The puzzle ratings certainly aren't elo-based; nobody should be rated 65535 (I think that it would have been even higher though it overflowed the max int). A rating of 65535 elo would take over 10^150 puzzles rated 3000 solved consecutively to achieve (which would take longer than the amount of time until the heat death of the universe). You can often gain 10 rating points for solving puzzles rated 400 lower than you; it would be 1-2 points in live chess. People also often have way higher puzzle ratings than live/daily chess ratings, which shouldn't be the case.
I get that the system is meant to avoid punishing people for solving puzzles and solving too slow (hence you get at least +5 if you get 50% or more in a puzzle), but there should be a punishment if you spend too much time, as this means that you will likely be unable to find it in time in an actual game. Maybe there should be blitz mode where you do get punished for taking too much time (since that's what would happen in an actual game) and a standard mode (where your score does not depend on time; only whether you solve the puzzle correctly or not). At the very least, you should lose rating if you get 2/4 moves correct in a puzzle rated 500 lower than you.
Also, I don't think people should be rewarded if they found the correct moves initially but did not find the full follow-up sequence, as a tactic often only leads to an advantage if you get the entire move sequence correct (though it is possible that the engine creates a surprise response that's hard to refute and you did get the essence of the tactic by making the correct first move).
I don't think that the guy with 65,000 points had to solve 10^150 puzzles. There reportedly are a limited number of high-rated puzzles. Some people have evidently memorized them.
And they used to subtract points if you took too long to get the right answer but that upset a lot of people. If you search the forum history you can see many of those posts threatening to cancel premium memberships. So now you can get at least five points on a puzzle if you solve it correctly-- even if you take all day to solve it.
And they knocked thousands of points off people's ratings at least once.
At any rate, they have tried many variations over the years.
They added Puzzle Rush and Puzzle Battle for people who like time pressure. They added the survival mode for those who want to spend more time.
No one solution will make everybody happy. No one solution will get people to buy and keep premium memberships.
I think they are trying to make as many happy as possible.
The puzzle ratings certainly aren't elo-based; nobody should be rated 65535 (I think that it would have been even higher though it overflowed the max int). A rating of 65535 elo would take over 10^150 puzzles rated 3000 solved consecutively to achieve (which would take longer than the amount of time until the heat death of the universe). You can often gain 10 rating points for solving puzzles rated 400 lower than you; it would be 1-2 points in live chess. People also often have way higher puzzle ratings than live/daily chess ratings, which shouldn't be the case.
I get that the system is meant to avoid punishing people for solving puzzles and solving too slow (hence you get at least +5 if you get 50% or more in a puzzle), but there should be a punishment if you spend too much time, as this means that you will likely be unable to find it in time in an actual game. Maybe there should be blitz mode where you do get punished for taking too much time (since that's what would happen in an actual game) and a standard mode (where your score does not depend on time; only whether you solve the puzzle correctly or not). At the very least, you should lose rating if you get 2/4 moves correct in a puzzle rated 500 lower than you.
Also, I don't think people should be rewarded if they found the correct moves initially but did not find the full follow-up sequence, as a tactic often only leads to an advantage if you get the entire move sequence correct (though it is possible that the engine creates a surprise response that's hard to refute and you did get the essence of the tactic by making the correct first move).