That doesn't follow (because "most of the stuff on top of HN" is a much stronger claim than what you said here) but I'm happy to answer anyway. The answer is that it frequently doesn't work. What we don't know is how many other cases there are where people are getting away with it. We can't know that; by definition, we're never going to know it.

Even there, though, one can make educated guesses, based for example on how many cases come up where people got away with it at the time, but then evidence comes to our attention later and we can figure it out in retrospect. Such cases are useful because then we can extend our software to catch them in the future.

I'd never claim that such manipulations never work; obviously we can't know that. It's possible that superclever manipulators are rolling HN in their hands like a piece of silly putty. All I'm saying is that if "most of the stuff on top of HN comes from [that]", then I don't know the first thing about my job.

Yeah, you are right. Sorry, I exaggerated there, HN is the best mainstream source for technical stuff honestly.

Still many posts that make it to the top feel like commercial advertisement and I am pretty sure they use (their?) sock puppets to get the first upvotes to cheat the algorithm.

It's easy to create 20 accounts and just switch between them (and IP) when you do your normal HN procrastination to validate them and then get those initial 20 upvotes to go up to the top.