Your network blocks the Lichess assets!

lichess.org
Donate

New Chess Tool - Novelty Hunter!

Nice. I assume you also know about https://chesspath.pro/ which is a similar project.

"Almost all the code was generated by my good friend called Claude, with some revision by me."
Yes, I see since beginning of this year an avalanche of new tools created by Claude. In fact last week I created myself the lichess broadcast downloader. That is a tool which allows you in one go to download all games of a player broadcasted here on lichess. This avoids the very time-consuming manual downloads and filtering per tournament. If interested then just send me a message.

Nice. I assume you also know about https://chesspath.pro/ which is a similar project. "Almost all the code was generated by my good friend called Claude, with some revision by me." Yes, I see since beginning of this year an avalanche of new tools created by Claude. In fact last week I created myself the lichess broadcast downloader. That is a tool which allows you in one go to download all games of a player broadcasted here on lichess. This avoids the very time-consuming manual downloads and filtering per tournament. If interested then just send me a message.

I appreciate the "ad" warning on your blog post, but I'm actually not seeing any affiliate links or similar anywhere!

I appreciate the "ad" warning on your blog post, but I'm actually not seeing any affiliate links or similar anywhere!

That toggle is a little bit confusing. If you promote your tool, is it "an ad"? What if it's totally free? What if it's not? What if it's free for now? What if it's not yours? What if there is no product promoted, but you're still paid to write the content?

Maybe someone who understands that toggle could write a post about it

That toggle is a little bit confusing. If you promote your tool, is it "an ad"? What if it's totally free? What if it's not? What if it's free for now? What if it's not yours? What if there is no product promoted, but you're still paid to write the content? Maybe someone who understands that toggle could write a post about it

The warning states:

Contains sponsored content, affiliate links or commercial advertisement

I'm pretty sure a dictionary can easily explain what these mean (except perhaps "affiliate link" but most people understand what that means too).

The warning states: > Contains sponsored content, affiliate links or commercial advertisement I'm pretty sure a dictionary can easily explain what these mean (except perhaps "affiliate link" but most people understand what that means too).

@mvhk said ^

Nice. I assume you also know about https://chesspath.pro/ which is a similar project.

"Almost all the code was generated by my good friend called Claude, with some revision by me."
Yes, I see since beginning of this year an avalanche of new tools created by Claude. In fact last week I created myself the lichess broadcast downloader. That is a tool which allows you in one go to download all games of a player broadcasted here on lichess. This avoids the very time-consuming manual downloads and filtering per tournament. If interested then just send me a message.

huh? I wrote ChessPath.
but yeah it's good for finding novelties.

@mvhk said [^](/forum/redirect/post/iCuyfTBf) > Nice. I assume you also know about https://chesspath.pro/ which is a similar project. > > "Almost all the code was generated by my good friend called Claude, with some revision by me." > Yes, I see since beginning of this year an avalanche of new tools created by Claude. In fact last week I created myself the lichess broadcast downloader. That is a tool which allows you in one go to download all games of a player broadcasted here on lichess. This avoids the very time-consuming manual downloads and filtering per tournament. If interested then just send me a message. huh? I wrote ChessPath. but yeah it's good for finding novelties.

@Zorbos from your docs (interestScore = 40% × earlyNovScore + 40% × efficiencyScore + 20% × rarityScore) I gather that the novelty score is:

  • 40% early move score
  • 40% efficiency score (based on evaluation of future moves and game result)
  • 20% rarity score (based on how many games there are in the db)

Isn't it strange that just a fifth of the score is actually related to novelty while 40% is just related to the ply?

I've tried to get the data from the Explorer items and I translated it to this:

  • early move score: Math.min(1,1.5-ply/20)
  • efficiency score: 2*(winPerc * winPerc) -1 where winPerc is (wins+draws/2)/total from the explorer item
  • rarity score: Math.min(1,1/followupGames) + 1 - Math.max(1,followupGames/mostPopularGames/rarityThreshold) where rarityThreshold is 0.1 and followupGames is total games per move item and mostPopularGames is the total games per most popular move

This removes the need for stockfish evaluation - although I have that too - and relies solely on actual game wins and losses. But I am still unsure about the 40/40/20 split.

@Zorbos from your docs (_interestScore = 40% × earlyNovScore + 40% × efficiencyScore + 20% × rarityScore_) I gather that the novelty score is: - 40% early move score - 40% efficiency score (based on evaluation of future moves and game result) - 20% rarity score (based on how many games there are in the db) Isn't it strange that just a fifth of the score is actually related to novelty while 40% is just related to the ply? I've tried to get the data from the Explorer items and I translated it to this: - early move score: Math.min(1,1.5-ply/20) - efficiency score: 2*(winPerc * winPerc) -1 where winPerc is (wins+draws/2)/total from the explorer item - rarity score: Math.min(1,1/followupGames) + 1 - Math.max(1,followupGames/mostPopularGames/rarityThreshold) where rarityThreshold is 0.1 and followupGames is total games per move item and mostPopularGames is the total games per most popular move This removes the need for stockfish evaluation - although I have that too - and relies solely on actual game wins and losses. But I am still unsure about the 40/40/20 split.