Why are the tactical ratings so high?

Hi, it seems pretty common to have a tactical rating much higher than your average rating (not just here on chess.com)

I was just wondering if there is a reason behind this? Why not just lower the tactical ratings to match blitz and rapid ratings for example?

I'm guessing the initial conditions caused it to happen this way (in other words it wasn't planned) and now it's a hassle to change it, so they don't bother.

The point of having a rating is that you get puzzles around that rating just like your chess rating puts you against other people within your rating range. The cause of the disparity is that solving a puzzle doesn't mean that someone else's score drops which is what you have when you play other people.

Pretty sure each puzzle has a rating. When you solve it, the puzzle's rating goes down relative to your tactics rating. When you fail it, it goes up.  That way they can be sure players get puzzles at their level.

And getting a puzzle wrong doesn't mean that someone else's score goes up.. I don't think that has anything to do with it, I think the error rather comes from puzzles being systematically overrated

Yes that's how new puzzles get their rating. but that still doesn't answer why puzzles are constantly overrated, in relation to other ratings.

Because some puzzles would need sub zero ratings otherwise, which this site doesn't allow.

Think about it: most people's puzzle ratings are about 800 points inflated. So what happens if we deduct 800 from every person and every puzzle?

The 800 rated puzzles are now 0 rated puzzles. What happens to the puzzles below 800 rating?

Because ratings aren't an objective measure like a person's height or speed. Ratings measure you relatively and relative only to others in that group.

That's why your blitz and rapid and bullet ratings can be different, and why blitz on one website is different from blitz on another website.

Makes sense, but also just a note... the ratings work fine at negative values, it's just site administrators set it so that doesn't happen.

But for example there's nothing wrong, mathematically or functionally, with starting people at a zero rating, and letting new players go down to -1000 (or whatever) and have the best GMs playing at 2000.

You could also compress it so that instead of 200 rating points being worth a 75% score, only 20 points is. So now maybe beginners are at zero and the best GMs are at 300.

That answer isn't sufficient, because it begs the question: why don't we make them comparable?

A few months ago, Chess.com raised everyone's bullet ratings by 150 points, because they wanted the ratings to be similar for all categories. So why didn't they also drop everyone's puzzle ratings by 800 points as well?

My first answer addressed that, I said it's because they're lazy.

But I like your answer for it better... although I wonder how much puzzles are inflated by noobs who fail a few then never try again. They probably could lower them all 800 points... and any puzzle that ends up below, let's say 500, could be removed from the rated puzzle pool completely, and they'd have a separate puzzle section for beginners or something... lol, yeah that sounds too messy. Keeping it as it is makes sense.

Allowing sub zero ratings would encourage sandbagging. People would have competitions about trying to get the lowest ratings possible.

As for the second suggestion: that defeats the entire elo/glicko rating system, which makes the ratings too difficult to compare to other sites.

There's a reason why UK stopped using their weird national rating system recently. They wanted their national ratings to be comparable to FIDE ratings. That's always the goal.

That doesn't make any sense.. puzzles being inflated by 800 points is ofcrouse an extreme case, and naturally puzzles rated below 800, aren't in turn inflated by 800.. 

For sure, it's more practical to keep it the way they're doing.

Joined 3 days ago
 

Just a note before we go further with this OP...

what does that have to do with the discussion? if you have a point go ahead, it's a discussion after all

Nah it's actually totally normal.

I'm 2000 blitz rating on this site and I'm routinely solving 2500-2900 rated puzzles correctly.

There are many 1500 blitz players on this site getting 2000-2500 rated puzzles correct more often than not.

A true 2500 rated puzzle should be difficult for titled players to solve, but in reality, it's not.

Thanks for the heads up.

While it's true, your still harshly generalising. the inflation may be proportional to the rating, so it would make sense that puzzles aren't as inflated in ratings such as 800, (which would prevent the paradox)