Psychology(?) Question

[Aitch]

OK, so I get the i on weekdays and on the puzzles page there are three Suduko puzzles which I always have a go at - look I'm not addicted, I can give them up anytime I want to!

So, the three puzzles are rated Simple, Intermediate and Advanced. The Simple and Advanced I nearly always get through OK - say 95% of the time. But, if I do mess one up, the odds are it will be the Intermediate one - say about 95% again.

Why should this be so? I can think of two reasons:

1. The ratings are inaccurate.

2. For the Basic one I am more relaxed (it's supposed to be easy, right) and work through it unstressed. The Advanced is supposed to be hard, so my brain kicks up a gear and I solve it without too much bother. But for the Intermediate, it's neither one nor the other, so my brain isn't working optimally. Plus, I suppose, there is a learned aspect, I have messed these up more than the others, so it becomes to an extent a self-fulfilling prophecy type thing.

Any suggestions?

Of course, if this was a poll, there would be a Planet X option, but I can't think for the life of me what that would be.

 

[Earthborn]

My guess: 1.

Sudokus can quite easily be solved algorithmically (I even have an Excel sheet somewhere that can solve them), so the rating probably reflects only the number of steps a computer program would need to take to solve it. That makes rating them easy. But humans don't think like that: when playing games they learn to recognise complex patterns, forming an "intuition" on how to solve them. The Sudokus rated advanced may very well contain the most recognisable patterns.

 

[Aitch]

Possibly; but I tend, in my rather unimaginative way, to use what I would call brute-force algorythmic methods. First I do sets of columns of 'boxes' of 9 squares, then rows of such, to pick off the low-hanging fruit, as it were. Then either individual rows/columns or individual boxes of 9. And then iterate. I suppose some intuition comes into it, but not much.

This is possibly a hangover from my years in computer programming. Mind you, I worked with a guy who could probably cobble something together in APL quite easily (I suppose Sudoku could (just) be considered a very complex version of the Eight queens puzzle^WP), he did the same for solving the Rubik's Cube.

 

[Dancing David]

Lots and lots of reasons, it would take a while to parse them out.
1. Confirmation bias.
2. Attention
2.1 Higher on the Advanced
3. Focus
3.1 Ditto
4. Mis-rating
5. Method of solving but I doubt that.

 

[Aitch]

Lots and lots of reasons, it would take a while to parse them out.
1. Confirmation bias.

In what way?

2. Attention
2.1 Higher on the Advanced
3. Focus
3.1 Ditto

Don't see a difference here. But it seems to be my second option above.

4. Mis-rating

Seems more less likely, going by Earthborn's post and the way I do the solving.

5. Method of solving but I doubt that.

Well, I use the same method on all three, so I doubt it too.

 

[Ziggurat]

My guess: 1.

That would make sense if it happened once or twice. But if it's really a general pattern, that explanation seems inadequate. While the computational difficulty rating isn't the same as a truly accurate "human" difficulty rating, the correlation should still be pretty good.

I suspect a combination of confirmation bias (failing on the intermediate will stand out more than failing on the advanced, since it's not expected), plus more effort/focus paid to the advanced problems. But misratings may help to create "noise" which can contribute to the confirmation bias.

 

[Joey McGee]

I would guess that they use different algorithms or rules to set the difficulty level, and the way your brain does math is tripped up. Your pattern-seeking or pattern-responding methods that were learned have their own rules. I would compare this to an optical illusion.

Try getting a batch of only intermediate puzzles and doing them. It could also be a subtle psychological thing about risk or something, I suppose. If it's because of the algorithms they use, your brain might wire up to solve those puzzles easily and then you might make up for it. That's how I have solved those kinds of problems in certain training before.

 

[Dikken]

I found this hints page - I can always sove the 9x9s but have problems with the "mega" versions.

Good luck: http://angusj.com/sudoku/hints.php

 

[Maurice Ledifficile]

A guess

Just guessing (from doing quite a few of them myself).

It seems that the label "Difficult" simply means that you have fewer revealed numbers to start with. They might leave over 25 numbers on an "Easy" puzzle.

This will make the harder ones harder on average, but you still might run into trouble with an easy one or breeze through a difficult one.

I don't know what is really happening, but it seems reasonable to assume that confirmation bias has a role to play here (you get a feeling that there is a pattern which is out of the ordinary and can easily find corroborating evidence).