It’s been over a year since I joined Stripe and I have amassed some thoughts. I’d never worked in a “Silicon Valley” (scare quotes because I feel silly saying it) company before, and I think most of these observations are applicable to the ecosystem at large rather than Stripe specifically. (Maybe later I’ll try to put on paper the things that I think are unique to Stripe; there is lots to write about on that.)

First, for context: I’m a simple prairie boy who is headquartered squarely in flyover country. I’m a half decent programmer who’s failed spectacularly in my business endeavours. Contrary to conventional wisdom, failure is not “learning things the hard way” when it comes to business; when success is an unlikely outcome and sources of failure are common, it’s easy for people to fail at the same thing repeatedly for their whole life, even if they take a different approach each time. So, instead of learning from failure, we should strive to learn from success wherever possible.

I think it’s fair to say that Stripe is an example of a successful company, and Silicon Valley is itself an example of a successful startup ecosystem, so I’ve been learning a lot.

The first thing that struck me is the level of institutional knowledge that exists in Silicon Valley. Nobody is confused about the business model of startups. Everyone understands, in a general sense, the lifecycle of a startup, and the risks associated with each stage. The liquidity of this knowledge is magnified in two obvious ways: (1) the fact that the workforce is extremely fluid (with an average job tenure of about two years) means that tactical knowledge is quickly shared between companies, and (2) the fact that the set of major investors (and therefore startup board members) is small enough to guarantee a certain amount of high-level information sharing. Both of these factors are sensitive at the micro level (after all, no company wants to leak information which is potentially proprietary) but evidently beneficial at the macro level.

The next thing that struck me is the extent to which software businesses are driven by compounding effects. This should not have surprised me, since I’ve worked in software all my life. But until you see it working at scale, it’s difficult to understand. The entire economic engine of software owes its existence to the idea that the marginal cost of software can be rounded to nil, but exploiting this fact correctly requires a mindset that is only possible in a company that was founded by a programmer, and in which its investors have seen this playbook executed 100 times before. (What does this mean? It means almost never spending human energy on the first-order effects of recurring problems, and instead developing automated–or at least computer-assisted–solutions to them. This is easier said than done, of course.)

After some time, I realized that the above two elements are in fact enabled by a third element which seems obvious but is not: getting lots of smart people into the same place. (I don’t know if geographic location is a hard requirement, but certainly the same “idea space”.) This sounds like a cliché but it’s in fact a strict pre-requisite for the sort of ecosystem that is described by the other two elements above.