Yes, very good and easily winnable. It means that there's no reason to suspect that there is a best solution for the aforementioned reasons that it's surprised by it's own evaluation. A true puzzle has an only continuation. I play against stockfish all the time and it decided that on second thought, most of my moves are on par with its in a drawn game. Still evaluating 0.00 In some cases the human move is better than it's choice. For some reason this is unbelievable to people. But they've largely given up and assumed that the computer is always right when it isn't. For a better understanding of this, Leela zero is proof of this. You can also get the book,"The silicon road to improvement".
That's impressive
@Farewell_Witty_Ailen said in #3:
> Holy man...... How much time did you use? I appreciate the work tho, great research, keep it up! It must have took you a lot of time to do it, wow, and it's a school project.
Yes good
@Kevin_M06 said in #9:
> > I like your way of analyzing the puzzles. How did you determine the most frequently played puzzles and their rating level and the most hated puzzles? Where is this data available?
>
> The data is available at: https://database.lichess.org/#puzzles, provided by Lichess themselves. The most played puzzle corresponds to the highest value in the field "Nbplays", the rating level is "rating and the "most hated" puzzle is the one with the lowest "popularity" field.
>
> I recommend clicking the link and see for yourself how popularity is calculated and why saying that the puzzle with the lowest popularity field is the most hated wasn't pertinent but just a way to simplify readability of my blog.
> Best, Kevin.
Hey Kevin,
thanks for your answer. I have recently constructed a few puzzle themes (Vukovic, Triangle Mate, Kill Box Mate, Balestra Mate, Blind Swine Mate, Corner Mate):
https://lichess.org/training/themes
(using Python) out of the above puzzle database, but have so far missed these points (see blog posts on my profile). I might also include some of these metrics when blogging about these mating themes again :-)
Um, actually this is the least played puzzle. https://lichess.org/training/q1lWU
I rarely comment on Lichess articles, but let me get this straight: you analyzed every single puzzle?
@Anonymous1590 said in #16:
> I rarely comment on Lichess articles, but let me get this straight: you analyzed every single puzzle?
No, I analyzed the Lichess Puzzle Database as a whole, performing an analysis by average, collapsing by themes / openings etc.
@Kevin_M06 said in #17:
> > I rarely comment on Lichess articles, but let me get this straight: you analyzed every single puzzle?
>
> No, I analyzed the Lichess Puzzle Database as a whole, performing an analysis by average, collapsing by themes / openings etc.
What do you mean by ". . . performing an analysis by average, collapsing by themes / openings etc."?
@Anonymous1590 said in #18:
> > > I rarely comment on Lichess articles, but let me get this straight: you analyzed every single puzzle?
> >
> > No, I analyzed the Lichess Puzzle Database as a whole, performing an analysis by average, collapsing by themes / openings etc.
>
> What do you mean by ". . . performing an analysis by average, collapsing by themes / openings etc."?
By “performing an analysis by average, collapsing by themes/openings,” I meant that I wasn’t analyzing each puzzle individually. Instead, I grouped (or “collapsed”) the puzzles by certain categories, like their theme or the opening they come from and then calculated averages for those groups.
For instance, I grouped the database by themes to count the number of puzzles that have the same themes. I can then do a descending sort to view the 8 most frequent themes. I can then compute their average rating, number of times played, etc. and compare the results with the rating and number of times played of all the puzzles (and not just the puzzles that have the most frequent themes). This gives an overall statistical picture of the puzzle database without manually analyzing each position.
@Kevin_M06 said in #19:
> > > > I rarely comment on Lichess articles, but let me get this straight: you analyzed every single puzzle?
> > >
> > > No, I analyzed the Lichess Puzzle Database as a whole, performing an analysis by average, collapsing by themes / openings etc.
> >
> > What do you mean by ". . . performing an analysis by average, collapsing by themes / openings etc."?
>
> By “performing an analysis by average, collapsing by themes/openings,” I meant that I wasn’t analyzing each puzzle individually. Instead, I grouped (or “collapsed”) the puzzles by certain categories, like their theme or the opening they come from and then calculated averages for those groups.
>
> For instance, I grouped the database by themes to count the number of puzzles that have the same themes. I can then do a descending sort to view the 8 most frequent themes. I can then compute their average rating, number of times played, etc. and compare the results with the rating and number of times played of all the puzzles (and not just the puzzles that have the most frequent themes). This gives an overall statistical picture of the puzzle database without manually analyzing each position.
That all sounds way over my head lol. I sort-of understand, though. How long did all of this take you? And how long did you have to complete the school project? By the way, although your blog is somewhat hard to understand, I can tell that you put in a lot of effort. Amazing job!