An optimistic founder needs someone on the team imagining disaster
Optimism belongs in strategy and in how a founder treats people. Paranoia belongs in the systems that fail when nobody is looking.
Assume the best but hire paranoid people.
In 2008 Twitter went down so often that its error page, a cartoon whale lifted by birds, became a symbol of the company. In 2009 an intruder guessed his way into a Twitter employee’s email, reached company documents stored online and sent them to a technology blog. Both episodes came from the same kind of gap: a small team moving fast, confident that the service would grow, and too few people whose job was to picture it breaking.
Evan Williams, who co-founded Blogger and then Twitter and was its chief executive until October this year, puts the lesson in six words. The two halves describe different jobs. A founder who expects the worst from people will build a suspicious culture and make slow, defensive decisions. A founder who expects the best from systems will be surprised by outages, breaches and lawsuits. The answer is to split the roles: stay optimistic about the direction and the people, and hire some people whose instinct is to ask what could go wrong.
It holds in any area where failure is quiet until it is total. Security, infrastructure, finance and legal work all reward people who assume that something is already broken and go looking for it. Andy Grove made a similar argument about competitors at Intel in the 1990s. A paranoid operations engineer is the same idea applied to servers.
The line is weaker when the paranoid people end up in the wrong seats. A security team that can block any release without explaining the risk will slow a company to a crawl. Paranoia pointed at colleagues rather than at systems turns into politics. And a company full of cautious people may never ship anything bold at all.
There is also a tension the line leaves out. Paranoid people are often not pleasant to work with, because their job is to say no. They need a founder who takes their warnings seriously even when the warnings are inconvenient. Hiring them and then overruling them every time is worse than not hiring them, since it creates the feeling of safety without the substance.
The workable version is specific: name the few places where a failure would be catastrophic and put a person there whose job is to worry, with the authority to stop things.