Mate impossible = draw.
Not here not now. In general yes.
> There are numerous positions that by this rule should be draws that are declared lost.
Let's attempt to quantify this. Claude Shannon estimates the number of 40-move chess games to be 10^120:
I estimate the number of possible dead positions to be less than 10^10, which is about 0.000000000000000000000000000000000000000000000000000000000000000000000000000001% of Shannon's number (and that number only covers 40-move games). Even if my estimate is wrong, for those positions I'm continuing to write a helpmate solver (although probably it's easier for anyone else to write it from scratch than for me to integrate it into Stockfish) although it's not yet performant:
Other research on the topic can be found at chess.stackexchange.com/a/22764
What matters is positions, not the number of games. State space complexity, not game tree complexity. So your count is off by a factor of about 10000000000000000000000000000000000000000000000000000000000000000000000000000x than the number you posited. Yeah - that is the correct number. Unspoke point there being let us use exponential notation, yeah?
And again this is a red herring. The question is does this rule make lichess a more or less enjoyable place to play? Does it provide a better or worse user experience for players? I think if you look at things remotely objectively, the answer to this is extremely evident.
@Toadofsky very interesting. But I think that there aren't that much legal positions with closed structure. (I mean a Knight blocked by 2 or 3 pawns of his own side can get back in some tricky ways, but a bishop can't).
Everything starts with at least 4 pawns for each side and each pawn blocked by another pawn, if more pawns then they must be connected and blocked. Kings can't attack the pawns in no number of legal moves. If extra pieces that are not bad bishops (same color as your pawns) or own pawn blocked knights then there's no draw.
And I think that covers all the legal positions.
It is extremely evident. This is why we are using the current system instead of the old shitty system.
To everyone who says this "problem" should be easily fixed by just letting a computer evaluate the end position - which is the most popular "solution" to this problem (which is NOT a Lichess problem, it's a FIDE problem) - have you considered how much CPU time would be devoted to "watching" every game in play? Yes, most positions could probably be sorted in a few seconds, but this could happen in ANY GAME AT ANY TIME. By all means, program a computer "arbiter", then purchase a ton of high-end hardware, pay for the running of said hardware and give the infrastructure to Lichess.... oh wait, it's probably not that easy otherwise the issue would be resolved.
Sorry, I just find these posts annoying; where people assume that a "bug" is Lichess's fault without doing any research: the top 2 are this and en passant. Both of these can be found with a quick search on the Lichess forum. For this, if you have to blame something or someone, blame FIDE.
@lovlas this peculiar implementation is literally making people think the site is buggy, and it occurs on a daily basis. If you're going to argue in favor of this rule it's certainly not because it improves the user experience! This feels very much akin to how a discussion about the chess player notes in a chess.com board room might go. Somebody who has the ear of management there clearly feels they do good. To literally every other person looking in from the outside, you can't help but wonder what in the world they're thinking. Bubbles.
En passant is also making people believe the site is buggy. People having poor internet connection also makes them think the site is buggy. We could literally be handing out $100 bills for free and people would find something to complain about.
@lovlas Completely agreed. Except there's only one of those that is unique, and by choice, to Lichess.