lichess.org
Donate

Puzzle rating

Well guys, to explain my point a bit better I bring you this example:
http://es.lichess.org/training/55237
This problem was categorized as hard not long ago. The fact of the matter is that it's been played more than 40.000 times, and due to multi-accounters and anonymous players that learnt it and memorized it, its score has decreased significantly.
We need the computer to select puzzles that are less voted so they are played from time to time and this stops happening.
So I assume you have some proof that this is why the puzzle's rating has gone down, and an explanation for why it can't possibly be that the puzzle just isn't as hard for most as it is apparently for you?
I'd wager the lowish rating comes from the moves being quite natural.

1. Most reasonable looking check
2. check with material win
3. check with fork
4. grab free queen

In other words, a lot of people probably get it correct intuitvely even when they didn't think it through.

Personally I found it not so easy because I was looking for something better all the time which you have to do with these puzzles on here when you want to get a very high rating. And once over 2200 you get extremely paranoid when you don't find a checkmate in a position like this.
I would agree that this could also be a problem with a rating >2000 if it had some sort of trick to it that sorts the "guessers" out.

I have seen way way harder problems than this one though.
"The fact of the matter is that it's been played more than 40.000 times, ... ."

@AlcianGreen, it was only 34,666 times when I looked at it a few minutes ago. The credibility of your assertion suffers when you're careless about such details (trivial as they may seem), especially when you have a conspiracy theory within the same line.

(I think I failed this puzzle some time back. I tend to suck at material advantage problems because I got a little used to solve-for-mate puzzles from books many years ago.)

Hmm, basically, you're asking for a feature that disregards popularity when serving random puzzles. Not really a proper workaround, but you can manually go through http://en.lichess.org/training/1 to http://en.lichess.org/training/60120 by simply changing the URL. Of course, this also disregards the rating differential among other factors.
@AlcianGreen

This puzzle has a rating of almost 1900 (!).
That's a puzzle rating that is hard for quite a lot of players.

The puzzle itself is in my opinion not so difficult, but I assume most people cannot keep their hands from moving the mouse and move too fast.
If you take a bit of time, then you will probably agree that this puzzle is not a 2400 rated puzzle.
@achja, I have yet to encounter a 2400+ puzzle. I think I've only seen a 2300+ puzzle so far and it gave me -3 when I failed it. :3
Here's some proof, I recently missed a 2400+ puzzle :
http://i.imgur.com/kKkyJFg.png

@achja, ah, sorry. I wasn't questioning your claim. I was just sharing/noting that I haven't encountered them naturally (through the random/popularity selection algo) yet. Then again, my rating might not be high enough even if/when I choose the hard setting.

And, haha, you just raised its rating some more! :P

Weird thought: what if these puzzles were actually genetic problems that evolved? That would be so much sci-fi fun.
@weakwithwords
I didn't feel questioned or something.

Just wanted to share the 2400 puzzle existence with everyone else reading here.

A few days ago I actually solved a 2400+ puzzle, and saw its rating go to 2393 or something. That was kind of weird and fun.

And surprisingly I fail some 1800 puzzles every now and then. I guess I still have some tactical blind spots to work on.

Here's a 2400+ puzzle I just solved : http://en.lichess.org/training/31905

I think the rating is so high because lots of players see one line they think is winning, but it is leading to nowhere.

When I realized that, I found the winning move after some thinking (42 sec). Not so very very difficult, and actually quite an elegant move.

This topic has been archived and can no longer be replied to.