The New Sudoku Players' Forum

[SudokuButabi]

SHORT VERSION:
A) How are newspaper/online-daily puzles rated for difficulty? Are they rated by the same method?
B) Is the only way to make a puzzle harder is to increase the advanced technique requirement count?

Greets. I have expanded my collection of sudoku techniques recently. My past experience has been limited to pencil and paper without solvers or hint software. As i search for a good source of puzzles,it seems to me a lot of puzzless ratings does not match their complexity.
SOURCE 1:
A book from my bookshelf titled "Second Degree Black Belt Sudoku" by Frank Longo. It's introduction states, "this book consists of 300 puzzles of super-tough difficulty."

Code: Select all

 *-----------*
 |.18|..4|...|
 |...|.8.|..3|
 |39.|5..|.4.|
 |---+---+---|
 |.5.|6.7|239|
 |...|...|...|
 |642|9.1|...|
 |---+---+---|
 |.8.|..9|.24|
 |4..|.6.|...|
 |...|4..|78.|
 *-----------*


As you can see, this example requires only a handful of locked candidates, nothing more complex. Hardly as scary as the title first had me believe.

SOURCE 2:
A co-worker gave me an ebook/pdf of 6400 sudoku puzzles with solutions. It has 4 classes of difficulty; easy, medium, hard and evil. There is no author or title page to this so it could have been someone who generated a bunch of puzzles and compiled a pdf.

Code: Select all

 *-----------*
 |68.|35.|...|
 |...|.1.|.2.|
 |..7|.6.|9..|
 |---+---+---|
 |1..|...|...|
 |3.5|.9.|1.2|
 |...|...|..9|
 |---+---+---|
 |..3|.2.|8..|
 |.6.|.7.|...|
 |...|.45|.31|
 *-----------*


This example is labeled evil, yet only requires nothing more than some locked candidates to solve. The only remotely evil aspect is you run out of hidden singles kind of early.

As i've been searching for a good source of puzzles (especially ones i can just paste into Simple Sudoku or Hodoku) for further technique practice and fun, I do find some hard ones too. It just seems like everything i'm finding is either super-simple, or labeled hard yet rather simple, or labeled hard and are CRAZY hard (loops, chanis, franken-fish, etc).
Another way of getting puzzles is thru generation with one of my choice solvers (Simple Sudoku, Hodoku, and Sadman). Hodoku seems to be the mosy powerful/useful to me. I've tried a number of generators and that adventure told a similar story, Diabolical puzzles to one program are medium to another.

I just want to be able to tell how proficient i'm getting. Do people copmlete those crazy hard type puzzles on paper, or are those designed to be done with software aid? Not necessarily a hint system, but things like filters and auto-candidating. One goal I have is to be able to solve daily sudokus or newpaper sudokus friends or co-workers may throw at me.
Is there any sort of standard to rate puzzles? Programs seem to assign points for technique requirements; more chains more points. Is that all that goes into difficulty definition? Which brings up a big followup question i have... Is the only way to make a puzzle harder is to increase the advanced technique requirement count?

Thanks and Peace,
--Adam

[daj95376]

Are you interested in puzzles like these?

Code: Select all

 +-----------------------+
 | . 9 4 | . . 5 | . . . |
 | 5 . . | . 2 . | . . 4 |
 | 8 . 7 | . . 1 | 5 9 . |
 |-------+-------+-------|
 | . . . | . . 3 | . . . |
 | . 7 . | . 4 8 | . . 2 |
 | 3 . 8 | 9 5 2 | . . . |
 |-------+-------+-------|
 | . . 3 | . . . | 2 . . |
 | . . 5 | . . . | . 3 8 |
 | . 8 . | . 3 . | . 1 5 |
 +-----------------------+


Code: Select all

 +-----------------------+
 | 2 7 . | 9 8 4 | 6 . . |
 | 5 8 4 | . . . | . . . |
 | . 6 . | 5 . . | 2 8 . |
 |-------+-------+-------|
 | 4 . 3 | 8 6 . | . 7 2 |
 | 7 . . | 3 4 . | 5 . . |
 | 6 . . | . . 7 | . . . |
 |-------+-------+-------|
 | 8 . 6 | . 9 . | . . . |
 | . . 7 | 6 . . | . 5 . |
 | . . . | 4 . . | . . 6 |
 +-----------------------+


If so, then you might check into the puzzles that I post here.

[SudokuButabi]

@daj
thanks for thr link! the puzzles i peeked at are a great match for my current skill level.

I'm sorry if my main query was clouded with excessive details and examples. I'm really trying to find out if the adjustable point system is the only way a sudoku is graded. And why is my book saying they're 'super tough' puzzles when they require nothing more than pairs and locked candidates?
Is it that many or some of the aadvanced techniques were not known when that book was published? So those puzzles WERE super tough in 2005, more towards the start of the American Sudoku affinity began?

[Pat]

Code: Select all

 . 1 8 | . . 4 | . . .
 . . . | . 8 . | . . 3
 3 9 . | 5 . . | . 4 .
-------+-------+------
 . 5 . | 6 . 7 | 2 3 9
 . . . | . . . | . . .
 6 4 2 | 9 . 1 | . . .
-------+-------+------
 . 8 . | . . 9 | . 2 4
 4 . . | . 6 . | . . .
 . . . | 4 . . | 7 8 .


this example requires only a handful of locked candidates, nothing more complex.

needs a duo in addition to a box-line interaction

i call it a nice 2-move puzzle --

each move alone solves nothing,
you must catch both to solve r7 5

Code: Select all

 6 8 . | 3 5 . | . . .
 . . . | . 1 . | . 2 .
 . . 7 | . 6 . | 9 . .
-------+-------+------
 1 . . | . . . | . . .
 3 . 5 | . 9 . | 1 . 2
 . . . | . . . | . . 9
-------+-------+------
 . . 3 | . 2 . | 8 . .
 . 6 . | . 7 . | . . .
 . . . | . 4 5 | . 3 1


This example is labeled evil, yet only requires nothing more than some locked candidates to solve.

looks like a 3-move puzzle

requires at least 1 duo

[SudokuButabi]

looks like a 3-move puzzle
requires at least 1 duo

i call it a nice 2-move puzzle --
each move alone solves nothing,
you must catch both to solve r7 5

I'm sorry iff you misunderstood Pat, but i'm afraid your reply was rather extraneous. My question was not asking how to SOLVE puzzles, but how they are GRADED. I am well aware of what techniques are required. I could have exhaustively listed every solution set, but that doesn't coincide with the spirit of this post.
Now if you would be willing to share your personal grading vocabulary, i'd be very interested! Or even what your experience has brought you to expect from a typical "evil" puzzle, and a "diabolical" puzzle, etc. If you didn't know the puzzles' source or grade, what would have been your guess? Easy? Hard? Evil? Hope to hear back.

[DonM]

Hi Adam,

To answer your question directly: There is no standard grading system and the relative terms used, easy, medium, hard, tough, super-tough, diabolical, evil and all the rest, only will make any sense for each particular puzzle source ie. a particular newpaper or website. There is no clear reason why a standard was never developed, but it likely has to do with the way Sudoku solving methods were discovered since 2005- in a rather haphazard, disorganized way with people often giving different names to the same method.

However, there are some guidelines regarding difficulty: All the newspapers will have their different names for Easy, Medium, Hard, Very Hard categories and once you get used to what that means for a particular newspaper, those categories will seem pretty consistent. One thing is likely true (at least from my experience): The most difficult newspaper puzzles, no matter what their name, will probably be of similar difficulty: The Diabolical of one newspaper will be fairly comparable to the Evil of another. When it comes to all the various Sudoku books, the terms used are only meaningful within that book- a Very Hard will likely be a little harder than Hard, but that's all. I can also say that the most difficult puzzles in all the Sudoku puzzle books I got in the past, never rose to a level of difficulty greater than those in newspapers, no matter how dramatic the name given to the puzzle.

However, when it comes to more advanced puzzles (well beyond newspaper or book difficulty), we do have some 'difficulty guidelines', though not real standards. A number of us on these forums are long past the newpaper category. The first general category of difficulty for us might be said to be the solving methods used by Simple Sudoku, one of the first sudoku solvers and one of the best designed when it comes to what we call basic methods. So, now when more advanced players solve the more difficult puzzles, they run the puzzle thru Simple Sudoku, having it solve all the hints it can give until it says 'No Hints Available' and then call the result, the SSTS position (Simple Sudoku Technique Set)- it is the puzzle at that point that we will try to solve and that point is a fairly consistent point of difficulty for any puzzle. (But keep in mind that Simple Sudoku, using the basic methods, will likely solve almost all of the most difficult newspaper puzzles.)

http://angusj.com/sudoku/

After finding the SSTS position of a puzzle, we'll now go one step further and use the program, Sudoku Explainer, to analyze it and give what is a sort of difficulty score referred to as ER or RE (for Expainer Rating or Rating [from] Explainer). The most difficult newspaper puzzles will typically score in the RE=6.5-7.0 range. Puzzles beyond the SSTS position, will typically score above and beyond RE=7 or so. It is at the level of RE=7.1 to 8.8 that more advanced players operate. (Puzzles at these levels will require methods beyond basic methods which usually means AIC chains and the like.) Only a handful of solvers presently tackle puzzles at the level of SE=9 and above. All that said, Sudoku Explainer is an old solver and the SE 'difficulty score' has some flaws and inconsistencies, but it's all we really have and in general, it works for us.

Find Sudoku Explainer at:
http://diuf.unifr.ch/people/juillera/Sudoku/Sudoku.html

[aran]

Just to follow up on what Don has said, with all of which I agree.
The question asked concerned personal grading systems, so this answer consists of personal observations :
the first step taken by many experienced solvers is to run a puzzle through Simple Sudoku.
If the puzzle has survived ie has not yet been solved, then it is considered worthy of "some respect".
The vast majority of newspaper puzzles do not survive.
In this sense, newspaper descriptions such as easy, medium, tough, evil, diabolical tend to have no meaning to the experienced solver, since whatever the label nothing remains after pounding by Simple Sudoku
Puzzles which do survive, ie remain unsolved post-SSTS (as explained by Don) and which therefore generally do not feature in newspapers, are then often run through Sudoku Explainer to obtain an SE rating, considered to be a reasonable indication of the degree of diificulty.
At this stage a new scale emerges : easy, roughly speaking, might correspond to an SE rating between 7 and 8, medium from 8 to 8,4, difficult say from 8.5 to 9, and very difficult beyond 9.
Within these categories say from 7.0 to 8.4, the challenge is not so much to solve as to solve elegantly (nice combinations of recognised moves) or innovatively (involving new or unexpected moves) or decisively (few moves).

[DonM]

Aran filled in some of the blanks I left. His last statement is particularly important and on point. It has not only become one of the more 'fun' aspects of solving sudoku puzzles, but is something that IMO can only be found in this particular 'hobby' (the sharing of elegant solutions for a puzzle).