Difficulty Levels

[Pritch]

I have played 16 games in each of the difficulties: Medium, Tricky, Fiendish and Diabolical and have noted with interest the average times that I achieved (all games were with auto pencilmarks on - for lazyness :P, but I never took any hints or complete guesses)

Medium: 06:28
Tricky: 10:59
Fiendish: 10:29
Diabolical: 07:59
(Beginner & Medium are both under 01:30 averages!)

I know that the game difficulties are based on the 'techniques' required to solve the puzzles, but it seems strange that there is such a difference with the Diabolical level times...Fiendish & Diabolical could be counted as the same time, after a bigger average...

Has anyone else got relatively better times for Diabolical than the others (the absolute times don't matter), or is there simply one technique that I can do better than others?

Do the difficulty settings need to be reviewed?

[Howard]

Do the difficulty settings need to be reviewed?

There are three broad categories of techniques, which are :

Easy ( single candidate, single position )
Medium ( Block-Line intersections)
Hard (Subsets, XWings, Forcing Chains etc)

The costs for Hard techniques are quite a bit higher than for medium techniques.

So, what I think is the case is that the Tricky puzzles tend to require several Block-Line intersections and the like, and the Fiendish/Diabolical puzzles may have just one or two of the harder (still) techniques, which if you spot them, can make it quite quick.

The medium techniques though, while somewhat easier to understand, can take quite some time to find because you have to scan quite a lot more.

In a future version, we'll be adding some extra features which will make it easier to spot some of the medium techniques, which I think will bring some of the medium puzzles further into line.

Once we've got those - and of course built up a large database of submitted times for puzzles from the online entries - I can tweak values for techniques a little further.

Further thoughts most gratefully received, of course!

Howard.

[Pritch]

Ok, I can agree with what your explanation - although it is only recently that I read the Astraware Sodoku site and found out that the ways I had worked out how to solve puzzles actually had names and associated 'difficulties'!

Istill have an issue with the rating of the Diabolical puzzles - as, by using auto pencilmarks, you can blast through the puzzle until there is only a relatively small number of unknowns, requiring a single hard 'technique' to solve the whole puzzle.
The difficulty is a question of interpretation - normally I use 'forcing chains / XWings' to solve the diabolical puzzles - simply a question of thinking and counting and remembering - so not particularly difficult.

The tricky and fiendish puzzle are more difficult for the very reason that there are more unknowns when you have to find a 'way into the puzzle'.

As such, needing more medium techniques is more difficult that needing one hard technique - especially when there are already fewer unknowns to consider...as is reflected in the times that I have achieved.

Incidentally, I believe that before I used the auto pencilmarks, I achieved the same relative times - as I used the same basic procedure as now...

Am I alone in solving the puzzles in this way and therefore finding this time vs. difficulty anomaly?

I guess as everyone thinks differently, rating the puzzles will always be 'difficult' :P

[icebox]

Difficulty ratings are like wine tastings. Some work and some don't. AW's Sudoku routinely ranks Michael Mepham's most difficult newspaper puzzles a step or two below what he does. And just yesterday it ranked a so-called 5-star difficulty newspaper puzzle as Medium. And then I ran into the phenomenon noted in the initial posting. It took 45 minutes to solve.

Mark