Outsiders

I’ve Misunderstood Tal Again

Once again, I have not fully understood Tal. Mikhail Tal is not just heroic because he trusts his instincts and Tal is not just great because he is bold. Tal is both of these things but is able to win because he has a unique and deep understanding of chess.

My current theory is that Tal was able to play unlikely moves that seemed like magic, because he understood the game beyond others’ with his own lens. This enabled him to see opportunity in deep, unlikely lines that others missed. While he certainly had a larger risk appetite and was bold in character, he based this boldness in having a unique view of the chess board that the other person didn’t. He was an outsider in chess that allowed him to make creative and powerful moves.

Insiders vs Outsiders

There is a famous Silicon Valley piece by Auren Hoffman comparing Naval (outsider) and Nirav (insider). Auren says that insiders have access and influence but are constrained by established rules, while outsiders lack insider access but possess the freedom to challenge the status quo and bring fresh perspectives. Tal was an outsider.

Auren says that most people fall at some point in between. He created this venn diagram to showcase the difference between the two types. They can both be winners, but they win in different ways and may be more likely to win in different environments.

“Outsiders need change to ascend. They often need a revolution. If change does not happen quick enough, they can be thwarted easily by sharp insiders who are networked together. Successful outsiders (like those in Ayn Rand novels) are deified. But most outsiders are unsuccessful and live their life mad that society has not recognized their talents.

‘But the memoirs consistently paint a picture of a willful kid, a loner, stubborn, maybe a little gloomy, and rough around the edges. But in these accounts, he also has a feisty streak. In one story, Napoleon led his fellow students in throwing their mattresses out the window; anything to irritate the monks who were their teachers.’

Not everyone fits neatly into being a Nirav or a Naval. Not everyone is a pure insider or outsider. We’re individuals. We’re grey.”

Change

This is a pretty helpful framework and why Silicon Valley thrives off of outsiders, who come in and shake things up a bit. They don’t follow the rules. They observe the world as it should be and will not conform.

And this also reminds me of one of my favorite quotes. If you have made it far enough to see the value in these people, you are going to have to learn to deal with them. Rulebreakers don’t make the best employees in most cultures, but they have immense creative potential when harnessed.

“One of the most significant mistakes made by observ­ers of the technology industry’s rise is to assume that the software produced by such companies is the reason for their domination of the modern economy. It is rather a set of cultural biases and practices and norms that make possible the production of such software, and thus are the underlying causes of the industry’s success. The central insight of Silicon Valley was not merely to hire the best and brightest but to treat them as such, to allow them the flexibility and freedom and space to create.

The most effective software companies are artist colonies, filled with temperamental and tal­ented souls. And it is their unwillingness to conform, to submit to power, that is often their most valuable instinct.”

Alex Karp, The Technological Republic

Has Y Combinator Lost the Plot?

Y Combinator has become a very important institution in the United States. The companies that have gone through YC have created a total value of $800 Billion in combined valuation.

Recently, YC has been receiving a lot of criticism for expanding the program and spinning out more and more startups that seem like they’re clout chasing and aim for virality more than they do a real business. It seems to me that what people are asking is whether YC is still a place that attracts authentic outsiders. My entire life I have placed people I have met along an axis of intrinsically to extrinsically motivated. Nearly everyone falls under extrinsically motivated nowadays, and I believe that’s often disastrous.

I took part in an accelerator at the University of Illinois and it was clear then that accelerators don’t help much. The great companies would be great with or without help from the staff. And further, you realize that anyone that does startups for the status doesn’t last long. If anything, it got in the way of deep work. But I also know people who have gone through YC and speak very highly of it. I guess this makes me question how valuable the institution is versus the individual. I’m always championing the value of individualism, but I can see both sides.

I don’t know why I included this section, but it has been something I’ve been thinking about. Part of me thinks that institutions today have decayed enough that maybe we could use a new institution like YC to take their place and that I welcome it. And then another part of me thinks that they have lost the plot. Maybe you only get outsized returns from outsiders. Look at the Thiel Fellows.

One of the questions we need to ask is whether YC is what makes these companies successful or whether it is just because they’ve created a successful index fund/fish net that funds great outsiders. Maybe software is no longer a game for outsiders? Perhaps YC is making the bet that we need more Niravs than Navals.

What is Valuable About Universities

I’m visiting and writing this from the Stanford Library. I’ve been looking for an environment that was as good to work at as Grainger Library from UIUC for years. I’ve finally found it. There’s something about University environments that make us better.

Of course it is helpful to learn new things and take classes you are interested in. I think many would agree with me that the value that universities offered them was exposure to new ideas and people. You learn a lot from being around other great people. By expanding your horizons, you find yourself.

There is something important about learning things you otherwise wouldn’t have. Learning new things and changing your view on the world is important for us in a way that I will come back to later. There is a certain serendipity that occurs in university.

Generalists vs Specialists

One way to think of people is as insiders and outsiders. Another is with breadth and depth, generalists and specialists, well-roundedness and spikiness.

I feel like tech has always portrayed outsiders as spiky and insiders as generalists. I think you can be a generalist outsider and a spiky insider. And so we are confronted with another question: which axis is more important? Insider/Outsider or Breadth/Depth.

Silicon Valley is obsessed with the archetype of the autistic kid who builds the next billion dollar business. But I don’t necessarily think there is value in being socially deficient itself. They are associating them with a certain type of personality. Someone who is completely deficient in most areas but as a result they are extremely spiked in one that makes them the best in the world. And I don’t think it is wrong that this spikiness exists in people and that it can be extremely valuable.

I’d like to introduce a new graph that captures another dimension. Because generalists are just as valuable, and I think more valuable over time if your business is to grow and bring a wide variety of problems.

Shallow & Wide vs Deep & Narrow neural network.

If we look at who really creates new unlikely change. It is usually the outsiders. But within this category, they don’t necessarily always need to be spiky. You can be a generalist as well. The other axis we should also pay attention to is breadth and depth. Are you a generalist or a narrowly focused specialist?

Breadth vs Depth

I only have been thinking about this because Sam Schapiro of Spiral Works published this paper on the importance of width and depth for creativity in Large Language Models. In this work, Sam finds that at least for LLMs, there is an optimal ratio of width and depth in neural processing for creativity.

Sam also pointed me towards this paper, The Associative Basis of the Creative Process, Mednick 1962.

This paper says that there is a difference between original thinking and creative thinking. While 1,033,239 is an original response to 12+12=__ it is not creative since it doesn’t have value. Tal makes original moves that are still valuable. He is creative.

Original thinking = generating novel ideas and challenges established norms by focusing on uniqueness

Creative thinking = forming new combinations from mutually remote associative elements

According to this associative theory, the more disparate the elements, the more creative the solution.

But when simply asked which is more valuable, Sam says “width,” and my intuition says the same. Humans are defined by our generalized intelligence. That’s the whole point of Politzki’s Law. And so to be creative, it is better to be a generalist than a specialist, especially in regimes of change.

Original Thinking vs Creative Thinking

If I think about my path so far, it has started as putting out a very original thing. However, that thing did not have value and looked more like a research project. It was a grain of sand in the universe. It was a very far out-of-distribution point in space. It has taken work since then through associative recombinations to take the aspects of my original contribution through General Personal Embeddings and combine them with other inspiration in the market. Since then, the structure has grown and been transformed as a shape in search of its ideal end form that has value.

I described a lot of my work in the last essay as a sort of searching, a searching for inspiration from the world. I think it has been something like this. Looking for associative combinations of 2-n different objects to combine and take the best of both. Kind of like taking the best values of multiple embeddings and combining them. The next stage of this journey will be a massive recombination that extends beyond the current state of the market’s technical solutions, market solutions, use cases, and combines it all into something that is the most “creative” thing I have ever built.

Zero to One

I remember reading this book and thinking it was amazing. However, the idea of a Zero to One product on its face, I think, is usually a romantic misconception. From the outside, it looks like it’s an overnight success, but I doubt it is usually.

My personal experience has been like a gradient of product-market fit. Where initial product shapes had no value, subsequent product iterations/transformations had some value, eventually people started paying, etc.

“In the AI era, product-market fit isn’t a binary state, but a dynamic, ongoing process that demands relentless curiosity, rapid iteration, and a willingness to challenge assumptions.”

• Bessemer Venture Partners, Mastering product-market fit: A detailed playbook for AI founders

But another very interesting point on all of this is that just as humans never achieve self-actualization, and evolution is not additive, it comes from ripping the old thing out and starting new, so too is the case for products. At any point, I could stop evolving the product and double down on a very narrow use case, and probably turn that into a lifestyle business. But I expect that I would be leaving money on the table for future better product iterations.

Failure Rate

In my last essay, I describe the versioning of my product as constant iterations of ripping apart the old product, taking the good, throwing away the bad, and reshaping it into a new and better form. It seems to me that this also mirrors how Auren Hoffman builds his team.

Auren also talks about how he brings on good people who grow fast and keeps them in a failure zone so that they are growing. He cites the ideal failure rate at 1/3 to 2/3, which is really out of the ordinary for most traditional companies. But like products, if you aren’t constantly failing, you are not learning enough.

These numbers are actually pretty accurate with how much I have kept and thrown away with product iterations.

The Full Picture

I don’t really like showing people my product, because I see so many flaws in it. When other people see my product, they see a snapshot of it in its current state.

But I see what it was, what it is, and path dependencies of what it can be. I see it as a very imperfect version of a product aiming to solve the larger vision of building the infrastructure for computers to understand us. I have many ideas for where to take it from here. For the first time, I feel extremely bottlenecked by the time and energy required for new developments.

In this sense, I’m receptive to feedback but it’s usually not very helpful. Most of it falls under categories of stuff I know I need to do but don’t have the time, things I’ve already thought of that don’t work for unintuitive reasons, or things they’ve parroted from Twitter as conventional wisdom that is not true for the specific type of business I want to create.

Unprofitability Signals Belief in the Future

I always remember hearing stories of tech companies like Amazon when I was younger, and people in the midwest always seem bewildered that companies can lose money and still be worth a ton of money.

In this video, Jeff Bezos confidently asserts that while it seems unusual that Amazon is such an unprofitable company, they are investing in the future. Amazon was a very different company than the one it is today. They have since invented many products.

They had more ideas for the future that they were excited about, so they held off on doubling down on who they were and invested more into who they could be. This is how I feel about my life as well.

Advice and Learning front the Market

I see many different forms this product can take. I saw this talk at Stanford by Shyam Sankar, CTO of Palantir, the other day and it captured my experiences so far. He explains the difference between selling into consumers and enterprises.

Consumers

Shyam describes learning from consumers as somewhat like a Gaussian Signal, where you see where some of the high signal is being used and you veer the product towards that. This is somewhat how our first iteration went. A bunch of people started using it in weird ways that were completely unexpected. You could learn from how they used it and focus the product around it. And you would need to make very real tradeoffs and cut a lot of stuff to focus your product for some customer segment. However, we mostly just added onto it to learn more instead, which brings us to the next category.

Enterprises

Enterprises follow the power law, where there are a small number of businesses that need your product the most. Ideally, these are businesses that you understand something the world has not yet where you know that in the future not just a small amount of businesses will need this, but EVERY business in the world will. Like LLMs several years ago.

For us, this has been developer activity, which in my experience so far has preceded broader adoption. Developers are tech-curious and understand things you can build with various new technologies. For us, developers came and built quite interesting things with the product, even before we introduced the API.

But developers can also be dangerous to pay too much attention to. If you look at MCP today, while that may be a long term signal of tool calling standardization, there are more builders than there are users of the tech itself. It is possible that it is an ephemeral fad.

The Frontier

New markets are in some ways an extension of the new frontier that Columbus ventured to explore in 1492. In this library, there is also this globe of how people thought the world looked before Columbus. You notice a couple of things. First off, there are obviously no Americas. Second but just as important, the rest of the map is also terribly incorrect. It wasn’t just that there was this missing piece. The fact that there was a missing piece was connected to the fact that general knowledge of the earth at that time was so incomplete.

In order to go outwards and be bold, we need to think that there is something more. Today, I would not set sail across the ocean blue. I don’t think there’s much more that contemporary cartographers haven’t mapped out. That frontier is relatively closed. The new maps look good to me.

However, when I look around the AI space, I see so many gaping holes. Even just in the AI Memory space, one could spend several lifetimes working on these problems. You can tell when you’re on the frontier when you can see the messiness of the status quo. I see endless white space.

My next piece will be a full mapping.

An Explorer’s Irreverence

Columbus was an outsider of sorts. Like Tal, he believed in something that was crazy. Interestingly, Christopher Columbus died a disappointed and bitter man, because he never found a westward sea route to the East Indies as he’d hoped. If he was as open-minded as a startup founder, he would realize that what he did find was much greater.

Columbus was a rule breaker. He didn’t accept the convention. He was an outsider that needed the world’s change like Naval.

Larry Ellison is one of the great business magnates of today and I’ve enjoyed catching up on his story. He says that it is his ability to question everything and think for himself that was never valuable in early life but has been the most valuable trait in starting a business. I agree. I’ve been so heads down that I’ve forgotten that’s what this is all about. Irreverent is a case study.

There’s a lot of noise in Silicon Valley. Many of the loudest companies from a year ago are now bankrupt. I’m perfectly happy tending to my garden, following my own individual belief system, viewing the world through my own unique lens, and moving forward. At this point, I’ve already solved the problem I set out to solve and now I’m just going beyond it.