"People who work in CS are known to work with scientific methods. "What if" is simply not enough."
That's hilarious. Try to remember this next time you put a print statement in the code or adjust some constant until it's right.
The problem here is that we have no way of either experimenting meaningfully or testing most of the things we'd like to know. We can describe tiny parts of reality and assume that some correlation is in fact causation, but without actually doing something it's impossible to conclusively prove what effect will any one action have.
That's nothing new - and it's not that unseen in programming, either. I think that "waiting for a proof" in this case is just an excuse; we need to get the best approximations, of course, but in the end we just need to find the courage to go and implement a solution without a certainty that it will be the best or even good.
In other words, when your service is on fire because thousands of users want to access it simultaneously (good for you) the first thing to do is to bring it back by any means necessary, including tools like intuition and wild guesses, and formalize the process afterwards.
That's hilarious. Try to remember this next time you put a print statement in the code or adjust some constant until it's right.
The problem here is that we have no way of either experimenting meaningfully or testing most of the things we'd like to know. We can describe tiny parts of reality and assume that some correlation is in fact causation, but without actually doing something it's impossible to conclusively prove what effect will any one action have.
That's nothing new - and it's not that unseen in programming, either. I think that "waiting for a proof" in this case is just an excuse; we need to get the best approximations, of course, but in the end we just need to find the courage to go and implement a solution without a certainty that it will be the best or even good.
In other words, when your service is on fire because thousands of users want to access it simultaneously (good for you) the first thing to do is to bring it back by any means necessary, including tools like intuition and wild guesses, and formalize the process afterwards.