A fighting retreat
Maybe the toughest thing about raising a Series A is the knowledge that if you succeed, a company you love will be very different at the end of it. Antithesis is one of the most unusual places I’ve ever worked. We prioritize the long term – in our business strategy, our technical decisions, and our relationships with our colleagues. We’re intensely deliberative and collaborative, happy to do things differently from the rest of the industry (when we have a good reason to), and maybe a teensy bit cult-y. But the most special thing to me is that it’s an exceedingly high-trust environment where people still feel that everyone is on the same side and politics is minimal. I’ve never seen that maintained at a company with over a hundred people before, and I still don’t really know how we did it, but it’s magical.
So I’ve been giving a lot of thought to how we can keep that going as we enter hypergrowth, hire a zillion people, and come under tremendous pressure to deliver consistent results. This is a letter I wrote to the company on the topic last month. I hemmed and hawed over whether to publish it, but ultimately decided that there’s probably another founder out there struggling with the same questions. If you find this useful or interesting, send me an email, and maybe we can chat over coffee the next time we’re in the same city.
Almost all small organizations feel different, and almost all large organizations feel the same. Why is this? I think there are three reasons.
-
Max Weber has a wonderful essay in which he observes that as small-scale societies grow in complexity, they tend to transition from what he calls “charismatic authority” to “bureaucratic authority.” Charismatic authority is social cohesion born out of loyalty to a particular individual, or more generally stemming from interpersonal relationships within a group. This serves as an astonishingly powerful glue for small groups, and enables them to run more efficiently than large groups, but its scalability is limited. Beyond a certain size, not everybody can work closely with the leader. More significant than that, though, is that people also know less and less about each other. There’s a size below which you can say: “treat me fairly, I’m a fellow X” and you’re almost certain to get a good reaction, and another size above which that stops working.
And so the society must transition to bureaucratic authority – rule by impartial procedures and laws and regulations, which are more reliably fair. But the key insight is that while “fairness” sounds great, you actually lose something while gaining something. For example, a family shouldn’t be “fair,” it should be better than fair, a place where people think nothing of their interests or what they’re due, but sacrifice for each other. In the same way, in small groups you might be lucky enough to be treated with grace rather than justice. That’s better so long as it works, since as Hamlet says to Polonius: “Use every man after his desert, and who should 'scape whipping?” But this only works so long as you can trust others to reciprocate, and at a certain scale trust is always lost, and so you must replace it with law instead.
-
There’s a curious fact in evolutionary biology (which has since become a funny meme) which is that things evolve into crab-like shapes over and over again. It’s an example of convergent evolution — many environments provide similar sorts of challenges, and if the “become a crab” solution works for one organism, it probably also works for another. Well, we see the exact same dynamic play out in software architecture, and also in institutional design. For example, the solution of having a tree-based hierarchical organization is an effective solution to certain coordination problems and easy to evolve into, so we see it recur over and over (though not everywhere).
The key to understanding why carcinization happens to companies as they grow is that small companies all have very different evolutionary niches — different customers, different competitors, different price points, and most importantly different types of employees. This allows them to also support radically different cultures. In contrast, every large company actually has a hugely overlapping set of concerns. One I’ve mentioned already is orchestrating and motivating a large group of people. Large companies also spend a large fraction of their energy on maintaining compliance with various laws and regulations, and evolve carcinized responses accordingly. There’s a more hand-wavy sense in which large companies are all eventually in the same sort of business, and get measured with the same capital efficiency metrics. Most large companies have essentially interchangeable employees. Finally, public markets punish any kind of deviation from the very standard large company playbook. All of this inevitably produces a sort of cultural carcinization as well.
-
The third reason I think is simultaneously the most powerful one and also the dumbest one. It’s just mimesis. Why do fads begin and grow? Why do Sneetches need to have stars on their bellies? Because human beings are comforted by doing what everybody else is doing. Or, if you want to be negative about it, because people are lazy and cowardly, because they would prefer a higher chance of failure so long as nobody can say it was their fault for being a weirdo. My own view is more moderate: this is neither a good thing nor a bad thing, but it’s a thing about human beings that you need to bear in mind, because it certainly is a thing.
It’s interesting to speculate about why large organizations are more affected by mimesis than small organizations (though not universally — note that basically every single YC company is the same company). I have two theories here. One is personnel: people tend to join small companies who for one reason or another don’t want to work at big companies (all of which are the same), and people tend to join big companies who like doing the same thing everybody else is doing. The other theory is scar tissue: anytime your company has a big failure or a giant disaster, it’s very easy and natural to blame it on whatever the weird thing was that you were doing, and decide in your post-mortem to become “more normal.” Often this is even correct! But people will tend to do it whether or not it’s correct, so there is constant pressure to evolve in the same direction.
Why do I care about all this? Well, I care because there’s a reason I decided to leave my happy and successful jobs at big companies and do this weird thing with all of you instead. I like the culture that we have. I think it’s more rational, more sensible, more humane, and more fun than the generic big company culture that you will get at most big companies. My guess is that since you joined a startup you feel at least somewhat the same way. So I would like to keep this going for as long as we can.
Notice that I say “as long as we can,” not forever, because I don’t think we can keep it forever if we become very successful. The three forces I listed above, which make every large company culture evolve towards a bland mediocrity, are just too powerful. If we get big enough, eventually they will win. (Hopefully none of us will care, because we will be lying on a beach sipping our beverage of choice.) The title of this post is “A Fighting Retreat.” You don’t stage a fighting retreat in order to win, but in order to lose more slowly. So that is my cultural goal for 2026 — see the company almost double in size, see it come under enormous short-term pressure to perform, see it exposed to many of the same market and regulatory pressures as everybody else, and lose culture slowly. Only get pushed back a tiny bit. Make the enemy pay for every step.
So how will we manage to do this?
Resisting the Weberian transformation from interpersonal decision-making to procedural decision-making is something that a small and select group of organizations have already managed to do, so we don’t actually need to break new ground here. I mentioned in the AMA last month that one inspirational example of such an organization is our new partners at Jane Street, who have scaled to over a thousand people while keeping their norms of informal and relational governance. The way all of these groups have managed to do this is by remaining “high-trust” — that’s actually a technical term from sociology but it’s pretty much just what it sounds like.
Basically, if you maintain very strict standards for belonging to group X, and you make belonging to it extremely distinctive, then people keep feeling like if you are a fellow X they can expect certain behaviors from you, even if they’ve never met you before. This lets you keep some of the “small town feel” even as you vastly exceed Dunbar’s number. A great case study of this is Hasidic Jewish diamond merchants who are able to lend on credit with vastly lower transaction costs than less insular groups. The good thing is this is just all the stuff we were going to do anyway — maintain a very high standard of behavior, maintain our esprit de corps, help newcomers to assimilate to us, punish infractions and fire people for gross infractions. That was all part of the plan anyway, but as a nice side benefit it will also help us push the limit of how far we can keep this vibe.
The next one, carcinization, is a little bit harder to avoid. Insofar as we come under the same evolutionary pressures as every other company, we are bound to become a little bit more like them, and some of those pressures (like regulatory/compliance concerns) are pretty much inevitable. I think the defenses here are twofold. First: by staying focused on our mission, our core competencies, and our immediate objectives, we will be protected a little bit. There’s a psychological concept called “keeping your identity small,” which helps people to be more adaptable because the fewer things you make a core part of your identity/personality, the easier it is to change. I think there’s a corporate equivalent of this which helps to guard against mission creep and associated cultural carcinization. Once again, we should be doing this anyway.
Secondly, we can just not go borrowing trouble. In software, premature optimization is often the root of evil, and I think this is even more true in human social structures. Yes, someday we will need strict rules for fairly managing access to the electric vehicle chargers in the garage, and an internal web app for coordinating it. But do we need that today? Or can the informal Zulip group that exists for this purpose last us another year? What about another two years? Very well-meaning people often look down the line, and put tremendous effort into making things more “scalable” for the future — more process, more bureaucracy, more levels of management. Fine, maybe we will need that stuff next year. In that case, let’s do it… next year. The key insight is that none of these things are costless. Each one costs us a tiny bit, and put together they cost us greatly, so we should always push back on them. Do we really need this right now? Can we go another few months without it? If so, let’s do that.
The last and most powerful force that will ruin our culture is mimesis. The good news is that this one is the easiest to fight. We just need to keep the courage to be weirdos when being weird makes more sense. This doesn’t mean we should ignore what the rest of the world is doing. Often the rest of the world is onto something! We should be intensely curious about why the rest of the world is doing certain things, and adopt their ways if we think they’re onto a good idea. But “everybody else is doing it” has never been a sufficient argument by itself for doing something at Antithesis, and it never will be.
On some level, the entire existence of startups is because the rest of the world is often wrong about something. If you believed in the Efficient Market Hypothesis, you would find a way to maximize your ownership of index funds, not scary illiquid Antithesis shares. You would work at a consulting company, not a company that thinks the entire rest of the world is wrong about how they test software. But you didn’t make those choices, you work here instead. I think that’s a good decision. Dave and I have both made careers out of doing stuff that the rest of the world thinks is dumb and wrong, so let’s add one more thing to the list and stay scrappy and high-trust and contrarian, way past the size that anybody thinks it’s possible to do that.
