What site would I get a good grading system or
a difficulty rating for a sudoku when imput the
#s myself:?:
Whats the best site for rating
4 posts
• Page 1 of 1
Whats the best site for rating
by thebull » Sat Jun 17, 2006 11:28 pm
- thebull
- Posts: 1
- Joined: 27 May 2006
by tso » Sun Jun 18, 2006 5:37 am
There is no good answer to this question as sudoku difficulty is subjective and non-linear. How could we compare these four hypothetical sudokus?
Puzzle 1) This puzzle can be solved using nothing but singles -- but during most of the solving process, only one single is available. (aka "high-stepper")
Puzzle 2) This puzzle can be solved using singles and a dozen locked candidates. Becase each locked candidate leads to placement of a number, most players will not require pencil marks.
Puzzle 3) This puzzle can be solved using singles and three consecutive locked candidates. Because one must resolve all three locked candidates before making a placement, most solvers would require pencil marks.
Puzzle 4) This puzzle can be solved using just singles until there are 15 cells left, at which point it can be solved with either BUG, and XY-wing or a short XY-chain. Up until this point there are many singles available throughout only requires pencil marks for this one advanced deduction, and then only for a few cells and only 2 or 3 marks per cell. (Pappocom software would rate this puzzle "invalid" though most readers here would find it nearly trivial.)
Can we simply define difficulty linearly -- the one that takes the longest to solve by a particular method is therefore 'harder'? Is it harder to walk 100 yards than 50 yards? I don't think so -- it just takes longer.
There is consesus that forcing chains are a much more advanced tactic than naked quadruples or hidden triples -- but it depends on context and the solvers choice of methods which of will be harder to spot.
A solver who routinely filled in all the pencil marks before starting would likely rate some puzzles harder and other easier than another solver who only used pencil marks as a last resort. A third solver who uses a computer to assist would have a different opinion. If puzzles had an intrisic hardness, all would rate them the same -- a 100 pounds weighs more than 50 no matter if you use a forklift or not.
That being said, try Pauls Page's for a simple five level rating with an easy import/export cut and paste feature.
I don't know any other websites that will rate a puzzle, though there are 100's that will solve it for you.
Puzzles pasted into library files will be rated by Pappocom software, but it's very clunky to use and will rate many puzzles invalid or "unfair".
Sadman's software will give puzzles pasted into it a rating on an 8 level scale.
Suexrate.exe and Suexrat9 will rate your puzzles on a numerical scale. An easy puzzle would be around 50, newspaper level would probably be under 80, but ratings in the 1000s are possible. The ratings beyond 200 or so may have more meaning to computer solvers than to humans -- your visceral experience solving a 300 or an 800 is unlikely to be subjectively different and the rating at higher levels will probably not predict how long it will take you to solve it. Though a 4 foot wall is harder to jump over than a 3 foot wall, a 40 foot wall is just has hard to leap as a 100 foot wall.
Puzzle 1) This puzzle can be solved using nothing but singles -- but during most of the solving process, only one single is available. (aka "high-stepper")
Puzzle 2) This puzzle can be solved using singles and a dozen locked candidates. Becase each locked candidate leads to placement of a number, most players will not require pencil marks.
Puzzle 3) This puzzle can be solved using singles and three consecutive locked candidates. Because one must resolve all three locked candidates before making a placement, most solvers would require pencil marks.
Puzzle 4) This puzzle can be solved using just singles until there are 15 cells left, at which point it can be solved with either BUG, and XY-wing or a short XY-chain. Up until this point there are many singles available throughout only requires pencil marks for this one advanced deduction, and then only for a few cells and only 2 or 3 marks per cell. (Pappocom software would rate this puzzle "invalid" though most readers here would find it nearly trivial.)
Can we simply define difficulty linearly -- the one that takes the longest to solve by a particular method is therefore 'harder'? Is it harder to walk 100 yards than 50 yards? I don't think so -- it just takes longer.
There is consesus that forcing chains are a much more advanced tactic than naked quadruples or hidden triples -- but it depends on context and the solvers choice of methods which of will be harder to spot.
A solver who routinely filled in all the pencil marks before starting would likely rate some puzzles harder and other easier than another solver who only used pencil marks as a last resort. A third solver who uses a computer to assist would have a different opinion. If puzzles had an intrisic hardness, all would rate them the same -- a 100 pounds weighs more than 50 no matter if you use a forklift or not.
That being said, try Pauls Page's for a simple five level rating with an easy import/export cut and paste feature.
I don't know any other websites that will rate a puzzle, though there are 100's that will solve it for you.
Puzzles pasted into library files will be rated by Pappocom software, but it's very clunky to use and will rate many puzzles invalid or "unfair".
Sadman's software will give puzzles pasted into it a rating on an 8 level scale.
Suexrate.exe and Suexrat9 will rate your puzzles on a numerical scale. An easy puzzle would be around 50, newspaper level would probably be under 80, but ratings in the 1000s are possible. The ratings beyond 200 or so may have more meaning to computer solvers than to humans -- your visceral experience solving a 300 or an 800 is unlikely to be subjectively different and the rating at higher levels will probably not predict how long it will take you to solve it. Though a 4 foot wall is harder to jump over than a 3 foot wall, a 40 foot wall is just has hard to leap as a 100 foot wall.
- tso
- Posts: 798
- Joined: 22 June 2005
by tso » Sun Jun 18, 2006 6:06 am
Ok -- maybe I'm being too abstract. Here's a example of the difficulty of objective ratings:
I'd rate this as easy -- just singles and plenty of them, then a BUG (or an obvious 5-cell forcing chain if you prefer) at the end. If you're not familiar with BUG or forcing chains, you might rated it much harder. Pauls Pages calls it "Outlaw". Pappocom's program calls it "invalid" though nearly everyone will find it much easier than any Pappocom Very Hard or even Hard.
- Code: Select all
+-------+-------+-------+
| . 1 5 | 2 4 . | . . . |
| . 2 . | . 1 8 | 6 4 5 |
| . 6 . | . . . | . . 8 |
+-------+-------+-------+
| . 3 . | . . . | . . 6 |
| 6 9 . | . . . | . 7 3 |
| 5 . . | . . . | . 8 . |
+-------+-------+-------+
| 4 . . | . . . | . 5 . |
| 2 8 9 | 7 3 . | . 6 . |
| . . . | . 2 1 | 8 9 . |
+-------+-------+-------+
I'd rate this as easy -- just singles and plenty of them, then a BUG (or an obvious 5-cell forcing chain if you prefer) at the end. If you're not familiar with BUG or forcing chains, you might rated it much harder. Pauls Pages calls it "Outlaw". Pappocom's program calls it "invalid" though nearly everyone will find it much easier than any Pappocom Very Hard or even Hard.
- tso
- Posts: 798
- Joined: 22 June 2005
4 posts
• Page 1 of 1
