Did Tactics Trainer just get harder?

Sort:
pstubley

Is it just me or did Tactics Trainer suddenly get much harder?  Almost all the puzzles recently have had very few attempts, and my rating has dropped 250 points in the last couple of days.  I just crashed and burned on one that has 15 moves!  That's not a tactic, that's a game ;)

I'm guessing that a bunch of new problems were just added and the levels have not really self-adjusted appropriately yet.  And there do seem to be errors.  I found at least one where there are 3 mates at the same move (even according to the analysis) -- I, of course, picked the wrong one.  Many of the long ones (the 15-move one is the longest, but there have been quite a few that are 12 moves, or 8) do not really seem to be forcing:  I certainly agree that Rykba may prefer one line to another for perfectly valid reasons, but they are seeming a little subtle for me.

I really enjoy doing tactics and learning from them, but I have to confess that the last day or so has been rather disheartening.  Maybe I have just done too many so only the new ones are left!

DeepGreene

There are a lot of new ones added a couple of days ago, and yes, it will take a while for them to "find themselves" in terms of an appropriate rating level.

For the most part, they should be fine, but if you come across any stinkers, please report them! Thank you!

DeepGreene
[COMMENT DELETED]
antnut

I disagree that "most are fine." Most of the new problems need significant editorial work. There are far too many of the "choose between +17 and a five-sacrifice mate in 11" variety, and others where the preferred line is not superior at all to alternatives. The new problems are not up to the usual tactics trainer standard.

Here are a few examples I've noticed in the last few minutes.

0437467 -- two mates, as stated in its own analysis and source; only one is accepted

0380163 -- you have to choose between +15.7 and mate in 11 at move 39

0357529  -- analysis and source identifies 3 different mating lines

I found these in about 5 minutes of play. The new problems are all like this. I hate them.

Metastable

Whew! Glad to know it's not just the scotch-damaged brain cells finally dying off on me. I've been getting killed the last day or two as well in TT.

antnut

0370755 -- multiple mating lines in analysis and source

0283333 -- must choose between mate in 9, +10.5, and + 11.6 at move 16

0220549 -- multiple mating lines in analysis and source

I've also lost 400 points on my rating in 2 days, very irritating!! And who has time to study problems that have an average time of 15 minutes plus? I come here for a quick tactics break at work, not to get out a pad and paper and laboriously work out 35 variations 11 moves ahead.

antnut

From analysis and source: 0397593

17. Ke3 Rh3+

({12:-Mat08} 17. ... Rh3+ 18. f3 Rh2 19. Ne2 Bxd2+ 20. Qxd2 Qc5 21. Qe1 Rxe2+ 22. Qxe2 Nb3+ 23. Ke4 $19)
({12:-17.88} 17. ... Rxd1 18. Rxd1 Qa6 19. Nb5 Qf6 20. Nxd4 exd4+ 21. Ke2 Re8+ 22. Be3 dxe3 23. fxe3 Qf4 24. e4 Qxg4+ 25. Kf2 Qxd1 $19)
({12:-14.49} 17. ... f5 18. Be1 Rxe1+ 19. Qxe1 Nxc2+ 20. Ke2 Nxe1 21. Rxe1 Bxc3 22. bxc3 Qxc3 23. Rd1 e4 24. Kf1 exd3 25. gxf5 Qc4 26. f3 $19)

This is at MOVE 2 OF 9 for heaven's sake!

ilmago

CryCryCry

Jebcc

you weep cm ilmago?  wipe your tears toughen up

antnut

I think ilmago might be teasing me Surprised

waffllemaster

The very long mates no one sees to the end, you play those kinds of moves intuitively (until you get close enough to see all the way to mate).

If the system faults you for clearly winning moves, then I don't understand why anyone uses it... because if it does there are better TTs out there that don't...

ilmago

no, not teasing you at all, nsimington.

If anywhere along the main line, the second best move is also clearly winning, giving you an advantage of +3.00 or more, the tactics is considered to be ambiguous and thus a bad one.

Bad ones can be reported by clicking on "report problem".

chess.com seems to be hoping that this will allow to get rid of most of the bad ones within a reasonable time.

 

However, I am afraid an automated fix will be inevitable.

JeffMGreg

I'm so glad I'm not the only one.  I dropped about 300 points with these problems.  I've been reporting problems but there are so many of them. Really hope these get fixed soon.

pstubley

I have every expectation that things will return to "normal":  coming up with TT problems is not trivial, so some growing pains are not surprising.  And ultimately (at least for me) it is not about the points, but about what I learn.

I have not had the confidence to report problems (I was proud of myself for leaving a comment yesterday about the multiple mates ;) ), but I will start doing that.

And to the staff of chess.com, despite my whinging about the problems getting harder in this new batch, I really do enjoy TT so thanks for all the efforts in finding problems for us to amuse ourselves with.

Jpatrick

A whole bunch of problems have been "autoapproved" recently, and from my experience, they are ALL ambiguous.  The last ten in a row have all had multiple wins.

My experience is that the problem set is completely poisoned.  I resent paying for this.

Jpatrick

You know, a lot of people have put a lot of time into cleaning up the problem set and now to have this load of crap dumped into it is really discouraging.  How can you do this?

baddogno

Thinking of canceling my membership over this.  I've done over 14,000 problems in the last year, but this new batch is awful.  Just get rid of them all.

Metastable

Maybe it would be more appropriate for new problems to be trickled in at a slower rate into the main TT problem set pool, rather than dumping a whole boatload in at once and disturbing the equilibrium of the whole system like that.

antnut

0237927

33. fxe6

({12:Mat08} 33. fxe6 Qd1 34. Qe4 Qh1+ 35. Qxh1 Bxe5 36. Qe4 Ke7 37. Qxe5 dxc3 38. Qxc7+ Kxe6 39. Re5+ Kf6 40. Qe7# $18)
({12:14.41} 33. Rxg7 Kxg7 34. fxe6 Rh8 35. Qxc7+ Kh6 36. Qc6 Kg7 37. e7 Rh6 38. Qf3 Qh1+ 39. Qxh1 Kf7 40. Bc4+ Kxe7 41. Qb7+ Kd8 42. Qa8+ Kc7 43. Qxa7+ Kd8 44. cxd4 $18)
({12:6.51} 33. Qf3 Qe3 34. Qf1 Bxe5 35. fxe6+ Bf4 36. Rf5+ Ke7 37. Rf7+ Kd6 38. Rxf4 Qxe6+ 39. Bf5 Qe5 40. Re4 Qd5 41. Qf4+ Kc5 42. Rxd4 $18)

Please fix this ASAP. TT is basically unusable now.

ilmago
nsimington wrote:

Please fix this ASAP. TT is basically unusable now.


I fully agree. This will need to be done in an automated way.

Auto-remove the autoinjected problems.

Fix the algorithm so that it does not allow such terribly ambiguous problems to get in.

And use the fixed algorithm to scan tactics trainer and weed out the bad ambiguous problems auto-injected on earlier occasions. Many of them are still in tactics trainer even after years of human weeding out.