confusion. ↩︎
There are a few randos who make claims, but without evidence. ↩︎
I thought it was Hunter Walk, but to be fair I didn’t read through theirsource code. Also a bit different circumstance: SLJ was a persona created byThompson and Satoshi merely applies techniques to conceal true identity. ↩︎

The Curious Case of the Y Combinator Bookmarklet

Sun, 28 Mar 2021 00:00:00 +0000

I frequent the /new page of Hacker News, usually to find interesting articlesthat were missed by the masses¹. In recent years I noticed that people weresubmitting “Y Combinator Bookmarklet”at least several times a week. From time to time, this would get a small number of upvotes.

After a while I started wondering why people were submitting this so frequently.Reviewing the bookmarklet page, it suddenly became clear². Logged-in HN users were clicking thelink which would then submit the bookmarklet page itself to Hacker News. Atsome point these links were simply automatically marked dead whenever submitted.

But, there is a better (and simple) solution! Just add javascript to preventthe default behavior. The yellow highlighted text below shows the change.

This allows users to drag the bookmarklet to their toolbar while preventing themfrom accidentally submitting to HN. An ancillary benefit is that, due to how YCtitles their pages, it will no longer be known as “Y Combinator Bookmarklet.”

This seems to be a problem with HN. First, there should be a preferencewhere landing on /new is the default. Second – I’m not sure the solution –there should be a way to allow for long-form articles because they rarelyif ever make it to the front page. ↩︎
Worry not, I did not click the link! ↩︎

There Should Be More Mergers

Sun, 23 Feb 2020 00:00:00 +0000

In February 2000, Peter Thiel and Elon Musk met at a cafe on University Avenue in Palo Alto to discuss merging Confinity and X.com, their respective companies. Here is how Thiel tells it in Zero to One:

If you can’t beat a rival, it may be better to merge. By late 1999, we were in an all out war… the focus was defeating X.com. In February 2000, Elon and I were more scared about a rapidly inflating tech bubble than we were about each other: a financial crash would ruin us both before we could finish our fight. Sometimes you do have to fight. Where that’s true, you should fight and win. There is no middle ground.¹

PayPal went on to become a powerful company that today has a market cap of $140 billion. We’ll never know what would have happened if Thiel and Musk hadn’t agreed on a 50/50 merger. There is a realm of possibility including the unspoken idea that both companies survived and competition drove innovation – something to keep in mind.

In recent years, I’m surprised that there aren’t more mergers of the same kind. Not paltry acquihires, but rather taking two companies that are otherwise forces with strong teams that would mesh well together.² It’s possible to object with a claim that maybe a weaker company ought to die out; however, we continue to see Y Combinator fund competitors who don’t seem to come to the realization that merging is an option. Y Combinator doesn’t fund ideas, it funds founders, and so it doesn’t follow that “the weak die out” unless Y Combinator was mistaken in their assessment of accepting founders into the accelerator.

One of the earliest examples of this that comes to mind is Balanced Payments, which launched in 2012 a couple years after Stripe. One of the distinguished features of Balanced was an API that allowed developers to integrate marketplaces. Stripe eventually added this feature, but Balanced was reportedly processing $370 million a year in early 2014.³ In 2015 Balanced shut down, and Stripe’s co-founder Patrick Collison called it “bittersweet.”⁴ What’s interesting is that Balanced and Stripe are both Y Combinator companies. Presumably they each had equal access to the resources and advice Y Combinator provides. Without knowing the details of the shut down, it seems odd that the company wouldn’t merge with Stripe. Were they drowning in lawsuits? The value of the merger wouldn’t exceed the cost of merging? Stripe’s valuation dwarfed Balanced’s?

Y Combinator seems to be the perfect place for mergers. Every winter for the past three winters, Y Combinator has funded a podcasting company. In winter 2017, Breaker. In winter 2018, The Podcast App. And in winter 2019, Brew. TechCrunch even makes a note of this:

The accelerator still admits many international copycats of U.S. successes, and YC is also repeating itself a bit. The Podcast App pitched the exact same product and strategy as Breaker, which debuted at YC exactly a year ago.⁵

So, great! Admit them. Advise them. Shower them with resources. But why not suggest a merger at some point? Is there room for 3 podcasting apps? In addition to these there is Stitcher, Marco Arment’s Overcast, and the native Podcast app (on iOS at least). So many apps in such a small market is inefficient.

One more instance of two Y Combinator companies that would be fit for merging are Yoshi and YourMechanic. YourMechanic started out with a bang; they built something people wanted and had some excellent blog posts which marketed their product, but somehow the leadership disappeared and the company appears to have gone sideways. Yoshi started out with fuel delivery straight to your vehicle, but quickly encroached on YourMechanic’s market. It seemed obvious in recent years that these companies combine forces and merge. Instead of a single, efficient mobile mechanic workforce in the Bay Area, for example, there are two workforces and the overhead and marketing costs that go with them yielding less margin for profit (if any!).

Yes, it’s hard to know whether mergers in any of these instances are a good idea without knowing the details of the companies and the cap tables that go with them. But what’s surprising is that we rarely – if ever – see a TechCrunch headline announcing that two companies have joined forces. It’s even more surprising that we don’t see this from Y Combinator given the prevalance of companies they fund in similar markets.

Paraphrased from Zero to One ↩︎
And, as noted by Thiel in Zero to One & Levchin in Founders at Work the X-Confinity merger wasn’t the best meshing of teams but somehow was good enough to succeed ↩︎
http://techcrunch.com/2012/11/16/balanced-marketplace-payments/ ↩︎
https://news.ycombinator.com/item?id=9200145 ↩︎
https://techcrunch.com/2018/03/20/these-are-the-64-startups-unveiled-at-y-combinator-w18-demo-day-2/ ↩︎

"How Do You Know?"

Thu, 12 Dec 2019 00:00:00 +0000

In spring of 2017 I was about to move from San Francisco down the peninsula. I had just decided to pause an ambitious side project that attempted to further unsupervised machine learning by automating clustering and instead turn my attention to another project dealing with art¹. Y Combinator had announced their first startup school; I applied and was able to join a group of other entrepreneurial types.

My art project² was going to sell physical art. I had developed an app that enabled people to browse and buy artworks with Stripe. Sounds prosaic but there was an angle of course. Even so, this endeavor was unfortunately not a sit-behind-the-screen and dominate³ type of project like my ambitious unsupervised machine learning project. It required that I go talk with customers and artists. 😱

It took time, but at the very least I was able to start emailing artists via a yearly Bay Area art event that produces a magazine with all of the participants’s contact info. I had a pretty good success rate and set up a meeting with one artist, with another meeting on the way. Not only was I making progress, but also discovered that when you tell artists that you want to sell their work, some want to hear more.

At this point it was about 4 or 5 weeks in, and things were moving too slowly. Yes, it is important to Do Things That Don’t Scale but I was treating this like it was Y Combinator’s actual program and if I didn’t have anything to show by presentation day, why bother?

So I pivoted to another iOS app. This one leveraged the Lob printing API to produce post cards with artwork on it. The hypothesis was that people would want to send a postcard with artwork from the location they traveled to.

At the weekly meeting, the group lead Aaron Glazer, CEO of Taplytics, had a guest from his former YC Group. We were discussing what was holding us back, and my contribution usually went like this:

I am thinking about trying to sell this concept to real estate agents to thank clients they’ve worked with. But I don’t think they’d want it.

This may seem dumb, dear reader, but I believe a lot of people suffer from this fallacy. You find a reason why it won’t work. Disappearing photos would be great! But that won’t work because you can always take a photo of the photo or perhaps the more familiar Dropbox is something that could be trivially built with rsync and big tech solutions will be pervasive soon enough.

So during this meeting halfway through startup school these guys asked me: How do you know? And they didn’t just say it once. Every time I made an assertion they would ask me, “How do you know?” There is only one way to know whether something will work, and that is to test the market. You needn’t build anything! You could. You could build a prototype. You could produce a video like Drew Houston did with Dropbox. But if you have any assertion without evidence to back it up, you may as well trash it. God doesn’t exist. Is that so? How do you know that?⁴ Likewise, I didn’t know whether real estate agents would be interested in artwork postcards.

After that call I did not follow up with real estate agents because I decided to pivot to augmented reality. But the lesson stuck with me. It is important to ask questions like whether it is the right time and what is the distribution strategy⁵, but you don’t want to make any assumptions without data.

I’ve heard people talk about eliminating items from idea lists, but I’ve never heard advice on how to do that. Consider how awful the idea of AirBnB was. Paul Graham tried to convince AirBnB to change their idea!⁶ Probably what made it work was Michael Seibel’s advice to leverage Craigslist on both the supply and demand side as a distribution strategy.

Whenever there is an idea I’m unsure of, it is now easy to think How do I know? If it is something reason would easily talk me out of like an AirBnB concept, I try to get some data. A handful of data points or a few posts on forums is not enough, just as it wouldn’t be enough in a scientific setting. And as with science, I should at least be able to walk away from the experiment learning something new.

Unsupervised machine learning to art? Yes, quite a leap. ↩︎
Not really a startup as there was no growth, right? ↩︎
Paul Graham may argue that you have to talk with customers either way. But a distribution strategy like that of YouTube (leveraging MySpace) or PayPal (leveraging EBay) did not mean having to arrange meetings with people, and in the case of PayPal the customer who sparked their genius emailed them. ↩︎
Atheism makes no sense unless you are claiming that there is no god in the same way that people claim that there is a god. You can participate in innumerable sophistry contests, but there’s no way you can win every time. Just because it seems like there is no god doesn’t make it so! ↩︎
These are perhaps the two most important questions for a startup. It’s ok to be a bit early if you can weather it (for example, AR industry is probably going to proliferate when Apple launches glasses in a year or so). Distribution is key for growth. ↩︎
https://twitter.com/paulg/status/1131337639540477954 ↩︎

How I Use Streaming Services

Fri, 08 Nov 2019 00:00:00 +0000

Participating in the transition from Blockbuster through Netflix online streaming service was an interesting lesson in change of market dynamics. Netflix realized you could be cost effective at sending DVDs through the mail, while Blockbuster snoozed assuming people would browse in store forever. Sure, browsing in real life is an experience, but it was also something of an annoyance for similar reasons why Amazon was successful¹.

For a while, around 2012 or so, it seemed like Netflix and streaming were the future. But it, too, has had difficulty negotiating deals for desired content with the competition heating up from other services such as Amazon and HBO. Turning to Hollywood to create original content, it seemed like it was off to a good start with the series House of Cards.

But as A-list content dwindled, the pressure was on. Netflix had to produce more original content and it tried to hype up shows like Marco Polo which ended in failure. Shows like Orange is the New Black, Narcos, and Stranger Things helped, but rising subscriber costs contributed to its struggle². In fact, it’s hard to see any recently widely talked about content as they ramp up efforts to try to see what sticks³.

Many people already had Amazon Prime when they launched Amazon Prime Video. Amazon has also had mediocre success with original content. Jack Ryan is probably the most talked about, but Amazon, too, has a lengthy list of never-heard-of-this-before original content⁴.

Another popular streaming service is HBO Now, which has been able to leverage its A-list content from the cable service with clout frequently advertised by hotels everywhere, HBO. Most well known content-wise here is Game of Thrones but has done well with newer series such as Westworld and Succession.

HBO Now is actually how I developed my current viewing habits. When it was launched I subscribed to the service just to watch Silicon Valley and then promptly canceled. An ancillary benefit was access to all the A-list content for the month.

I started doing this around 2015 or 2016, and fast-forward to 2018 when Amazon Prime was due to be renewed. Amazon had just increased the price for a Prime membership, and the benefits were questionable. Shortly after declining to renew Amazon Prime, I also axed Netflix.

I haven’t reactivated any of these services since, but I know the desirable content has started to queue up. Jack Ryan was indeed a worthy series. So at some point, just like HBO Now, I’ll spin up a Prime Video membership for the month before canceling. This will happen with Netflix as well. Ideally I’d subscribe to a service one month and cascade other services in subsequent months.

An exception here is Disney Plus. The content, at present, seems to be so good that it will be worth it to just pay for the service month to month without giving it much thought. Disney Plus is also interesting because one of Netflix’s pillars is children’s content. Imagine busy parents that need an hour a day and let their kids have some screen time. Would they prefer to just pay $10 a month for Netflix or risk subjecting their kids to content from randos on Youtube⁵?

In the end it may seem rather miserly to forgo a whole $360⁶ a year on content and services that could prove beneficial in situations like being stuck in an airport or on a long car ride. $330, however, is a lot of money for most people. You could go out to a nice dinner for two a few times over, or buy a nicer car, or contribute that money to Watsi⁷. But I think it speaks more about how the business model of streaming services is broken. It’s a commodity⁸ and the only reason to have the service is if you truly like content that hasn’t entered the zeitgeist like House of Cards or Game of Thrones, or if you really feel compelled to have water cooler conversations about the latest episode on Monday morning. Disney Plus will be an interesting service to follow. It seems like it has the hallmark of success with so much A-list content, and if you’ve been paying attention Star Wars is everywhere.

Update: I forgot I had wanted to contrast video streaming with music services (I wrote this post quickly because otherwise I wouldn’t have written anything). Music services like Spotify and Apple Music are perfect for platform lock-in. It makes sense to pay $10-$15 a month.

Amazon took the pain away of not finding the book you want in bookstores. Similarly, Netflix took the pain away of settling for some B-list movie because there were no other options in the Blockbuster you were browsing. ↩︎
https://www.forbes.com/sites/greatspeculations/2019/07/19/netflixs-original-content-strategy-is-failing/ ↩︎
https://en.wikipedia.org/wiki/List_of_original_programs_distributed_by_Netflix ↩︎
https://en.wikipedia.org/wiki/List_of_original_programs_distributed_by_Amazon ↩︎
https://theoutline.com/post/1239/youtube-has-a-fake-peppa-pig-problem ↩︎
Services vary from $8 for standalone Amazon Video to $15 for HBO Now, but let’s just say the services on average are around $10, and for three services for 12 months it is around $360. ↩︎
Do it! Right now! https://watsi.org ↩︎
It’s a commodity in the sense that people aren’t loyal to a service (except perhaps Disney Plus!) but instead just want to “watch something.” ↩︎

You Can't Top Bitcoin

Fri, 03 May 2019 00:00:00 +0000

I consider myself a crypto altruist; I don’t subscribe to any sort of maximalism of any cryptocurrency. I want to see everyone have access to this technology regardless of the particular implementation, so long as it is entirely open source and has been peer reviewed by computer scientists¹.

When I began studying cryptocurrencies a few years ago, I recognized that they were flawed. They require untold amounts of data to store the timechain, and perhaps worse the computing power and lengthy confirmation time make monetary transactions as we know them impossible. Digging through research papers, I learned how challenging of a problem it was to ensure speed, security, and decentralization.

It seemed like Bitcoin – and especially Ethereum – were unsustainable in terms of space and computing power required, not to mention they practically violate the ethos of a decentralized money system with mining being highly concentrated². To me, it seemed the path forward was something that worked well on mobile devices, or else how would cryptocurrency be adopted by billions of people? Since then a few solutions have surfaced that intend to work on mobile or by leveraging trusted computing centers in conjunction with a mobile solution. I’m not sure if any of these will ultimately succeed, but few if any people in the crypto community seem to support them. For reference, Celo is approximately what I’m talking about here, if it works.

Fred Wilson said a couple years ago that he wouldn’t be surprised if Ethereum surpassed Bitcoin in value³. This makes sense as Ethereum has the potential for people to develop applications on the platform that grow in value. Ethereum has an excellent ecosystem and the tooling is fantastic⁴. It is likely Ethereum will end up being of some non-trivial value in 10 years because of the wealth created via Ethereum apps. I don’t doubt, however, that there will be plenty of competition. Programmers are unfortunately fickle. Ruby on Rails was the hottest thing 10 years ago, and now we have Go, Rust, Swift, Node, and dozens of frameworks that associate with those⁵.

Imagine a Magic the Gathering is built on Tezos or EOS or Ethereum, &c. It has a lot of followers that know all about the cards and play the game, creating value. A friend sends a card (token) to my wallet as a gift. Thank you, but I’m not into Magic the Gathering. My wallet tells me it is worth $250 – awesome! Do I put it up for sale, hold on to it, or buy more Magic cards in hopes they will be worth even more one day? As a gift I may hold on to it, at least for some time. But likely I’ll sell it or trade it for something I am more familiar with. Cash? Gold? Those things are minted in central banks and possibly mined in asteroids⁶. If I want to buy a used Apple Watch or I want jewlery, sure. Instead maybe I trade it for a token that is pegged to a stock with solid fundamentals, or maybe I put it in a store of value that has proven historically to increase in value over the long term. Those seem immensely easier options. And if you are someone who dismisses crypto, read Green Eggs and Ham and then try donating to Watsi or the Internet Archive with it; less friction always wins.

Bitcoin is a store of value. It has brand recognition. The social challenge of overcoming its dominance is like campaigning Kay Jewelers to be more valuable than Tiffany. Fork market caps pale to Bitcoin, and even if they succeeded you’d still own it. Exceptional value is also Bitcoin’s biggest problem. Lightning was designed to enable instantaneous low fee transactions, but you need to stake Bitcoin. As an appreciating asset why stake it? People who were early to Bitcoin and have spent it might casually participate, but the users being onboarded today are unlikely to do this. In the US there are tax implications which is a huge turnoff to spending an appreciating asset like Bitcoin. This is just the reality. Perhaps legislation changes the landscape here.

The development that has me convinced Bitcoin will continue to be a store of value and the dominant of all cryptocurrencies is the rise of stable coins⁷. Stable coins are ideally pegged to the US Dollar, and while there are small fluctuations I suspect there will soon be some kind of exemption for taxes in the US. Stable coins will be used for day to day transactions. No rational person is going to want to trade their Bitcoin if they think it is going to continue to rise in value!

People will transact daily in stable coins, they will buy virtual goods on Ethereum and it will wax and wane in value, new networks will emerge and also wax and wane, but Bitcoin will remain the strongest store of value.

By “computer scientists” I don’t mean people who hold degrees necessarily, but anyone who is capable of building a blockchain from scratch. A consensus on the merits of the technology is what matters, not nitpicking. ↩︎
I realize this is a contested point. Ideally you’d have x nodes and 1/x nodes odds of earning a reward, but I digress. ↩︎
People misquote him as saying “he thinks it will” when in fact he said he wouldn’t be surprised. ↩︎
I built Glossy.cards, an ERC721 creator to sell collectibles for free, using Golang with the assistance of geth, web3, Metamask, Infura. (Sure, infura is a crutch, but Ethereum is still in development). Was impressed so much production-ready infrastructure was built in such a short time. ↩︎
Would be great if y’all just decided on one language for professional purposes and used the rest for hobby. I vote for a LISP; Clojure is ok I guess. Risp? Maybe something that supports tail call optimization. ↩︎
Seems like science fiction, but so does decelerating rockets and landing them on small pads in a rough ocean. ↩︎
It’s not lost on me that Bitcoin could one day be measured in terms of stable coins which themselves will have intrinsic value. ↩︎

Grocery & Meal Delivery

Sun, 23 Apr 2017 00:00:00 +0000

I still go to the grocery store in part for the experience and the discovery process. Often times, however, I’ve wished I could just skip the store and have the groceries appear within a day. The problem with that is then I need to sign up for some delivery service and still spend time paging through an app to find dozens of necessities.

Retail as end-point for physical logistics is often doomed. Retail as experience, curation, service, discovery is a a different matter.
— Benedict Evans (@BenedictEvans) September 5, 2017

I have tried a few services including Instacart and Amazon Now, and both worked out fine except for the economics. On top of the cost of groceries is an added percentage, fee, and tip. What’s hard for me to understand is why these companies aren’t able to tap into the fixed costs of a traditional supermarket. While supermarket net margins tends to be around 1.7%, gross margins are in the 20% range.

There are over a dozen Safeway supermarket stores in San Francisco

Considering the costs of real estate, utilities, employees, logistics, and insurance, it seems there should be enough for this service:

Instead of 15 stores, situate 3 warehouses in the vicinity of Excelsior, Dogpatch, and Inner Sunset (i.e. highways and center of mass)
Automate the warehouses so that pallets come off the trucks from supplier and are robotically transported to their row and bin; individual items are robotically retrieved to fill a shopping bag
The store employees are now delivery drivers. Bags are robotically loaded into the vans so the driver just has to get in and follow an efficient route and unload the bags to homes, offices, car trunks, &c.

The above seems reasonably doable, so I wonder why no one does it. Even if certain aspects aren’t automated from the start, it can be done over time.

What about perishable items? I think instead of bags reusable crates could be used with built in icepack. Since I consistently demolish large boxes from Amazon I’m not sure why they haven’t gone reusable yet, though with their own fleet it seems likely to be on their roadmap.

What about stolen groceries? I think stolen groceries would be about as likely as stolen Amazon packages. Some neighborhoods may be more susceptible, especially in San Francisco, so having items delivered to work or a car trunk may be a solution.

Lastly, if the vision weren’t already over the top, I think combining groceries with a “chef meal” service makes sense. That is, you get chopped vegetables and other ingredients along with a recipe so you can make a fancy dinner with ease. Maybe this would be a premium service, but hopefully far cheaper than current services with this offering.

Hyperloop May Create New Cities

Sun, 27 Nov 2016 00:00:00 +0000

Hyperloop concept from the original SpaceX white paper.

Were I to create a hyperloop company I may call it R-Loop to emphasize keeping out the hype. If I was lucky maybe one day it would be known simply as “The R” just as elevated trains in Philly and Chicago are known as “The El.”

It’s been just a couple years since two major hyperloop companies have been in existence and the outlook is dubious. Hyperloop One had a cofounder breakup which is often an ominous sign in startups, and there hasn’t been much news of the other company. Even if these companies don’t succeed someone will take the opportunity and figure out how to execute.

In a place like the Bay Area, the hyperloop could mean an alternative to local governments who refuse to build new housing. Why commute by car from Berkeley over to San Francisco which could take over 45 minutes each way, when you could hyperloop into San Francisco in 20 to 30 minutes? Even assuming self-driving vehicles eventually earn a comparable time, housing costs are still problematic.

Brisbane, a city adjacent to San Francisco, recently rejected a proposal to build thousands of new units on this large plot of land which has two 101 exits and a Caltrain station. Driving through it at night you’d think you were in wine country.

I asked my friend who is a city planner from MIT what San Francisco could do to relieve the problem. He replied that properties need to stop being used as unoccupied investment vehicles and the city should let loose on high rise construction, particularly in former industrial areas like SoMa, which allows for a compromise to preserve the character of neighborhoods like Noe Valley. So far few Bay Area cities are making efforts to allow for vast new home construction.

I realized, like many others probably, that hyperloop could solve this problem by connecting San Francisco to more rural areas and building an expandable city there. I ran this idea by my friend who told me newly planned cities almost never succeed, and it is far better to improve existing cities. But you know, Wall Street traders from Wharton and venture capitalists from Stanford and city planners from MIT excel at learning from history; one thing they have in common is they can’t predict the future very well.

The kind of effect the hyperloop has on society may be predicted by recalling the effects of new technologies like the railroad or air travel, but history can’t tell us what happens when we have multiple technologies that are only a decade or two old working together. Just as Tesla was able to mark billions in preorders for the Model III, we could use the Internet to propose a new city for which a hyperloop company has worked out a route. A developer could take preorders for home plans, and the hyperloop company could get to work. Some attrition may occur, but also augmentation once the city becomes a reality. Would people buy a home site unseen? For the current generation of homebuyers the answer may be yes. Everything is bought online today, even cars. What is more important is if friends are also buying in the city and how easy it is to get to San Francisco or Portland.

Being in such a new city provides more opportunity. Let’s assume the new city is around Eureka, CA. Someone who lives there decides to leave their San Francisco job for a job in Portland, so their commute goes from 20 minutes to 30 minutes one way. After work on Friday they decide to head to San Francisco to get an Impossible Burger at Cockscomb with old friends. No big deal, just a 50 minute hyperloop ride from work and a 20 minute hyperloop ride back home.

Hyperloop could create new cities between San Francisco and Portland.

All of this is predicated on a hyperloop company successfully creating the technology and making money. But I think once that happens, public and private sectors will recognize the incredible opportunity and devote resources to seizing it. People will buy land, create crowdfunding campaigns, secure bank loans, new homes and parks will be created, new cafes and businesses will rise, and the quality of life will be among the best.