Go West, Young Man

If you’re in tech, you need to move to San Francisco. The advice holds especially if you’re a student, not having dug your roots too deeply anywhere yet.

There are several reasons why you want to come here:

The Opportunities
All the coolest companies you’ve heard about are here. In a radius of roughly 30 miles, you’ll find Airbnb, Amazon, Apple, Facebook, GitHub, Google, LinkedIn, Salesforce, Slack, Stripe, Twitter, Uber…  And all of them are hiring (shoot me an email!). There is no place in the world with a comparable density of exciting, impactful projects, maybe to the exception of Beijing — but certainly nowhere in Europe.

Any engineer here with even a little experience at a startup receives a constant stream of phone calls and emails from recruiters. I’m not saying this to show off: ask anybody from San Francisco to show you their inbox. 1-3 recruiting emails per week is normal.

Beyond the opportunities you’ll receive personally, tech is currently reshaping the whole world — and San Francisco is at the center of it. I feel lucky every day, getting to witness what’s happening here. It’s often compared to the renaissance in Florence during the 15th century.

Some people get super annoyed when you say this. They’ll tell you talent is everywhere and curse about those damn Americans who think they’re the center of the world. I don’t mean to be insulting, and I agree talent and energy are everywhere. It is true that several tech ecosystems around the world are blooming — including Paris. But statistics like the amount of Venture Capital invested support the thesis that nowhere comes close to the Bay Area yet. Which leads me to…

The Money
You’ll earn more here than you would anywhere else. The difference is especially stark if you don’t come from the US. The entry salary for a software engineer who’s straight out of school hovers around $110k per year. If you’re good, it’s not uncommon to see software engineers with 5-6 years of experience make $300k-$400k per year in total compensation (which includes your base salary, yearly bonus and stocks). I know several people making this kind of money in big companies. Looking at it from the outside, getting into Apple / Google / etc looks intimidating, but it’s a lot easier than you’d think.

Compare these salaries to what you’d get anywhere else in the world. Straight out of school, a software engineer in France would be lucky to earn 45k€ ($52k). Interns make more than this here — those in Paris will make around $1k per month or less.

And yes, life in SF is much much more expensive than in, say, Paris. But in my experience the difference in compensation more than makes up for it. Don’t take my word for it: check out Numbeo, a crowdsourced database with the cost of living of a bunch of cities in the world. Its page comparing San Francisco to Paris states that you need to earn 46% more to live the same way in SF as you would in Paris. So these $52k in Paris translate to $76k in SF — still less than many interns here. And there is no way you’d ever get these $300-400k packages I mentioned in Europe.

The People
If you decide to move here, it means you probably have more ambition than average. Consider that everybody else you’ll meet in the Bay Area will share that with you. The self-selection effect is strong, and you’ll met the smartest, most energetic and ambitious people you’ve ever seen. The phenomenon tends to feed on itself — enthusiasm is contagious.

When I arrived, I met a guy who told me “the difference between French and San Franciscan entrepreneurs is that the French build startups to have a job, San Franciscans to change the world.” The insult stung, and was definitely an exaggeration, but the second part contains a little bit of truth. It really is surprising how big Californians allow themselves to think. Entrepreneurs here will tell you they want to cure cancer, colonize Mars, or disrupt the trillions-of-dollars car industry. They’ll make you reconsider what you can aim for.

California
California is stunning. It’s sunny year long — some people actually say they miss seasons, though I certainly don’t. You can drive just 4h and see some breathtaking landscapes in Tahoe or Yosemite, or just 1.5h North or South and find yourself in the middle of gorgeous woods. The food is great, San Francisco’s streets surprisingly walkable (something rare in America), and people very nice and welcoming.


But… The Rent
Of course, San Francisco’s got a couple of downsides too. Everybody complains about the rent, for good reason. Because of some absurd zoning regulations forbidding anybody to build anything in about 70% of the city, there’s a massive housing shortage, translating into some of the highest rents in the world (and some of the lowest skylines for a city that prosperous). Expect to pay easily $2500 per month for a studio. If you find a roommate, you can find a small room for about $1600/month (as of 2018). Again, the salary more than makes up for it if you’re a software engineer, but you’ll have less living space than you would anywhere else.

The Monoculture
People complain about what they call the “monoculture” here. The high-rents chased away anybody who couldn’t earn the big bucks, like artists. If you hang out in New York or Paris, you’ll meet lawyers, teachers, artists… Here it’s all tech, tech, tech. Not necessarily a bad thing if, like me, tech is your passion. I for one love hearing stories about how my favorite products are built “behind the scenes.” But some people crave more diversity of conversation topics.

Starting a family
The high cost of living makes it very hard to start a family too. I sometimes feel like SF is a game where you have 10 years to become rich. If you win, you get to stay longer and start a family. Otherwise you have to leave and the mayor personally comes to kill your tiki torch. Statistics do show that the Bay Area is one of the regions with the fewest children, relatively to its population, in all of America.

The Public Infrastructure
San Francisco is the poorest managed city I’ve ever visited. The homelessness rate is very high, the sidewalks disgusting (if you live or work near the center, you’ll see human feces and syringes on the ground almost every day), and you often find tents lined up in the street. I don’t mean to blame the homeless; their situation is absolutely tragic, and is largely a result of, again, the artificially inflated cost of housing. But it’s not just the homeless. Roads are in a terrible state, public transportation is slow and unreliable, there is a lot of traffic, and even the electric grid is bad (there can be occasional power cuts during the couple of weeks of rain every year).


A pattern I’ve noticed is that newcomers here tend to fall in love with the city at first sight. That “honeymoon phase” will last 1-3 years. After that, they stop noticing the kind of people they encounter and the opportunities they’re exposed to on a daily basis here, and start minding the rent, bad public transportation, and human feces.

But the good news is: if you don’t like it here, you can always leave! 3 years in SF will not only have been an amazing experience you’ll always remember; they’ll pay dividends for your whole life. I know some people who went back to France after 6-7 years (often to start a family), and they told me that having a “Silicon Valley” stamp on their resume earns them a 10-20% premium on the offers they get in Europe.

And it’s surprising how quickly one can build a network here. I met someone the other day who used to live here and moved back to Paris. Common acquaintances kept popping up in the conversation. I was astounded when he told me he’d only been living here for 3 years, and left more than 10 years ago! Maybe it’s true that “you can check out any time you want, but you can never leave.”


The Roots
“But moving to San Francisco is so hard!”, I often hear. And it’s true — especially getting a visa can be quite an ordeal.

My objective in this post is to convince you that you need to move to SF. There are already tons of articles out there on your options as far as visas are concerned, so I won’t get into too much detail about that. I’ll just say that if you qualify for it, the J1 visa is probably your best bet to get your foot in the door. O1 visas are great too, and easier to get than most assume.

Nobody said it would be easy, but it’s more than worth it. And if you think it’s hard now because you have a lot of things going on wherever you are, tell yourself that it’s only going to get harder with time, especially if you’re a student. Roots only get deeper. When I moved here, I jeopardized my small development agency (ended up closing it shortly after moving — remote is hard) and my relationship with a girlfriend I’d moved in with and had been dating for 3 years (broke up shortly after moving too — remote is hard!). All for a place where I had no network, and no plans.

I tried hard to convince a couple of friends to come. One of them is a technical genius, and I don’t use the word lightly. He’s like Rainman smart. These friends told me something like “I agree, I’ll come eventually, but I can’t now. I’ll come when the time is right.” 6 years later, as I predicted, none of them have come, and their roots have gotten deeper. They built companies or careers, took mortgages, got married…

Today, a couple of them admit they made a mistake.

They wish they’d seized their chance to go West.

Mind the Moat, a 7 Powers Review

The Lindy Effect has become my top heuristic to decide what to read next. This phenomenon describes how some things tend to live longer, the longer they’ve lived — to speak plainly, if something has stood the test of time it must be important. When you’re tempted to read a book published only 2 years ago, chances are you’re just sitting downstream of the author’s self-promotion efforts. Not that there’s anything wrong with self-promotion, but it doesn’t correlate well to the quality of the book.

But one must know when to override their favorite heuristics, which is what I did with 7 Powers. Written by Hamilton Helmer and published in 2016, the book came to my attention as a recommendation from an avid reader who I knew shared my preference for tried-and-true publications. It became more compelling as I saw its glowing reviews from folks I respect like Peter Thiel, Patrick Collison, Reid Hastings, Daniel Ek, and Mike Moritz. But what really convinced me is that it’s about competitive strategy, and more specifically moats. Moats fascinate me, and I know my understanding of them still has important deficiencies.

Moats are those barriers that protect your business’ margins from the erosive forces of competition. It doesn’t matter how revolutionary your product is: even if it literally changed the face of human civilization, you’re going to get nothing for it if anybody can sell it too, arbitraging profits away. The whole surplus of the revolution will go to the consumer, and none to you. Charlie Munger phrases it this way:

When we were in the textile business, […] one day, the people came to Warren and said, “They’ve invented a new loom that we think will do twice as much work as our old ones.” And Warren said, “Gee, I hope this doesn’t work because if it does, I’m going to close the mill.”

[…]

He knew that the huge productivity increases that would come from a better machine introduced into the production of a commodity product would all go to the benefit of the buyers of the textiles. Nothing was going to stick to our ribs as owners.

[…]

And it isn’t that the machines weren’t better. It’s just that the savings didn’t go to you. The cost reductions came through all right. But the benefit of the cost reductions didn’t go to the guy who bought the equipment. It’s such a simple idea. It’s so basic. And yet it’s so often forgotten.

Silicon Valley tends to reduce moats to network effects. But a simple look at Fortune 500 companies will tell you that there exist some other very powerful moats out there. The success of commodities sellers like Kraft Heinz, Procter & Gamble, or the Coca Cola Company has certainly shown me there was a gap in my understanding of them.

7 Powers did help fill that gap. It’s a simple, straight to the point framework with enormous explanation power. Let’s see what those different “powers” are.

1. Economies of Scale

This one is pretty straightforward, but has become so well-known it’s almost overlooked. If you have very low marginal costs, you should benefit enormously from scaling — an advantage you should juice for all it’s got. That leads to companies trying to “get big fast,” not because of scale for its own sake, but because it can serve as a weapon and bring enormous profits.

Helmer uses the example of Netflix (him and Netflix CEO Reed Hastings seem to be best friends or something). One pivotal strategic moment for Netflix was their investments in originals. Before that, they had to negotiate exploitation rights on a case-by-case basis. Their holders were largely oligopolistic and hence in a position to suck all the margins out of their business. Spotify had the same problem with labels, and it’s not super fun to have to hand 70% of your revenue to a handful of suppliers, before all other expenses.

Tren Griffin calls this “wholesale transfer pricing power“, and this is Porter’s supplier power at play: once one of your suppliers gets a monopoly over one of your necessary inputs, they’ve basically turned you into their serfs. You’ll still have to sweat to find customers, raise money, and do all the other unpalatable things that come with operating a business. They’ll get to keep the best part, which is the money. That’s one big reason why restaurants aren’t profitable: they’re in a commodity business, and their rent is a fixed cost that they have to pay every month no matter what. This is also why the most successful restaurants tend to own their walls.

Being #1 pays, especially in those industries with very low marginal costs. Suppose you’re Netflix. You have 100M subscribers, and each original costs $100M — so, $1 per subscriber. Now, consider your competitor Hulu, with 20M subscribers. Delivering the same value as Netflix costs them $5 per sub, 5 times as much! That means that you can allow yourself to spend more on marketing; better content; superior user experiences; more experienced people; and even be less efficient, and still beat them.

Economies of scale are the defining forces of the software industry. Marc Andreessen, paraphrasing Jim Barksdale, puts it this way:

Here’s the magical thing about software: software is something I have, I can sell it to you, and after that, I still have it.

Building software costs the same regardless of the number of people buying it. That results in a massive scale advantage, as described by Bill Gates:

At Microsoft, our only ‘hammer’ is software…. It’s all about scale economics and market share. You can afford to spend $300 million a year improving it and still sell it at a low price.

2. Network effects

Network effects are very à la mode these days. Either because some entrepreneurs are desperate to look like the next Facebook, or because they don’t understand network effects, they’ll tend to twist the word in all sorts of ways until it seems like their business has a network effect (if you squint really hard).

The most common misconception is that network effects = virality = positive feedback loops. But none of those 3 things are the same. The simple definition is that your business has a network effect if the experience for your customers gets better as more customers join in. Virality means that each user you sign up will tend to attract more users in turn. For example, those terrible online games that used to spam your friends’ walls on Facebook didn’t have network effects — they had virality. At the end of the day, your own experience as a player was the same regardless of how many other people played.

Network effects are a very important topic in tech businesses, and expanding on them is worth its own blog post — if you’re interested in digging deeper, I recommend this a16z deck.

3. Counter-Positioning

Counter-positioning is the practice of developing your business model such that incumbents have conflicting incentives preventing them to compete effectively. For example, competing with you could hurt one of their other businesses. As Helmer admits, it’s very close to Christensen’s disruption theory, with less of an emphasis on technology and “attack from underneath.”

One example of counter-positioning is what digital cameras did to Kodak. Most people blame those failures on the incompetence or complacency of the incumbent. “Everybody knew digital photography was coming! Those dinosaurs were stupid not to get onto it first!”

Theories like Christensen’s, and Helmer’s counter-positioning, propose an opposite explanation. The incumbent is unable to respond to the disruptive threat not despite but because of its expertise and business savviness. In Kodak’s case, the disruption was two-fold: with the technology, and the business model.

From a technological perspective, digital cameras sucked, even for the mass market. The first models I saw as a kid had a resolution of 320×240 (that’s 0.07 megapixels), terrible lenses, and could hold something like 20 pictures (on a floppy disk!). It was an interesting “toy” for the super early adopters market, but not for professional photographers, Kodak’s bread-and-butter. Of course, they knew digital photography would get better. But put yourself in the shoes of Kodak’s execs during the yearly budget planning. It’s hard to take away some precious dollars and talents from your cash cow, and give them to that inferior thing which your customers are scoffing at.

The gotcha is that those technologies are often actually superior on some other dimension, and can offer features that the status quo just could not match, even with all the R&D in the world. That often comes at a cost on some other dimension — which can be fine for the market segments that don’t care about them. Christensen says these segments are “over-served” and that the disruptor is “attacking the incumbent from underneath.”

The exclusive feature of digital cameras was developing your picture right after you took it and, most importantly, for free! Digital pictures would get sharper with time, but analog would never get free, unlimited roll. And this was a very big deal, especially for professionals.

My father used to own a publishing house with a dozen small newspapers. He lugged around _entire trash bags full of camera roll_to be developed, every week-end, spending tens of thousands of dollars a year on roll and development. When digital cameras became good enough, he immediately spent the big bucks on them. It was a shock to learn he’d spent $20k on a camera and a bunch of lenses, until he explained to me that it would actually pay for itself in just a couple of months.

Which leads me to the other kind of disruption, that of the business model. Not only were digital pictures low-quality, that “free roll” also made it a super lousy business for Kodak. They made most of their profits selling high-margin camera roll. That gave that division a lot of political influence inside the company — and Kodak got successful to begin with by cutting unprofitable lines of business, to instead focus its limited resources on cash cows.

That’s why it’s often said that Christensenian (which is totally a word) disruption is not technological, as much as a disruption in the value chain. Technology can change as much as it’d like, as long as your customers remain the same. Your company grew around satisfying these customers, so if a better way to serve them emerges, chances are your organization will be able to go after it.

But some paradigm shifts are different. At first, the new tech is completely uninteresting to your existing customers, and replacing your existing product with it might reduce your margins and result in a net negative. But, on the very long term, it grows in performance and satisfies more and more of your buyers, and the larger total addressable market often more than makes up for the lower margins. Those are the shifts that can bite you, if your natural answer is to stick to your guns and “flee upmarket.”

4. Switching Costs

Stickiness of your product can protect you from the forces of competition if it locks your customers in. You can milk that moat and even deepen it by up-selling extra features, integrations, consulting, overpriced training, etc… 

Helmer uses the example of SAP which, over time, can build deep roots inside your organization. Switching to another solution can be a months-long effort costing several millions of dollars, and much more in missed profits if done wrong.

Vendor lock-in is a big reason why enterprise software is so terrible and overpriced, other reasons including:

  1. The person who buys the software is not the same as the one who pays for it, and is not the same as the one who uses it. Per Milton Friedman’s classic matrix of “whose money’s being spent” / “who’s it being spent on,” that’s a sure way to buy crap for a lot of money. Sometimes, on the contrary, the software’s purchaser is over-incentivized on the measurable costs savings, at the expense of the immeasurable product quality.

  1. The software is optimized for must-have factors like compliance, complicated workflows, rule engines, permission management, etc… and UX takes a backseat to them.


  2. Companies often lack processes to re-evaluate their software purchasing decisions. I’ve seen some spend millions of dollars a year on software that hardly anybody ever used. But it was nobody’s job to reconsider the decision, and so they kept paying for it. Would you work hard on improving your product if you had a solid bedrock of such revenue?

Some other examples of customer lock-in include:

  • IBM, which still made over $1B per year with Lotus Notes in 2013. I heard the IRS’ codebase also largely runs on IBM’s PowerPC mainframes. Selling to the government can often be the guarantee of a lifelong rent for a tech company.

  • Gmail. You’ve already given your email address to all your friends and used it to sign up to hundreds of services. Gmail’s UI isn’t bad enough to justify the hassle of switching all this.

  • Photoshop, or complicated tools in general. In this case, the customer lock-in rests in the sunk cost of mastering a complex tool and learning its idiosyncrasies.

5. Brand

One of the most powerful and durable kinds of moat, as well as one of the most complicated and lengthy to build.

Brands are so powerful that they seem to be Warren Buffet’s favorite kind of moat. He’s placed famous bets on household-name brands (Heinz, Coca Cola, See’s Candies…) that build upon fundamental human preferences, and have been around for long enough to capture distribution channels (a closely related kind of moat I’ll talk about later). This keeps them growing by default, as long as they don’t screw up (Munger’s famous via negativa definition of success: not failing for long enough).

Some of the most powerful brands in the world sell commodities, like L’Oreal, Louis Vuitton, Diesel, Levi’s, Rolex, etc. This is no coincidence, and highlights one of my favorite mental models: to develop a skill, look for lifeforms that had no choice but to excel at it. The Wright brothers, designing their 1st airplane, famously spent a lot of time observing and studying birds.

(Another example is 3M. They reportedly developed compression bandages to fight ulcers, inspired by the skin of giraffes’ legs, which never develop any despite the extreme pressure on the veins of their 18-foot bodies. Likewise, if you want to learn how to differentiate and build a world-class brand, seek inspiration from the companies making a killing selling ultimate commodities.)

There is a positive feedback loop at play with brands and distribution channels. The more people know your brand, the more they expect to see it on the shelves of their favorite store, giving you more leverage over it. That allows you to get better deals with these stores — which, in turn, act as a channel for people to discover your brand, furthering its reputation, etc etc…


A lot of ink is flowing these days about the impact of the Internet on brands.

First, brands have that faculty described above to grant you leverage over distribution channels. This effect is practically nullified by the Internet’s infinite shelf-space.

Second, one of the fundamental functions of a brand is to signal quality, by getting the company to play an iterated game instead of a single move game. In game theory, this can be an effective way to get out of Pareto suboptimal equilibria. For example, turning the famous Prisoner’s Dilemma into an iterated game changes the optimal, rational strategy from (Defect, Defect) to (Cooperate, Cooperate).

Single move interactions can lead to market failures, as in the example of the Market for Lemons. You should read the Wikipedia entry if you want the whole spiel, but the gist is that there is going to be a race to the bottom in quality in a market with these 2 characteristics:

  1. Varying degrees of product quality
  2. Information asymmetry: the seller knows the quality of their product, but the buyer cannot inspect it.

Brands break that dynamic and make it a rational move to sell high quality items, since skimping on quality would hurt sales over the long run by damaging the seller’s reputation. By investing a lot on advertising and exposing itself to you again and again, a seller wants to signal: “we care about our reputation. We’re looking to build a long term relationship, and we’re never gonna give you up, never gonna let you down, never gonna run around, and desert you.”

The Internet demolishes this effect, by connecting a product’s prospective customers with its current ones. Who needs a brand when you’ve got ratings? I don’t need you to spend millions of dollars on ad campaigns. I can just ask all your current customers over there exactly how satisfied they are with your product, and how it’s holding up over time. This is a much cheaper and more direct way to break the information asymmetry.

This does not mean brands will die. The recent emergence of successful brands selling commodities online, like Anker with batteries, is a sign that there is still room for them. But it means that they might be less powerful than they once were. I suspect that the era of monster brands might have been a short parenthesis in the history of markets, closed by Amazon and other aggregators.

Another reason why brands might always be a thing is that the Internet isn’t touching one of their other fundamental functions: signaling. People want to show that they have money, or that they care a lot about the environment. And there will always be companies like Tesla to cater to these needs. The only way I could see the Internet change this is the extreme scenario of complete cultural fragmentation, where people share so few references that signaling becomes impossible. But I don’t see this happening any time soon. We need cultural Schelling points, and I see no reason why we’d be unable to develop them, short of us becoming a galactic civilization.

6. Cornered Resource

A company corners a resource when it somehow gains preferential access to it. Resources can be material as well as human: some firms have gotten so good at acquiring and retaining extremely educated talent from microscopic pools that one could argue they’ve effectively cornered that market — like Google with AI PhDs.

One famous example of a cornered resource is DeBeers and diamonds. Other examples include:

  • Intellectual property, like Disney’s on their characters, or Apple’s on iOS
  • Regulatory capture, an especially vicious form of cornered resource
  • Stronghold on distribution channels

7. Process Power

Leaving the least obvious power for the end: process power. Helmer uses the example of Toyota, which beat GM on its own turf despite incurring sizable tariffs and shipping costs. Worse, Toyota’s leadership was always extraordinarily transparent about its processes, giving interviews and even tours to American automotive executives, telling everybody about the Toyota Production System (TPS), Kanban-based just-in-time manufacturing, “Kaizen” continuous improvement, and workers allowed, nay expected to stop their part of the production line to fix the deficiencies they identified.

They even created a joint venture with GM in Fremont, California, and trained some of their workers in Japan. But they kicked their ass all the same.

Helmer’s point is that process can contribute to product quality, customer satisfaction, sales… And remain completely irreproducible. As such, it can constitute a powerful strategic asset.

A process is not just boxes and arrows on a PDF file, or a fancy list of principles in bullet points. It is deeply embedded inside the organization, permeating its culture. The bullet points are only very high level descriptions of the principles at play. These principles are developed and applied in a myriad of ways which form their own ecosystem within each company, and are not documented in any form. At Toyota, one could say that nobody, including Toyota’s own leadership, understands their process. They just know they’ve set up certain conditions, recruited certain people early on, encouraged some broad categories of behavior while discouraging others, rinsed and repeated for 10 or 20 years, and now they have this ecosystem that produces these results. One could no more reproduce it by visiting their plants and reading HBR than they could learn to play tennis by reading books about it.

The funny thing is that, from an insider’s perspective, you just do thing as usual, not noticing anything special happening. Meanwhile, on the outside, the process looks like some voodoo producing inexplicable results, with outsiders engaging in a cargo-cult replication of your ways and clumsily translating your values in their own lingo. I certainly feel that way with tech. I have no direct experience in any other sector, so I can’t speak for what we do differently than them. But I do see all those old conglomerates spending dozens of millions in consulting, to get presentations with words like “digital convergence” and “innovation economy” all over the place, and end up promoting some guy “Chief Digital Officer” or worse, “Chief Innovation Officer.”


I’d urge any entrepreneur to get familiar with the 7 powers and adopt them as one of their key mental models. You can use it like you would go down a “moats checklist,” and see if a business has the potential to sustain high margins. That works to analyze the structure of whole industries a well.

But before closing, I need to make the mandatory side notes about the interplay between culture and strategy, and the relative importance of strategy and execution.

People have been repeating Peter Drucker’s “culture eats strategy for breakfast” so much that it’s become almost trite. Strategy isn’t something one can design from the top of their ivory tower, spewing out a massive Powerpoint presentation and corporate slogans like Moses coming down from Mount Sinai. Your actual strategy is the one you end up implementing, and it’s concretized by the way you spend your money and attention. Like Gloria Steinem said: “we can tell our values by looking at our checkbook stubs.”

All this starts with culture. When there is a clash between the Tower’s strategy and the Square’s culture, the latter will almost always win. A recent example is Alphabet and Project Maven. The company’s top leadership decided they would start working with the Pentagon on a contract worth $250M/year and possibly unlocking a much larger, $10B deal to build the US army’s entire cloud infrastructure. But Google employees disagreed, and ended up having the upper hand. Google dropped the contract.

It’s not like culture and strategy were these two separate things, one coming from the bottom, the other from the top. Instead, they are deeply interwoven. Strategy is a million small decisions made every day at all levels of the organization. Companies document few of those, and can’t even measure them. In this way, one can look at culture as a sort of meta-strategy: shaping your culture is the strategy that makes the strategy.

Finally, about the importance of strategy and execution. First, the line between those is not always that clear, as explained just above. But the distinction does exist. There is this dogma in Silicon Valley that “ideas are a dime a dozen, what actually matters is execution.” Sure, strategy is worth exactly zero without some muscle to back it up. But I dissent with the majority, and believe more and more that, on the margin, strategy is far more important than execution. Cemeteries are full of startups with poor strategy and brilliant execution — and the Fortune 500 full of giants building exactly the right products, though in a very crappy way.

IBM executed on the IBM PC with the discipline of navy seals, reaching market at scale with an amazingly executed product in just 12 months. But they made the mistake of leaving the OS to a tiny startup called Micro-Soft, which made a killing from the blunder, and made the other good decision not to give exclusivity to IBM. On the contrary, they built their OS such that it could run on equivalent PCs. As a result, IBM got stuck with its beautiful commodity in a low-margin business that it recently had to exit. And Microsoft built its empire on Windows, which was inferior to the competition in many regards, but benefited from strong network effects (software running on top of it, today we call them apps), high switching costs, and economies of scale.

I find that the principle also applies on the level of one’s individual career. If you want to be successful, working on the right thing is what matters most. I’ve seen brilliant engineers who cared about their craft like Japanese sushi chefs — documenting every aspect of their work, testing their feature thoroughly on several platforms, instrumenting and optimizing their code to make sure it runs smoothly and has no memory leak, etc… who got stuck for years, while other engineers with poorer technical skills worked on projects with a direct impact on the business, and earnt back-to-back promotions. Some cynics perceive this as dysfunctional politics — I think it’s a healthy replication of the way markets actually work. Deciding what to work on is the most important part of your job, whether you’re an entrepreneur, a manager, or an engineer.

Nobody Cares

A friend was commenting on how active I was on Twitter, telling me how he held back from posting anything online, out of shyness. I answered that that was my default mode as well, and, I suspect, that of most people. I get shy and afraid of what people could think about me, or I self-censor after imagining the worst way a tweet could be interpreted.

(It sometimes does seem like some people make a deliberate attempt to misrepresent what you’re saying. That, or they’ve spent so much of their lives imagining monsters that they now see them everywhere. Speaking to them, you feel like they’re fighting a mental chimera, instead of addressing your points.)

When I find myself engaging in too much self-censorship, I remember one of the most liberating facts I know: that nobody cares.

An extreme example I can think of to illustrate this point is politicians, who seem able to get away with anything. Embezzlement, sex scandals, you name it. They make national headlines, are the shame of the country for a couple of days, and everybody is sure their career is destroyed forever. Then, they disappear, come back onto the scene after a year or two, run for senator, and win (I hear it can even work when you run for President[reference needed]).

I don’t mean you should be an immoral scoundrel — rather, I want to demonstrate how people really don’t care. These politicians’ entire careers rests upon their reputation, and that of their enemies upon making sure that they don’t get back on their feet. If even they can recover from those scandals, what do you think is the worst that could happen to you (whom, again, nobody cares about) after you’ve made some stupid statements?

Nothing at all. I don’t mean that people won’t remember — I mean that they won’t even notice. Most content is already bad, so yours will just be drowned in that ocean of mediocrity that people scroll through all day. Even better: since it’s bad, it’s not going to spread very far. You’ll be benefitting from a selection effect, where your best content will receive a lot of exposure, and your bad content simply go unnoticed.

This is an asymmetrical game where bad moves cost almost nothing, and good moves are worth a lot. Logic dictates that, playing such a game, you should roll the dice as long as they let you. The board game Monopoly was designed to give people an intuitive understanding of the way markets could supposedly tend towards toxic monopolies — I wish somebody designed a board game to make people understand that life is positive-sum; that the downside of most moves is never really as steep as it seems; that the upside can be unbounded; and that the best thing one can do is make as many moves as possible. In Marc Andreessen’s words, “optimize for the maximum number of swings of the bat.”


Adding a bit of nuance: It’s not that people will never notice your screw-ups — they’ll just do so at such a low frequency that it can safely be rounded to 0 when you’re getting started. And, even when they do, their noticing is largely inconsequential. But, sure, you may get slapped on the wrist from time to time if you post often enough, or to a large enough audience.

I’ve been the subject of Twitter mobs at least 3 times — but that’s after 10 years and almost 14,000 tweets (I also happen to be a child-eating monster, which some people object to). That’s less than once every 3 years, and nothing material happened to me or my career; if anything, it’s allowed me to grow a thicker skin. Even without this silver lining, the upside of “thinking out loud” is more than worth the bit of controversy you’ll have to endure from time to time.

By posting more of your thoughts publicly, you get to steelman your theories, meet people who think alike, build more shared context with those you already know, have interesting conversations, and receive great recommendations (you’d be surprised how much better humans still are at making those than machines).

I do wish more people were active online. The whole promise of the Internet was that of an infinite, vibrant, open forum of ideas. Instead, there seems to be a mass exodus towards private communities — they surely are where I get the most meaningful, genuine interactions, and hear the most original ideas today. This makes me wonder how many eye-opening insights we’re all missing out on, uttered in conversations that should have happened in public. Some people welcome this as a natural evolution of the Internet — I perceive it as the tragic entering into a new Dark Age, and find the silence of some of the most brilliant minds out there deafening.

So, the low frequency of backlashes mentioned earlier does amount to a lot, when you get as much visibility as these folks do, or start representing something to the world instead of just being taken as an individual. When you’ll have written as much as they have, inspired as many people, and taken as much crap, you’ll have an excuse to get shy. Until then, remember the good news: nobody cares about you.

A Long Week-End in Mexico

I took a trip to Mexico City over the long weekend — it was my first time there, and I absolutely loved it. The city surprised me in many ways, and didn’t match at all the picture in my mind. I was expecting dry landscapes scattered with cacti, and instead I saw what was undoubtedly the most green and lush city I’ve ever visited. Almost every street is bordered with a continuous line of tall, luxuriant trees, and there are potted plants  in every shop, café and restaurant.

The general mood in the streets is calm, peaceful, and at the same time vibrant, joyful, and insouciant. People don’t hurry around or push each other like they do in New York City. They smile easily and don’t get impatient if you take too long to decide what you’ll order, or if you don’t speak their language. There is a general feeling of dolce vita: people stroll, kids play, and there is often music playing from storefronts and balconies… The street is a place of life, instead of a mere corridor used to get from A to B, like in the US. To be fair, the amazing climate certainly helps.

(As an aside, there is something to be said about this insouciance, inside a country reputed to be dangerous. In America this is contrasted with downright paranoia and over-caution. A friend recently told me the story of an acquaintance of his, who got arrested by the police for leaving his 9-years old child unattended for a couple of minutes in a public park in SF. Compare this to the many groups of kids who were safely playing football in squares in Mexico City, with minimal adult supervision. Coming to America, I was expecting to meet reckless cowboys chewing tobacco. Instead, I saw skinny-jeans wearers who will never jaywalk, leave their kids unattended, or go to work if it’s raining outside — though that last one might be specific to California.)

To my embarrassment, I was also expecting a dangerous city. I wasn’t exactly thinking I’d get mugged in just 3 days, but we can all be nervous when our only exposure to a new place has been through network news outlets. Even in the most cosmopolitan cities one can always feel when they’ve wandered into the wrong neighborhoods. I felt none of that in Mexico City, even walking in small dark alleys at night — I certainly felt a lot safer there than I do in San Francisco. I also saw much fewer homeless people, and the few I encountered did not seem as terribly mentally ill, or drug addicted to as an extreme degree. 

Speaking of not speaking their language, this was one thing that disappointed me: almost everybody there whom I encountered speak absolutely no English — and by that I mean, are unable to count up to five.  I don’t mean to sound elitist, rather I’m genuinely curious as to why this is. You would assume a relatively poor country (their GDP per capita is $9k, one sixth that of the US), sharing a border with two of the most wealthy states of the wealthiest country on earth, would deem it useful to learn their language — especially when it’s so grammatically simple, and shares the same alphabet as them. Incidentally, this English illiteracy also exists in Spain, the only European country I’ve visited where people did not commonly speak English. I speak almost no Spanish myself (although a huge percentage of words are cognates in French), which made it quite hard to communicate with them. This hindered my trip, cutting me off from any meaningful exchange with locals — next time, I’ll try going with a Spanish-speaking friend. I enjoyed the city so much that I’m considering learning some Spanish, just in case I decide to extend my stay there someday.

One thing that will certainly bring me back to Mexico is the food. There really aren’t that many Mexican restaurants where I grew up in France, and I’ve just learnt that what San Franciscans call Mexican food is, indeed, very different from the real thing. First, the food in Mexico City is generally a lot less spicy than one would expect, and a lot less spicy than its Californian counterpart. Second, the ingredients are fresh, and of very good quality. The guacamole especially tastes completely different, probably as a result of using avocados of better provenance. 

A lot of these meals are consumed in Izakaya-like shops in the streets. Those abound, contributing to what seems to be a huge informal economy. One can buy just about anything from corner vendors, most of which don’t accept credit cards — I’d be surprised if much taxes were paid on these transactions. I couldn’t help but think of the inefficiency represented by all those vendors (about 2 per block!), standing all day next to their carts, only there to put cash in a leather pouch and hand a product to a customer who could easily have taken it themselves. Their omnipresence reminded me of Tokyo’s vending machines — I guess the reason they haven’t been automated yet is because their labor is just so cheap. In any case, they add a lot of charm and life to the streets.  

Despite the relative pleasingness of the city, some things are still there to remind you that you’re in a relatively poor country. Although the streets were extremely clean (much cleaner than SF’s), some neighborhoods right in the center of the city have a pungent stench of sewer throughout the day. Everybody warns you not to drink the tap water, huge construction machines are active right next to pedestrians, electric and phone cables installed in a hodgepodge manner, and, the unforgivable sin, the Internet connection was dreadful.

Despite any expectation I may have had, I had a blast, and I can’t wait to explore more of that beautiful country. I can’t believe it’s taken me this long to get there in the first place, especially when it’s a mere 5h flight away from SF. Next time, maybe Puerto Vallarta, or Cancún.

(all pictures by @ludoviclandry)

Favorite Books of 2017

Biographies

Einstein: His Life and Universe

Definitely the best biography I’ve read this year. It combines details into what’s probably the most incredible intellectual adventure of the 20th century — we’re talking of a guy who single-handedly upended physics as it was understood, alone from his patent office in Switzerland — and insights into the personality of the man behind it. Shed light on aspects of Einstein I knew nothing about, and taught me what an awesome human being he was, just as as much as he was a genius physicist.

Benjamin Franklin: An American Life

Another one from Walter Isaacson. The advantage of reading the biographies of people like Benjamin Franklin is that they make you learn about both the life of a man, and the historical period they had a part in shaping — in this case, the American Revolution, which I know way too little about — in a way that I find much more relatable than history books.

How Breakthroughs Happen

Makes the case that innovation is more about the combination of existing ideas in novel ways, than the fundamental breakthroughs people imagine, and talks of “technology brokering:” gathering ideas, practices, and technologies from seemingly foreign fields, so as to bring them together, and sometimes create a whole new one.

Technology brokers are characterized by their connections to multiple small worlds and, further, by their connections to worlds that otherwise share few connections between them. Network theorists use the concept of range to describe the breadth of connections any one person may have in a network, as measured by the number of non-redundant ties […]. Technology brokers attempt to maximize their range of connections because by doing so they are in a better position to be the first to see how the people, ideas, and objects of one world may provide valuable solutions in another.

This made me think of this article, The No. 1 Predictor Of Career Success According To Network Science: 

According to multiple, peer-reviewed studies, simply being in an open network instead of a closed one is the best predictor of career success.
[…]
In the chart, the further to the right you go toward a closed network, the more you repeatedly hear the same ideas, which reaffirm what you already believe. The further left you go toward an open network, the more you’re exposed to new ideas. People to the left are significantly more successful than those to the right.

I believe this highlights a weakness of the Silicon Valley ecosystem, which is becoming increasingly monocultural. Much of this is due to the insane, NIMBY-induced rents (a 1 bedroom in downtown San Francisco now costs easily $3,500 per month), making it impossible for all but the wealthiest to stick around — those wealthiest tending to be the ones who work in tech. The silver lining is that you not only increase the concentration of tech people, and with it the frequency of valuable serendipitous encounters, you also make the most driven people self-select to migrate to this small area, ready to sacrifice personal comfort, for the hope of participating in building something impactful.

But this also comes at the high cost of building this echo chamber everyone complains about. I don’t have any data to support this claim, and the human memory is notoriously bad to detect that kind of pattern, but I do feel there’s been a noticeable decrease in the originality / craziness of people around, in just the 5 years I’ve been here. You still see wicked smart, brilliant, ambitious, hard-working people, but you’ll find those in NYC too, whereas the Bay Area used to have its own, special breed of weird people, in a good way.

That article above also contains an amazing quote, related to technology brokering, from a Wired interview Steve Jobs gave:

Creativity is just connecting things. When you ask creative people how they did something, they feel a little guilty because they didn’t really do it, they just saw something.

It seemed obvious to them after a while. That’s because they were able to connect experiences they’ve had and synthesize new things. And the reason they were able to do that was that they’ve had more experiences or they have thought more about their experiences than other people.

Unfortunately, that’s too rare a commodity. A lot of people in our industry haven’t had very diverse experiences.

So they don’t have enough dots to connect, and they end up with very linear solutions without a broad perspective on the problem. The broader one’s understanding of the human experience, the better design we will have.

Exit, Voice, and Loyalty

One of those classics I just had to read, after having it recommended to me one too many times. You can check out my full review here.

History

The Lessons of History

The best way to describe this book is through its introduction:

Since man is a moment in astronomic time, a transient guest of the earth, a spore of his species, a scion of his race, a composite of body, character, and mind, a member of a family and a community, a believer or doubter of a faith, a unit in an economy, perhaps a citizen in a state or a soldier in an army, we may ask under the corresponding heads—astronomy, geology, geography, biology, ethnology, psychology, morality, religion, economics, politics, and war—what history has to say about the nature, conduct, and prospects of man. It is a precarious enterprise, and only a fool would try to compress a hundred centuries into a hundred pages of hazardous conclusions. We proceed.

From Third World to First

The memoirs of Lee Kuan Yew, the hero (and I don’t use the word lightly) who served as Singapore’s Prime Minister for over 3 decades, bringing about the miracle he describes in the title of his book: transitioning “from third world, to first world, in a single generation.” Lee thought of the issues of his time in a crystal clear, all-encompassing fashion that’s nothing short of fascinating, and his book is replete with interesting anecdotes, giving a rare peek on the small quirks of the world leaders he frequented closely over several decades.

Of course, it’s also quite a lesson in leadership, from someone who led what must be one of the most amazing success stories of the 20th century, and Lee’s philosophy of extreme pragmatism is a challenge to the most ideologically minded out here — I know it did challenge me.

Self-Help

Principles

This is probably the one best book I’ve read in 2017, and will be one of those books I revisit every couple of years.

Ray Dalio is the founder of the famous hedge investment firm Bridgewater, and this is a concentrate, in a mere 300 pages, of the principles he’s built for himself over a lifetime.

Principles has, by a very wide margin, the highest signal to noise ratio I’ve ever seen in a book. Dalio has an extremely analytical, methodical mind, and it shows: the book is structured in bullets and sub-bullets, written in a clear and concise fashion, no word being unnecessary nor prose overly flowery. I’d share some excerpts here, but would end up quoting half the thing.

Peak Performance

A recommendation by @sknthla. It’s one of those easy-to-skim books that try to pound a couple of simple, valuable insights into your skull:

  • Progress happens by alternating periods of intense stress with deep rest, the latter being just as important as work — and world class performers taking it just as seriously.
  • That “stress – rest” cycle happens at all levels, and, at the lowest one, the authors recommend alternating short periods of work (around 1 hours) with 5-15min breaks.
  • All ways of resting are not made equal, the best ones being:
    • Meditating (doesn’t have to be long, 15min).
    • Social activities, i.e. hanging out with friends.
    • Walks in nature. Apparently, and I’d be curious to see if those results replicate consistently, merely staring at pictures of nature have a positive impact on the brain’s learning ability.
  • The work that counts is the one that’s painful. It’s going to sound cheesy, but “no pain, no gain” is definitely true, or, like Mohamed Ali said “I only start counting when it starts hurting.”
  • An excellent way to make progress is to deliberately seek out challenges that exceed your ability, just enough to be challenging, and not so much as to be overwhelmingly stressful. Make it harder if you’re completely in control, and dial down a notch if you’re overly anxious.
  • Perceiving stress as a challenge instead of a threat will make a huge, direct impact on the progress you make. Growth mindset yada yada.
  • Routines and habits are important. To get in the zone faster, develop routines for before and after your work sessions.
  • Never multitask: put your phone away (just having it in sight, or feeling it in your pocket, apparently has a measurably negative impact on your focus), and if necessary block websites using utilities such as SelfControl.

Heroes on the Shoulders of Giants

Epistemic status: I’m outside of my circle of competence when it comes to AI. That said, most of this post is just logical reasoning, independent of AI-specific knowledge.

François Chollet (whom you should definitely follow on Twitter) writes about “The Impossibility of Intelligence Explosion.” Here’s his TL;DR, although I’d encourage you to go read the whole thing:

  • Intelligence is situational — there is no such thing as general intelligence. Your brain is one piece in a broader system which includes your body, your environment, other humans, and culture as a whole.
  • […] Currently, our environment, not our brain, is acting as the bottleneck to our intelligence.
  • Human intelligence is largely externalized, contained not in our brain but in our civilization. We are our tools — our brains are modules in a cognitive system much larger than ourselves. A system that is already self-improving, and has been for a long time.
  • Recursively self-improving systems, because of contingent bottlenecks, diminishing returns, and counter-reactions […], cannot achieve exponential progress in practice. Empirically, they tend to display linear or sigmoidal improvement. […]
  • Recursive intelligence expansion is already happening — at the level of our civilization. It will keep happening in the age of AI, and it progresses at a roughly linear pace.

I think the piece makes a couple of brilliant points — I love the idea of “civilization as an exocortex” — but I disagree with some of those points, or think they still allow for the possibility of a future intelligence explosion. Here’s my take on this.

1. We’ve already gone exponential

François Chollet writes:

[Y]ou may ask, isn’t civilization itself the runaway self-improving brain? Is our civilizational intelligence exploding? No. Crucially, the civilization-level intelligence-improving loop has only resulted in measurably linear progress in our problem-solving abilities over time. Not an explosion.

I disagree, and think that yes, we’ve already been going exponential, by a lot of metrics.

This is the historical population of the world (although some say that this should soon start looking like an S-curve, the world having already reached peak child):

Now, here’s the World GDP (or GWP) per capita across history, in 1990 dollars (from Wikipedia’s “World Economy” article):

Remember, this is per capita, so the curve represents growth in productivity itself, rather than in population. And this is over a very short time span! 500 years isn’t much compared to the ~10,000 years that human civilization’s been around for.

And this is limited to humans, who, in the grand scheme of the Universe, are an extremely recent phenomenon. You may already have heard of Carl Sagan’s “Cosmic Calendar,” compressing the Universe’s chronology from 13.8 billion down to a single year. On this timeline, life doesn’t appear until September 21st, multicellular organisms December 5th, and anatomically modern humans before December 31st, 11:52PM. Agriculture, which many historians consider the beginning of human civilization, doesn’t start until 11:59PM, and 32 seconds.

In his article on “quantitative macrohistory,” Luke Muehlhauser writes:

[M]y one-sentence summary of recorded human history is this: Everything was awful for a very long time, and then the industrial revolution happened.

This industrial revolution, on the “cosmic calendar”, doesn’t start until 570 milliseconds before midnight.

So, sure, if you look at it on the timescale of a single human, things are moving kind of slowly. But, remember: it’s a miracle we’re seeing anything move at all! For most of human history, one could have time-traveled 100 years into the future and been able to function normally, barely noticing any difference (except for the way those punks are now tying their hair with 2 bones instead of 3).

To wrap this point up: on the grand timescale of the Universe, nothing happened for a very long time, and then all of a sudden, life developed. From this point already, staying on the scale of the Universe makes it hard to tell events apart, so we need to zoom into this thin sliver. There, we can see, again, not much happen, and then all of a sudden, mammals develop, including apes. Zooming further, not much happening, and then human civilization! Zooming in, not much, and then industrial revolution!

You get the idea: not only do I think the singularity can happen, I think we’re already living in it.

If there ever is a computational singularity, where nanoseconds will be to those machines what millennia are to us, they’ll summarize not only human history, but even our modern times, as “nothing happened, and then there was the seed AI”. Our “can you imagine Earth was around for 600 million years before any life developed?” will be their “can you imagine they used to sleep for 8 hours a day?”, 8 hours being as unfathomable a duration to them, as 600 million years are to us.

(I realize I’m grossly anthropomorphizing what would actually be completely alien entities, but it’s fun, and you get my point.)

2. Individuals are the ones pushing things forward

Chollet writes: (emphasis mine)

An individual human is pretty much useless on its own — again, humans are just bipedal apes. It’s a collective accumulation of knowledge and external systems over thousands of years — what we call “civilization” — that has elevated us above our animal nature. When a scientist makes a breakthrough, the thought processes they are running in their brain are just a small part of the equation — the researcher offloads large extents of the problem-solving process to computers, to other researchers, to paper notes, to mathematical notation, etc. And they are only able to succeed because they are standing on the shoulder of giants — their own work is but one last subroutine in a problem-solving process that spans decades and thousands of individuals. Their own individual cognitive work may not be much more significant to the whole process than the work of a single transistor on a chip.

Those “standing on the shoulders of giants” points never sit quite right with me, especially when it’s used to de-emphasize the importance of individuals. It’s true, to some extent: we all leverage the progress civilization has made before us, so that we can move it forward ourselves.

But to the extent we’re all standing on those giants’ shoulders, why do so few of us achieve anything noteworthy? Is it only about luck and circumstances, aka “being in the right place, at the right time”?

I think circumstances matter, but they’re not the main factor. Again, at the time when scientists made some of the greatest discoveries, many people were arguably in the right circumstances, indeed, often much better than the ones of the researcher who ended up making the breakthrough. Einstein is the canonical example: how come this clerk, stuck in some patent office in Switzerland, was the one who discovered special relativity, when so many others had tenure, with more time to think about that stuff than him, access to more content, more brains to pick from and have fascinating conversations with, and sometimes with entire teams and laboratories at their disposal?

Newton, too, had his famous “annus mirabilis”, when he discovered the laws of gravity, and differential and integral calculus, while trapped at his parents’ place in the countryside, to avoid a plague epidemic.

If anything, it sometimes seems like the only way to have any impact is to know when to stop shoulders-climbing, and start working on your own thing, like the researchers in those two examples did. After all, those shoulders know no top: you could dedicate your entire life to a single field, without having exhausted the discoveries that have been made in it. In the words of Richard Hamming:

There was a fellow at Bell Labs, a very, very, smart guy. He was always in the library; he read everything. If you wanted references, you went to him and he gave you all kinds of references. But in the middle of forming these theories, I formed a proposition: there would be no effect named after him in the long run. He is now retired from Bell Labs and is an Adjunct Professor. He was very valuable; I’m not questioning that. He wrote some very good Physical Review articles; but there’s no effect named after him because he read too much.

I agree the work civilization has done before you matters, but the importance of individuals still seems under-appreciated by most people. It takes only one person, sailing further than others did, to push the frontier forward for everybody else. In this sense, civilization actually moves as fast as its fastest member. We only need one Columbus to unlock access to a whole new continent, one Einstein to get the insights required to develop lasers and nuclear power, one Darwin to gain a new understanding of life.

(As an aside, this “bias towards action”, required to have any impact, appears to me as one of the greatest strengths of Americans. It is so easy to lose oneself in a pointless hoarding of knowledge, and be frozen into inaction by “what ifs” and “things we could have missed”. Americans, with their “shoot first, ask questions after” ethos, seem to appreciate that one should start doing something before they completely understand it.)

3. Brains are the bottleneck

François Chollet continues:

[T]he current bottleneck to problem-solving, to expressed intelligence, is not latent cognitive ability itself. The bottleneck is our circumstances. Our environment, which determines how our intelligence manifests itself, puts a hard limit on what we can do with our brains — on how intelligent we can grow up to be, on how effectively we can leverage the intelligence that we develop, on what problems we can solve. All evidence points to the fact that our current environment, much like past environments over the previous 200,000 years of human history and prehistory, does not allow high-intelligence individuals to fully develop and utilize their cognitive potential.

I agree with this, except when it comes to the frontier. If you broadly define intelligence as the ability for an entity to control its environment then, by and large, one’s intelligence sits outside their brain: 99% of one’s ability to control their environment is dependent on the ticket they pulled in the lottery of “in what century will you be born”, with less than 1% determined by their IQ.

But that civilization was, in the first place, built and shaped by individuals, just as much as they were in turn shaped by it. When it comes to pushing the frontier, and redefine what the next centuries will mean for the generations living them, the bottleneck lies in these individuals. Here, I echo Peter Thiel’s rejection of technological determinism, according to which “progress will happen, no matter what.” For better or for worse, I think that, regardless of the current cultural momentum, the fate of the world still lies in the hands of individuals, and the actions and decisions they’ll make.

From a more concrete and “micro” standpoint, ask entrepreneurs what’s the current bottleneck holding their companies’ growth back. Some will say it’s market adoption, or capital, proving Chollet right in his emphasis on contextual factors. But in AI, almost all will say it’s the brains: there just isn’t enough talent to go around, and articles abound about the piles of money they’ll throw at the good ones when they find some, with commonplace 7-figures packages.

Like Charlie Munger says:

I like artificial intelligence, because we’re so short of the real thing.

4. AI scales in a way humans can’t

Still from Chollet’s post:

An overwhelming amount of evidence points to this simple fact: a single human brain, on its own, is not capable of designing a greater intelligence than itself. This is a purely empirical statement: out of billions of human brains that have come and gone, none has done so. Clearly, the intelligence of a single human, over a single lifetime, cannot design intelligence, or else, over billions of trials, it would have already occurred.

But of course, everything had never happened, until it happened for the first time. You could equally have said “it’s an empirical observation that no human can run 100 meters in less than 10 seconds, since nobody’s ever done it before”, until Usain Bolt showed up.

Chollet continues:

Will the superhuman AIs of the future, developed collectively over centuries, have the capability to develop AI greater than themselves? No, no more than any of us can.

But once we’ve got that Seed AI, nobody’s talking of making only one of it. The thing with artificial intelligence is that, unlike its natural counterpart, it can actually scale, horizontally as well as vertically. When AlphaGo won its matches against world-champion Lee Sedol, many were quick to point out that:

The distributed AlphaGo system uses about 1 megawatt, compared to only 20 watts used by the human brain.

Which, according to those observers, makes the comparison unfair. However, like Benedict Evans says, “unfair comparisons are the most important kind.” The game doesn’t care how energy-efficient you are: you either win, or you lose. If anything, this seems to earn AlphaGo more points: in AI, once we find something that works, we can just throw a bunch of chips and megawatts at it, and watch it become even better. It’s often pointed out that the technologies behind today’s AI renaissance, like gradient descent, convolutional nets or deep learning, had actually been around for a long time, and just started becoming effective once we had enough computing power to feed them.

On the other hand, how do you think 50,000 humans, consuming a total of 1 megawatts, would have performed against a single AlphaGo? The closest thing I can think of is the Twitch Plays Pokemon experiment, where 1.2 million players took 394 hours to complete a game that a single kid can complete in about 25 hours.

(For the record: I don’t think all we need for the singularity to happen is more computing power. We probably need a lot more theoretical breakthroughs to get there. All I’m saying is: once we’ll have made those breakthroughs, and built this “seed AI”, we’ll be able to flip a switch, and get a million more of it, something you can’t do when you find a really smart human.)


To summarize:

  • Looking at history over a large enough time scale, it’s clear the complexity of the Universe is already exploding exponentially—from subatomic particles, to atoms, to molecules, to life, to organisms, to civilization. An intelligence explosion would sit in the continuity of this phenomenon.
  • Individuals are the ones pushing civilization forward, although they’re also leveraging the progress it’s made before them. Civilization tends to move as fast as its fastest member, since it only takes one of these individuals to push the frontier forward for everybody else.
  • Hence, the quantity, and especially quality, of available brains are the bottlenecks standing in the way of progress…
  • … and the first AI to surpass human intelligence would lift those bottlenecks, unlocking an intelligence explosion.

Thanks to Dan Wang and Kevin Simler for reviewing the draft and providing suggestions.

Book Review: Exit, Voice and Loyalty

Exit is an important part of capitalism: by allowing consumers to take their money anywhere they please, it trims out inefficient firms, drives productivity and quality up, and prices down.

“One thing people underestimate is how markets don’t allow anyone to do anything except make better and better products.”
– Bill Gates

“If we don’t produce better shoes, faster, and at a lower price than Bob just next door, consumers will flock to him and we’ll be bankrupt before we know it!”

The mechanism is so crisp, so simple, and yet so powerful, that it’s one of the first ones most people, including billionaires and economists, resort to when thinking about markets.

But Hirschman makes the case that exit doesn’t have the monopoly over quality-improving, price-reducing forces, and proposes an alternative: voice. The reasoning goes as follows: even if I have a monopoly, if I keep producing goods or services of low quality, or at too-high a price, my customers will keep complaining about it, which can have unpleasant outcomes, such as attracting the attention of the regulator. Voice is a lot more messy than exit, but it works – in fact, it works because of its messiness.

The rest of the book is about the interplay between those two forces, voice and exit. In most cases, exit gives strength to voice: nobody cares about your complains if you have no better alternative. But in other cases, exit can downright be counter-productive, i.e. hurting quality and driving prices up. Those are the cases where firms or institutions are insulated from the signals exit send them, most of the time because they’re public entities, and so just don’t care much about profits.

The problem (and this is one of the most powerful points made by the book): in those instances, exit can provide an escape hatch to the most demanding customers, who then don’t make their voices heard by the firm, where voice is precisely the only thing that could get it to change. Those entities, for this reason, could even want the exit of those most demanding customers, who, if they stayed, would be nothing but pains in the ass. In the case of private firms, the existence of alternatives not only helps their case in anti-trust lawsuits (“we’re not a monopoly! look at those little guys in the corner who kinda compete with us if you think about our market in this arbitrary way”), it also drives the most demanding – and so potentially most vocal – customers away, leaving them with a bunch of docile customers who care a lot less.

But sometimes, the customers who remain don’t do so just because they’re docile, or non-demanding, but because they just can’t afford the higher quality, more expensive alternative. In this case, voice becomes the last resort of the powerless, and it can, when they organize themselves on a sufficient scale, actually be quite powerful.


Hirschman gives the example of public and private schools: if a public, free school decreases in quality, the most quality-conscious (or least price-conscious) parents will be the first ones to exit. This would result in a lower income for public schools, who would give a damn if they were, well, private. But since they’re funded by the state, mostly regardless of their financial performance, they don’t care about running at a loss – in fact, maybe they prefer it this way, so they can present themselves as the “steward of the poor, last bastion of the powerless” etc etc.

But what they do care about, since this deep-pocketed funder of theirs is subject to democratic forces, is the discontent of its customers. If a bunch of angry parents were to go in the streets tomorrow, chanting slogans about how public schools were but of bunch of stupid, corrupt incompetents, that would attract attention, and they might lose some funding or, more likely, be put under more stringent supervision. Hence, exit, by first trimming down the people who care the most, at the expense of their voice, actually contributes to the public school’s worsening quality.

This also means that there is something qualitatively different about the higher-quality, higher-price option in a market, which will gather the voices of the most demanding customers, serving as fuel in its quality-improving engine. Perhaps this is the reason why we hear so many people complain about problems with their iPhone, and so few with their Android, despite iOS being so obviously superior to it *ducks for cover*


An interesting application of Hirschman’s theory is how it applies to politics. If the market’s got exit, politics has votes, and, to a large extent, voice competes with votes just as much as it does with exit: if a voter isn’t happy with their party, they can either cry about it and try to “change it from the inside”, or give their vote to somebody else.

And, in the same way that, in the markets, voice will be the last resort of the captive customer, it is too the one of the captive voter. Now, since votes don’t cost anything, and everybody, rich and poor, get exactly one of them, who are those “captive voters”?

Think of political offerings as a one-dimensional spectrum, with everybody equally distributed along it, from left to right, depending on their opinion. Political parties need to get as many votes as possible to get elected, and people just vote for whatever party sits closer to them on the spectrum. In that case, political parties will try moving as close to the center of the spectrum as possible, which will make them capture more of those middle-votes. That also means they’ll drift further from their most extreme constituents, but nobody cares, since they’re still closest to them and so will still get their votes.

And this, according to Hirschman, is the reason why the most extreme voters are the most vocal: not just because they’re pissed off and there is something soothing in having everybody hear about it (although I’m sure there’s some of that too), but because they’re the ones least empowered by the democratic system, and since their vote isn’t going to get them anywhere, they’re left with nothing but their voice.

Now, like firms, political parties are both vote-maximizers, and, to a lesser extent, discontent-minimizers. Sometimes, they’ll overshoot in the latter direction (listening too closely to the extremists), at the expense of their votes-gathering.

(I usually try to read as little about politics as I can, since so much of it is pure signaling. For that reason, I’ll let people who are more knowledgable than me in this topic expand on how this applies to the last US elections, and more specifically the Hillary-supporters-Bernie-bros interplay, which I know next to nothing about.)


Now, loyalty. By delaying exit, loyalty tips the scale towards voice. Loyalists will tend to exit less, and complain more, than non-loyalists. Even if they exist, they will do it for radically different reasons than other people: when non loyalists exit, they then “couldn’t care less” about the fate of the firm they left behind. On the contrary, loyalists often resort to exit as another form of voice, and, in the case of the boycott, even make it explicitly contingent upon some action taken by the firm.

In that case, it could be argued that the person wasn’t loyal to the firm in the first place, but rather to some ideal it perceived the firm to stand for. This to me highlights the power of the narrative, that can become a self-fulfilling prophecy: by taking a public commitment to some mission of social value, a firm will attract those believing the most in it, who, through their voice, will hold the entity accountable to this ideal. This made me think of Muhtar Kent’s “A brand is a promise. A good brand is a promise kept.”


By and large, I don’t think I agree with Hirschman’s main point, i.e. that voice can be a substitute to exit, and that exit can sometimes even hurt quality and prices by muting out voice. Sure, you may be able to find some situations where, on the margin, that is the case, but I find them more interesting as counter-intuitive edge cases than as sizable phenomenons in the world. You can find those edge cases if you look hard, but the vast majority of the time, preventing exit is a net-negative, even accounting for the voice that comes to replace it.

The fact that exit could hold certain firms back says nothing about the counterfactual, which is the improvement in quality that would happen in the firm customers would be leaving for, and that is incentivized to fill up the gap left by the incumbent, to get those dissatisfied exiters’ money. Maybe Western Europe would have slightly improved, had people stayed put instead of leaving to America in the 18th and 19th centuries, but that doesn’t mean it was worth missing out on the opportunities offered by the new continent.

Prosperity, offered to you by exit

Even if voice was more effective, I wonder about its negative externalities. Voice implies flooding the meme space with messages of dissatisfaction, spreading them everywhere, including to people who aren’t the least concerned about the matter. Who would want to live in that negative cacophony? Exit, by comparison, is a lot quieter, cheaper and doesn’t even require people to be literate or articulate enough to voice their concerns. All it requires is for them to put the detergent bottle back onto the shelf, and pick another one instead.

You could also argue voice to be a zero-sum game, to the extent that influencing public opinion is the very way it gains its power, and that there are only so many messages large groups of people can be exposed to in a given day. Exit doesn’t know those limits on its bandwidth, and so scales much better.

Another power of exit, especially when it comes to community, is the ability to strengthen both sides: the deserted and the new group. Two factions stuck together in the same group will be doomed to pull in reverse directions, each canceling the efforts of the other in a tug-of-war that yields nothing but immobility (ohai congress). Decoupling them can allow each to evolve in the direction it would prefer, and so to increase their coverage area, and speed of exploration, of the tree of possibles. Many people describe this as one of the greatest forces of cryptoassets: like Chris Burniske said, Bitcoin’s forks have made it transition from 2 warring parties in a single country, to 2 different countries with a single party each.

(Imagine how great Twitter would be if, instead of being stuck in intestine wars, which I assume is the reason why its product is at a standstill, true believers could get their “Twitter classic” that’s stuck in v0.1, with 140 characters, favorites still being favorites instead of likes, chronological timelines, etc… And the rest of us an actual modern product, with full algorithmical timelines, no character limit, and God knows what else innovative people could come up with if they didn’t have to spend their time arguing.)

Finally, even assuming that voice could bring improvements to existing firms, exit works in a radically different way: instead of trying to improve unsatisfactory, inefficient firms, it kills them, until only satisfactory ones remain. If voice is like a single firm doing dead-lifting, exit is like an entire ecosystem being subject to natural selection. And, no matters how much crossfit a fish does, it’ll just never be as good as lizards when it comes to crawling on rocks – if rocks-crawling is the last trend.

For that reason, I think exit is one of the greatest forces for good, and creating new escape doors is one of the highest leverage activities one can engage in. That’s one of the reasons why I’m so excited about blockchains: they break up monopolies, and create a new permissionless paradigm for innovation. Likewise, VR can break monopolies over space, by creating an infinite amount of it, allowing more experimentation, and models we can’t even think of today.