Search "user:Berder"

Absolutely - chesstempo uses a threshold of +1.75 for alternate moves and does not have this problem. lichess needs something similar.

Thibault could give an option (enabled by default) to use your browser to help analyze other people's games. The way it could work is, when you press "analyze" the moves of the game are split up and s…

Larissbug, that's likely just because they had calculated that move in advance, so they could make it quickly when the time came.

4) Pawns need at least 1 empty square on all sides so they can move freely. Try to avoid placing them too close together.

I don't see any problem. Black resigned and lost. Maybe someone already fixed your game.

Yep, browser analysis confirms broken puzzle.

This puzzle is a complete mess. Analyze it with the computer: you see that Rxd6 is a blunder by black, turning a +7 eval into a mate in 8. And yes, h5 instead of Rxf8 is a blunder by white. Rxf8 is ma…

Can you link to them? I didn't happen to notice any suspicious games.

Looking at some of Think-Chess's recent rated games he does seem to have plenty of inaccuracies, mistakes and blunders. If he was actually cheating he's been very clever about hiding it.

Rather than any of these features I'd prefer if all the training puzzles were fixed so that every move with a winning eval is at least accepted as an alternate move.