This is an excellent question, and I'm disappointed this hasn't received more attention.
Secretary problem - applicable to anything where you have a lot of choices, such as marriage, getting a job (not just hiring), picking a business partner, anything you have to accept/refuse.
The solution is you reject the first 37% of applicants, use that to establish a benchmark, and then pick the first one who is better than any rejected. You're 37% likely to find the best candidate. But such is life. In other words, once you marry someone out of 10 choices, there's 63% odds you'll find someone better later.
There's Amdahl's Law/Gustafson's Law, explained nicely here: https://www.embeddedrelated.com/showarticle/1033.php
Basically, upon first observations, Amdahl's Law seems true... even if you put 10x more effort into something, it seems like the total output is 42% higher. But this total output being 42% higher means that you can also do X, Y, and Z, 42% faster. Having Z 42% faster leads to A being 20% faster, leading to B being 8% faster, and B is a feedback back into the loop.
In more practical terms, much faster computers means software is written slightly faster, but slightly faster software allows for more complicated electronics CAD, which allows us much faster computers, which allows for even more advanced technical developments like Internet and cryptocurrency.
In short, if you're locked on Amdahl's Law (minor gains for tremendous effort), find a way to utilize Gustafson's Law (looping the gains back).