Thoughts on (Amazonian) leadership

The secret sauce to Amazon's success is an obsessive compulsive focus on money.

> Amazonians talk about "one-way doors" and "two-way doors", and it is quite true that many decisions are can be reversed... but that doesn't always mean that there is no cost associated with reversing a decision.

Not many companies get this right. They seem to be very polarized on this mentality: There are companies that are too cavalier about decision making, have this "do something, anything" attitude, try too many risks, and then end up breaking things that are hard to reverse or painting themselves into a corner. And then there are companies that just get paralyzed by every little decision they have to make and end up defaulting to doing nothing out of caution.

I think it's really rare to find a company that can consistently apply the right amount of caution and care, but still actually get things done. There's often no internal framework or thought about whether something is a "one way" or "two way" door, as the article puts it. It's just (for example) Product saying "let's do it" and Legal saying "no, don't do it" and nobody knows how to solve this without having a default decision or flipping a coin.

“Paxos as a service”

Ok. I have just spent 5 minutes noodling on that one and … I love it even if I cannot work out how to make it work …

Definitely coming back to that

To be clear, I'm not saying it has to be literally Paxos as a service. There's a bunch of ways you can take operations from different sources in a distributed system and produce a consensus view of what happened first. But aside from trivial and low-performance options (e.g. everyone contends on a single DynamoDB item and uses conditional writes) Amazon hasn't made any of the possibilities available.

In my experience, Amazon genuinely is focused on the customer.