Ask HN: How do you solve big, complex problems?
My process is to:
- put all the problems/insights/concerns (that I am aware of) in front of me;
- stare at them until I have a broad-strokes solution;
- then, refine the broad strokes solution, and see if it still works;
- keep doing that until I feel like the only thing left to do is to implement the solution.
I find that the "stare at it until I see how to solve it" part tends to be very slow and incompatible with collaboration.
How to approach problem-solving better? Any recommendations? Books?
10 comments
[ 2.1 ms ] story [ 35.5 ms ] threadI think it's a bit more positive in outlook but it has a chilling description of the scariest way high-tech projects go wrong... By approaching an asymptote.
When your solution is not adequate to solve the problem you can keep investing in it and feel you are getting closer and closer and closer but you never get there. It's a high risk in modern AI projects, and it was also amusing to see in people who tried to fly before the Wright brothers
https://www.youtube.com/watch?v=fw_C_sbfyx8
In particular people thought the problem of flying was about getting thrust and getting lift. It's really about controlling the motion of the plane in 3 linear degrees of freedom and 3 rotational degrees of freedom. You need each and every control surface on a plane so you don't tumble. If you arrogantly presuppose 5 degrees is enough you could keep improving your machine... you might even get off the ground but you will encounter it again very quickly.
Also if its about personal problems that involves your own values or considerations, I would suggest the WOOP method. Read "rethink positive thinking"; its a slow read but important to fully understand the purpose.
There's something to be said about mindless increases in observation and familiarity even when focused mindful action is fairly urgent.
Either way you will often come to the same conclusions but if not, the relatively mindless session will sometimes yield ideas that you could not come up with any other way.
>How do you solve big, complex problems?
The most direct thing I try to do is have straightforward simplifying solutions that are no more complex than necessary. And that will lead to further simplification if possible.
Looking at all the variables at once may make things seem unsurmountable, but there is no other way to get the big picture even when nobody will ever fully understand it at first glance.
Edit: this is basically what AI is supposed to do, come up with worthwhile conclusions by just staring at the data. If you're also a living breathing subject matter expert and you do this in parallel that should be even better.