[robotelvis]

In my experience the best approach is to first try to solve the problem without having read the prior work, then read the prior work, then improve your approach based on the prior work.

If you read the prior work too early to you get locked into existing mindsets. If you never read it then you miss important things you didn’t thought of.

Even if your approach is less good than the prior work (the normal case) you gain important insights into why the state of the art approach is better by comparing it with what you came up with.

[dpatru]

A decade ago I read this same advice in "The Curmudgeon's Guide to Practicing Law": spend at least a little time trying to solve the problem before you look to how other's have solved it. One benefit is that occasionally you may stumble on a better method. But the more common benefits is that it helps develop your problem-solving skills and it primes you to understand and appreciate existing solutions.

[brookst]

What if you’ve already read the prior work before trying to solve the problem?

[kortilla]

Then you’re very unlikely to come up with a novel approach. It’s very difficult to not let reading “state of the art” research put up big guardrails in your mind about what’s possible.

All of the impressive breakthroughs I saw in academia in the CS side were from people who bothered very little with reading everything related in literature. At most it would be some gut checks of abstracts or a poll of other researchers to make sure an approach wasn’t well explored but that’s about it.

The people who did mostly irrelevant incremental work were the ones who were literature experts in their field. Dedicating all of that time to reading others’ work puts blinders on both your possible approaches as well as how the problems are even defined.

[agumonkey]

Maybe some people tried to develop out-of-the-box sessions to force investigating absurd axioms and see how it goes.

[HelloNurse]

Worst case: you don't have a fresh perspective, but you have learned something and you can try plenty of other problems.

There's also a fair chance of finding possibilities that are "obviously" implicit in the prior work but haven't yet been pursued, or even noticed, by anyone.

[ibejoeb]

In all seriousness, if you're cool with it, LSD. Or anything else that can take you out of the ordinary course of though.

[cubefox]

> If you read the prior work too early to you get locked into existing mindsets.

I agree, though in some cases coming up with your own ideas first can result in you becoming attached to them, because they are your own. It is unlikely for this to happen if you read the prior work first.

Though I think overall reading the prior work later is probably still a good idea, but with the intention not to become too impressed with whatever you come up before.