Work Somewhere Dysfunctional
The rewards are great if you can survive…
The other day I was talking to an engineer about developing and refining good process. My most effective and valuable tricks for getting stuff done were developed in dysfunctional places (i.e. the government.) If your technique needs ideal conditions in order to work, then it doesn’t work at all. At some point you need to stress test it, and you might as well start off that way before you’ve gone all over town talking yourself up.
This is a big part of why I decided to advance my work on software safety by working for a defense company. No approach to AI safety is worth discussing until it’s operational in a place where unsafe outcomes are likely. Before it has yielded results in that kind of environment it’s just noise, white wash and wishful thinking.
People who want to study earthquakes need to go where there are fault lines. People who want to solve hard problems need to go where those problems live.
Unfortunately, places with hard problems that haven’t been solved are also — inevitably — kind of dysfunctional. If you don’t have a ‘good enough’ solution and you don’t have innovation in progress then the only way you deal with a hard problem is with maladapted strategies. Solutions that rob Peter to pay Paul in some way. Perhaps the organization is choosing to just let…