lichess.org
Donate

Tactics Trainer on lichess app

For some reason, the tactics trainer on the android app keeps giving me puzzles which are several hundred points below my tactics rating, hence keeping my rating continuously provisional. Would love to see a fix for this! Thanks in advance
Hi Aladeen,

I'm really glad you posted this! I have had the identical experience and it is absolutely RUINING the tactics training for me. I was going to start my own thread about this until I saw your post.

I'd estimate that 95% of the problems I'm being given are rated at multiple hundreds of points lower than my rating. The remaining 5% are problems near my level. As a user, I would expect the total opposite: being primarily given problems at my level, and occasionally given problems below or above my level. This current situation sets me/up only to fail! Being bombarded with problems that offer very little reward and carry extreme punishment rating wise. It is akin to an 1800 level player only being allowed to play 1400 level players, while being denied 1700 and above players.

Specifically, if I'm around 1900, I will be assaulted by a flood of 1500 level problems only.

Makes no sense! Let me play problems at my level!

Anecdotely I have earned and sustained a 1900 tactics rating several times over many many hours, only to watch it be evaporated in literally a few minutes by a barrage of lower rated problems.

I don't have a problem with the rating system concept in general, however, I have a significant complaint with the problem-selection algorithm. Its current premise and implementation (or at least my individual experience) is deeply flawed and degrades the otherwise enriching tactics training into a psychologically devastating mockery.

Make Lichess Tactics Great Again!
" Being bombarded with problems that offer very little reward and carry extreme punishment rating wise. "

If these problems are too easy for you, why are you failing them?
hahaha great reply BaronVonChickenpants.!!
I was thinking exactly the same.
Your respective criticism, while generally sensible, is completely unrelated to the issue described and is thus an irrelevant distraction. This is a bug, regardless of whether or not a particular tactic was solved correctly.

The scenario concerns being offered only problems that are substantially lower rated than your current rating. This intrinsically dictates a small, trivial reward for a correct answer (+2, +3, etc) and a large punishment for an incorrect answer (-14, -20). Thus, in such an arrangement, the player will need to get problems correct at a rate of perhaps 10 to 1 in order to just break even.

It may be the case that the system is designed to deliberately operate in this pattern, which I would be interested in hearing more about.

I can report that to test this, I recently used another older account and thankfully did not experience this beyond hellish nightmare. I was offered a healthy humane dose of both higher and lower rated problems and felt completely comfortable with the evolution of the point delta. So, perhaps this bug is isolated to this particular account, similar to the "tactics rating stuck" reports.

For now, I'll train signed out, until fairness is restored!

If someone gives you a "flash" car would you complain the color was wrong.
Most tactics that win games are simply easy tactics at least on this site,if I'm wrong please post the evidence to teach us all.
The tactics rating is very often different then performance rating people tend to think speed and rating numbers mean something they do not. (at lest not for speed games and tactics of speed games. I define speed game as any game less than 20 minutes.)
Some one said "For now, I'll train signed out, until fairness is restored!" that should fix it.
This bug is now impacting both of my accounts; I am being given ONLY problems that are way lower than my rating. For example, I was just shy of an 1800 rating, wanting to pass it, and am given ONLY problems at the 1200 level. If I get it correct, I get +2; if I get it wrong, I get -30.

This is a profound bug in the "pairing" algorithm. It's the same as an 1800 level player in a blitz game being Always paired against a 1200 level player and NEVER being paired against an 1800 level player.

Many others have vocalized this exact same bug in these forums, and have expressed it in terms of "pairing." Clearly, this is a bug that needs fixing.
Yeah I don't know if it's a bug or if that's how it's intentionally set up. I mostly get puzzles in the 2000-2100 range even though my puzzle rating is over 2600. I thought this was maybe because there aren't a lot of high rated puzzles but it seems to happen at all ratings.

I understand what you mean when you say high risk for little reward. I have to be very careful not to get these puzzles wrong otherwise I'll lose a ton of rating points that will take a very long time to recover. Not that these ratings really mean anything but it's still a little annoying when you only get 2-3 points for a correct answer and then lose 20 points for an incorrect answer.

The overall point of solving puzzles is to learn new tactics and patterns and to practice your calculation skills. These lower rated puzzles are still good for calculation but I still almost always get them right so I feel like I'm not really learning as much as I could be.
I've had just the opposite experience. I'm at ~2100 and the difficulty of all puzzles except for the "puzzle of the day" have gone way up. Puzzling has become an exercise that requires an annoying amount of time, focus, and calculation to find deep tactics. I enjoy the challenge, when I have the time, but part of me misses being able to pull out my phone and quickly solve puzzles as a distraction in class or in a meeting. I suppose the higher rated puzzles are slowly improving my calculation abilities, but the repetition of spotting tactics that are apparent on the board in the lower rated puzzles also serves a useful function in training. I don't see those puzzles anymore and sometimes I'll log out just for this purpose.

This topic has been archived and can no longer be replied to.