The biggest difference between what I read and what my parents read is that my parents subscribe to newspapers, while I subscribe to individual people. This simply wasn’t possible before the internet, but it also took easy online payments (like stripe) to be fully realized.

I get my tech strategy from one person, China-US news from another, California city planning from yet another. My architecture, design, venture capital, bitcoin, autonomous cars, bay area wind forecasts, and startups readings all come from individuals.

This is interesting, but how do these people get paid? Blogs break the typical payment model of buying the daily newspaper or subscribing monthly. Instead, I’ve found that I start by reading one or two free posts a week from a person. Once I’m hooked, I’m much more likely to purchase an annual subscription for around $100.

If each person I read can be the best at what they do and get 10,000 people paying $100 a year, that’s $1 million in annual revenue. If 10% of weekly readers convert to paid, that means this individual might only need 100,000 people reading their content. Granted, that’s a lot of people, but it’s not anywhere near the order of magnitude needed for an advertising-supported newspaper. 

However, not everything is perfect. There’s still an issue with discovery. None of these blogs are on facebook or medium or twitter. I’ve generally found each because another one of my blogs (or friends) will recommend them. I wish there were a better way to find these people.

I’ve always felt that cities have two important structural advantages over Lyft and Uber. First, cities can run transit at a loss, and second, cities control the design of the streets. From SFMTA:

The SFMTA got the green light to paint red transit-only lanes in 2012 from the California Traffic Control Devices Committee and the Federal Highway Administration as an experimental measure to improve driver compliance with transit-only lanes. If proven to be a success, the red transit lane treatment could become a national standard for transit-only lanes, similar to green bike lanes.  The red paint treatment is already recommended in the National Association of City Transportations Officials Urban Street Design Guide.

For the time being, venture funding allows Lyft to run at a loss just like transit, but no amount of venture money can change the fact that cities can leave Lyft in traffic while buses are given dedicated lanes. Or can it? From CNN:

With the help of architecture firm Perkins+Will and transportation consultants Nelson/Nygaard, the ride-sharing company has reimagined a street for the future. The teams reenvisioned a concept for Wilshire Boulevard in Los Angeles, a notoriously car-centric city. The average L.A. driver wastes over 100 hours a year sitting in traffic.

Wilshire Boulevard’s design is typical of the city boulevards – there are currently 10 vehicle lanes, including two lanes that buses share with vehicles.

But Lyft’s design includes trees, protected bike lanes, a loading zone for ridesharing vehicles, three narrowed lanes for vehicles and lanes for autonomous buses. The concept also rewards buses with exclusive travel lanes.

In a classic case of “if you can’t beat them, join them”, Lyft is pushing its own vision of the future of city streets, complete with features that make Lyft better like loading zones. I expect Lyft to push this hard and get users to add pressure on governments. As long as a Lyft is a superior experience to the bus, they might have the popular support to change street design to create their future.

Detroit is following in China’s footsteps. From the WSJ:

Developers and designers complain that, like many cities, Detroit’s onerous and outdated rules make it too difficult to rebuild or repurpose long-neglected retail areas. To try to reduce those obstacles without a time-consuming and expensive rezoning process, the city is proposing a handful of “pink zones,” whereredtape will be cut to help small developers and entrepreneurs open new businesses and revive aging commercial strips. The goal is not to eliminate zoning but to ease some of the constraints faced by new projects, like minimum-parking requirements orenvironmental-impactreports.

This model of development, reducing red tape but only in certain geographic areas, reminds me of China’s special economic zones. From ThoughtCo:

Since 1979, China’s Special Economic Zones (SEZ) have been beckoning foreign investors to do business in China. Created after Deng Xiaoping’s economic reforms were implemented in China in 1979,  Special Economic Zones are areas where market-driven capitalist policies are implemented to entice foreign businesses to invest in China.

These zones were successful not only because they changed the rules, but also because they provide a strong focusing power. If you were someone who wanted to do business in China in the 1980s, your choice was clear, go to one of the special economic zones. This concentrated all of the economic activity rather than spreading it out across all of China, and concentration is the key to what makes geographically-dependent things (like cities) valuable.

What’s also interesting about these policies is that they are examples of experiments that do better when kept smaller rather than larger. This is partly because a critical mass of concentration is necessary for the changes to be effective. Choosing 8 cities rather than 4 would have spread out the influx of investment, and probably would not have been as effective. Similarly, reducing red tape across all of Detroit seems unlikely to spur much development in a city that is already avoided by many developers.

This means that for an experiment like this, there is a tradeoff between keeping it small enough to achieve critical mass, and large enough to achieve critical mass. The conventional wisdom only recognizes the minimum threshold of critical mass, not the maximum threshold.

My friend recently joined the Grammarly team. One really interesting comparison to Grammarly is the set of tools that help programmers write better code. Programmers have code linters that check for stylistic problems, compilers that check for invalid code, templates that help with common patterns, and more. Grammarly basically does this for the english language.

Another example of this is Dropbox, which took tools well known in the programming community like git and rsync and made them easy enough for anyone to use. 

I wonder what tools and best-practices are hidden in other fields that would be good candidates for simplification. 

I’m struck by how visible technology is in Sidewalk Lab’s Street Life After Retail article. Every single one of their visualizations has some questionably-useful piece of technology in the foreground, including holograms, 3d printers, mini delivery robots, and robotic arms. While this is certainly a possibility–just look at how visible cars are–it’s not at all a certainty. 

Furthmore, I’m struck by how business-like the community-led organizations they mention seem to be. Their examples include a senior center, public kitchen, montessori school, family health (doctor?), bike share, child care, and game room. These all sound like businesses to me. What’s really striking about this depiction of a street is the lack of cars and the presence of people. The transportation and delivery mechanisms must be hidden from view. Subway? Roads with self-driving cars on the next street over? A daytime ban on delivery drones? Hard to say.

This leaves me asking a different question: What should be centrally planned and what should be left to businesses and individuals? In today’s cities, the streets and sidewalks planned centrally, leaving almost everything else to individuals. It’s time to re-evaluate that balance. Do central planners simply need to change roads and sidewalks for this new world or do they need to take on more (or less!) responsibility?

Both founders and artists need to sell the story of their future success. Early angels and early art buyers give their support because they hope this future success will materialize, greatly increasing the value of their investment.

Arthena realizes this and wants to become a VC for art, letting other people buy into their fund, which Arthena turns around and invests in risky artists. 

Angels like Jason Calacanis also realize this and promote the startups they’ve invested in similar to how one might promote an up and coming artist.

Given that it’s always possible to spend more and get more healthcare, some interesting questions arise:

• How much of a country’s income should be spent on healthcare?

• How much of your income should be spent?

• If everyone suddenly made twice as much money, but costs remained the same, then do the answers to the first two questions change?

Amazon wants to open a second headquarters. This means a distributed team, and I’ve found that not every team or project lends itself well to being distributed.

When I was traveling with Google a few years back I was stunned how every single office seemed to have a Chrome team. When I returned home, I looked for another team at Google that was similarly distributed, and couldn’t find one. 

Chrome can be distributed because it’s built on properties that work well without much communication. Namely:

• the goals are clear. Speed, simplicity, security

• who to ask for permission to contribute is clear: there is an OWNERS file in every directory and owners comments in every file. 

• the review process is public: https://codereview.chromium.org/

• the code is public and immediately available: https://www.chromium.org/

This structure makes it easier for people to contribute code without a bunch of communication overhead. From what I saw, Chrome engineers tended to communicate less than engineers on other teams at Google. I think this is a direct result of their organizational structure that makes contributing possible without much communication. 

Earlier this year I tried 30 days of yoga at home. My instructor never got tired, or expensive, or scary because she was a youtube video. At the end of the 30 days, I decided that I’d rather put more time into rock climbing than yoga. But I’m glad I tried, I’ve wanted to try yoga for years but never mustered up the energy to do it until now. 

I think this illustrates two important learnings.

First, that having a set time period associated with a new behavior makes it easier to commit to that new behavior even when it’s not obvious that it will be good.

Second, that by introducing a decision sometime in the future, you avoid the trap of continuing to do something that’s not that helpful just because you started and never really considered whether it was worth continuing.

Earlier this year I talked about how making streets feel more dangerous can actually make them safer. 

Thinking, Fast and Slow by Daniel Kahneman gives an explanation for why this might be. Basically, the high =-energy state of the brain is activated when danger is perceived, and this state is better at avoiding pitfalls that fool the low energy state that the brain is in most of the time. Daniel calls these system 1 (low energy) and system 2 (high energy). 

Interesting to ponder: what happens when all of the streets feel dangerous? Do fewer people drive because of the stress it invokes? Do people become immune to this? Do they become fatigued? It’s one thing to be in a high energy state for a few seconds at an intersection. It’s quite another to be living in it for hours. 

Computers can recognize traffic signs, but it’s possible to fool them by adding some tape. Similarly, computers can understand speech but can be fooled by sounds that are too high for humans to hear.

An active and growing area of research is how to design systems that can’t be fooled so easily like this. This will be critical expertise going forward, just like how network security is today. Great for up and coming security researchers wondering what to specialize in. 

Another interesting angle is that the very fact that these systems can be fooled by things that wouldn’t fool a human means that these systems perceive the world in a different way than we do. These are the special effects and optical illusions of the computer world, and they look nothing like what we’re used to. 

I hate feeling like I’m missing out. This feeling is unhelpful because there’s no way I can do everything and be at every event. But somehow it persists. 

Originally I thought that an unstructured life leads to more FOMO, because the gaps and decisions about what to do next give time to wonder what I’m missing out on. However, I think this misses some of the nuance.

For me at least, the core feeling of missing out is triggered when I feel present, but not a part of, an activity that I want to participate in. The classic example of this is working out of a cafe on a Saturday. Everyone else is enjoying their weekend, but here I am, working. 

I tested this hunch last weekend by working out of my Wework office. I had drastically less FOMO and got vastly more done. The few other people crazy enough to be here on a Saturday were also working hard, and I surprised myself by staying the entire day.

FOMO is one of the worst feelings because it can give a good activity a sour taste, making life dramatically less satisfying. If you have your own methods for avoiding FOMO, I’d love to hear from you.

When I first started working on my startup, I set up a desk in my bedroom at home. This was cheaper than coffee shops and way cheaper than office space, and I figured that cost was the most important factor.

I knew something was wrong I started sleeping badly. I’m usually blessed with an easy and restful sleep, but suddenly I couldn’t fall asleep and woke up feeling awful. I guessed this might be something about not having a separate space for work, so I tried coffee shops.

If you’ve ever tried to work at a coffee shop you know that it’s really awful. Interruptions every minute, with loud noises, music you can’t control, and the expectation to keep buying food. Add to that a tiny laptop screen and a different seat every day and it’s a recipe for disaster. I don’t know how anyone can do it.

Finally, I bit the bullet and spent the money for a private office at a Wework. My productivity went through the roof, and I started sleeping again. It’s truly amazing. I estimate I get 3 times as much work done, and still have more energy in the evenings and weekends for studying Chinese and machine learning. Previously I was too exhausted to do these kinds of intellectually-intense activities outside of work. 

The key qualities for a workspace are:

• Zero interruptions

• Same space every day

• Separated from home and other non-work spaces

• A comfortable environment (temperature, noise, smell, lighting, furniture, bathrooms, etc…)

As my company grows, I’d love to offer everyone her own private space. That might be against the wisdom of open-desk layouts, but it seems critical for productivity.

I use moleskin notebooks for most of my thinking and ideas. They’re useful because they’re durable and I can add sketches in addition to just text. They’re also way better for notes in 1 on 1 meetings, because they demonstrate a professionalism missing from notes taken on phones and ipads. 

Unfortunately, once the information is in the notebook it’s pretty much locked in there and unless I spend lots of time devising systems to index and retrieve it like Darwin and other scientists did back in the day. But I feel like a computer can do this faster and better than me.

I’d like to see a service where I can mail my finished notebooks/journals. They’d be scanned, with an app to browse and search. I’d also like to be able to add digital snippets and ideas to this app so that it’s all in one place when I want to search for something that I’ve written down. 

Does anything like this exist? Some simple google searches returned lots of photo scanning services but nothing that seems like a good fit for journals. 

I’ve noticed that when I ride a bike share bike, everyone gives me a lot of space as they pass me. This doesn’t seem to happen when riding my road bike down the same streets. My guess is that people assume bike shares are used by less frequent and less skilled riders.

This serves as two important reminders. First, that signaling matters, but also that signaling can be wrong. If possible, better to watch actual behavior. Watching me cycle a few blocks would probably better identify my skill level than assuming based on the type of bike.

The same holds with business. Think that a startup is hot? Watch how it grows over the next few months. Think that a salesperson is stellar? Look at the sales numbers, rather than how she speaks or how confident she is.

When I started offering a subscription to artwork through Home Gallery Project I assumed people would be rational when considering the price. Big mistake. 

First, price anchors are very real. Customer after customer anchored their purchase of artwork to other purchases in their home, like their couch, TV, and bed. However, when considering a monthly subscription to artwork the anchor was always Netflix. $7.99 per month is just too little to pay for shipping, let alone the artist.

This has an important implication: people with more expensive furnishings also buy more expensive art. If as an artist you want to price your painting at $5,000, sell it to someone who has bought a $5,000 couch.

Next, people in apartments guessed (often incorrectly) that they would only be in their apartment for at most another year. This means that when considering an annual plan, they compared it to purchasing artwork directly. And it turns out that owning the art is important, even if it will be thrown out after a year.

The elasticity of demand is also interesting. While a specific piece might be a must-have for someone, an unknown piece is at best a nice to have. This means that there is significantly more price leverage once someone has decided that they want a specific piece. With a subscription that isn’t tied to any specific piece, the leverage simply isn’t there. 

The successful art sellers all seem to have the same playbook. Build up a list of people interested in art who are in the same socioeconomic bracket and similar tastes, then periodically invite them to events. At these events, have a great salesperson who can convince someone that they want one of the pieces.

San Francisco’s bike share is expanding by an order of magnitude from 700 bikes to 7000 bikes. As new stations are added the value of the system grows faster than a linear rate. This is a classic example of network effects. But today, I want to focus more on some other aspects of bike shares.

Bike shares, at least in the USA, have two important properties. First, they are typically free at the time of use, relying on annual subscriptions. Second, they’re mostly immune to traffic because it’s so easy to go between lanes and around stopped vehicles. And because biking is three to five times faster than walking, bike shares open up a wide range of trips that were too expensive to lyft and too slow to walk.

This makes it easier to have a business that’s off the main strip, as long as it’s worth going to. Even if the surrounding streets feel unsafe to walk on, a bike might be fast enough for this not to matter. 

And it makes each station a hub. If I ran almost any kind of business I’d want a bike share out front. More eyeballs, and more convenient.

I expect this to be a common perk that companies offer their employees. I also expect the next generation of bikes to be electric, making hills and sweat a thing of the past.

This is a great direction for dense cities to move into, especially for larger cities like New York and Taipei where the urban core can be many miles across. 

Swappa makes it easy for developers to purchase older hardware for testing. This works best for ongoing testing week after week.

For rapid prototyping, another approach might work: purchase and return. I just returned an iPad pro after spending the past week prototyping a feature that I thought might improve the user experience of study groups. After testing prototypes with real users, I decided that other avenues were a better use of my time and money.

But what does a product manager actually do?

I’ve gotten this question a lot, from friends, family, mentees, and even coworkers. If you really want to know, go read Inspired. It’s a decade old, but has aged well and contains everything I wish I’d known about product management when I started at Google in 2012.

When Taiwan was designing their new health system in the 90s, there wasn’t yet consensus on how it should work. And as anyone who reads Dilbert knows, when the decision maker doesn’t understand the space, bad decisions are made. So, how to educate a decision maker?

“[We] had a conference in Taipei, over three days, and all these [experts] debated each other. And I required cabinet ministers [aka decision makers] in Taiwan to chair the various sessions, like the minister of finance, the minister of planning, of health. See, when you chair a meeting, you have to sit through the whole thing and listen. That’s how you educate decision makers.”

(from The Healing of America)