The New "Brilliant" Move Algorithm in Action!!
You have two choices of a mate in 7. However, only one of the two options awards a "Brilliant!!" from the Chess.com Game Review. Why is that? It's simple: Chess.com's new algorithm for classifying Brilliant moves is working.
The Old vs the New Definitions
Chess.com's ambitious evolution of the Game Analysis and Game Review features have historically had one particularly easy target for criticism: Brilliant moves. From 2020 until recently, it was easy to find handfuls of forum threads declaring, "Yay! I made a brilliant move... but why was it brilliant?" Some assignments were useful - most notably in endgames were deep analysis and calculation confirmed that only one move in a position secured a win or draw. However, for every useful Brilliant move, there seemed to be 10 mysterious or bizarre ones with seemingly no value.
In the March 2021 State of Chess livestream, founder Erik Allebest confirmed how Brilliant moves used to be determined:
"The previous way of delivering on Brilliancies was a little bit about what the engine didn’t expect and found later... as it parsed through the line."
Truly, the old definition let Brilliant moves be assigned by machines for machines. Compare that method vs the new definition as outlined in the November 22 update:
"Brilliant Moves must now sacrifice material in some way and must be the best, or nearly the best, move in a position."
This promised a more human-friendly algorithm. But how does it hold up in practice? A recent casual club game provided two useful examples for showing the new algorithm in action.
Example 1: Sacrificing Material for an Attack
Black was gravely behind in this Italian Two Knights game. White just played Qa4?? allowing Black to come crashing back into the game. Can you find the only good, "Brilliant" move?
This is not a very difficult move to find, but nonetheless it fits the new definition perfectly well; Black is sacrificing material by "hanging" the knight on a5 (it is most certainly poisoned), and this move is the only good move in the position.
Example 2: Two Options for Mate in 7, One is Brilliant
The ending of this same game helps illustrate the new Brilliant definition (and perhaps to a fault). White ignores the immediate threat and falls into a forced mate. Two options exist for carrying out at mate in 7 moves, but Game Review only marks one as Brilliant.
Option A - Great Move
Option B - Brilliant Move
Comparing Option A to Option B, we see the new Brilliant move assignment in action. Consider Option B; it fits with the new definition perfectly well.
- Both 18. Qh4+ and 18. Rf3 are the best moves (here, tied for best)
- 18. Rf3 sacrifices material
One Last Caveat
One statement from the November update does peak some interest:
"Moves will also be a bit more generously awarded for newer players, recognizing that some strong sacrifices that may be standard for experienced players are quite an achievement for newer players."
And this begs the question - if someone of higher rating ran a Game Review on this same game, would these moves still be marked as Brilliant? Does this part of the algorithm only affect a particular rating threshold? Regardless, this procedure is not necessarily bad - it might even have the most utility. After all, the moves I would consider "Brilliant" may differ greatly from the opinions of a much stronger player.
