Whatever problem has you frustrated today, it's almost certainly worth a few moments to consider "what would need to be true for us to not have this problem at all?".

Generally speaking, the way to not have a problem is to look upstream, and see what conditions lead to you having the problem in the first place.

Running out of gas is a real world example of a problem best solved by not having the problem at all. Upstream of running out of gas is habitually stopping to get gas when the tank is below half.

That example calls out something else: There's problems, and there's problem patterns. Looking upstream doesn't solve the acute problem of being out of gas right now. Too late for that, now the problem must be solved with a call to AAA or a walk to a gas station or a helpful Samaritan in a pickup. But the pattern can be curtailed: always getting gas when the tank is below half will make it so you never have to solve this "I'm out of gas" problem ever again.

Heading upstream is also good at preventing related problems you don't have. I think of this as preventing families of problems. Lets say a natural disaster hits your community. Hurricane, fire, flood, take your pick. Authorities issue "GO NOW". If you habitually fill the gas tank when it dips below half, you're good. If you don't, you might now be running out of gas somewhere AAA isn't going to help you, and now you have a Life In Danger problem on your hands.

ChatGPT, Claude and Gemini are all really good at helping us brainstorm ideas. What would need to be true for this to not be a problem in the first place is a great prompt to wrap up a thread with an LLM on how to solve a problem.

Part of that is being willing to timebox outlandish-sounding ideas and earnestly consider them before (usually) discarding them. Twenty minutes is often enough. If you do that five times a month, you'll have spent a little over an hour and a half. If you think you're good at generating ideas now, it's a good test. Ninety minutes is less than 1% of a standard work month.

You can be a great problem solver, and great problem solvers are almost always rewarded in software development. But much more valuable is the one who prevents problems or whole families of problems from needing solving at all. if you're interested in your job being AI-proof as an engineering leader - and I don't know if we should or shouldn't be worried about that, but plenty of people seem to be - I can't think of an easier way.