The New Sudoku Players' Forum

[PaulIQ164]

Not trying to show off (well, maybe a little,) but am I the only one who doesn't really use pencilmarks when solving Su Dokus? I do the Times' fiendish ones (and did the super-fiendishes) hardly ever writing very much aside from the big numbers (occasional notes around the perimeter, but often not even those). I find it more satisfying, and all those little numbers tend to confuse me more than they help anyway.

[lunababy_moonchild]

Well, if you do have an IQ of 164 I can't imagine you having problems with anything

Luna

[PaulIQ164]

Wasps I have problems with. I hate the little blighters.

[Guest]

...and perhaps modesty.

[PaulIQ164]

Maybe that too.

[Karyobin]

I don't have to use pencil marks for any sudoku other than Very Hard and a few Hards. Like you, I find Fiendish and anything marked Diabolical in the Telegraph fairly straightforward to do mentally, though some take a little more patience than others.

I'd certainly be impressed if you can solve Very Hards without pencil marks. To be fair, I'd be impressed if anyone short of a savant could do that.

I told the kids at school that IQ stood for 'Idiot Quotient'. They're not so impressed by it anymore.

[tso]

My experience is similar, though I don't think it's because I'm so brilliant. I believe it is because most of the puzzles in newspapers aren't hard enough. I've heard that although Pappocom's software will create a "Very Hard", newspapers that carry his puzzles max out at "Hard", labeling them as "Fiendish". Some tactics which would usually require most people to use pencil-marks to implement, including, but not limited to, forcing chains, swordfish, coloring, etc, are not needed in the vast majority of these puzzles. Other tactics may very well be made more confusing by putting in pencil marks before needed. I infer from various Pappocom statements that he solves puzzles the way you do -- he uses few or no pencil marks and doesn't like puzzles that cannot be solved this way and/or feels that the solving experience is ruined by doing so. I agree with this up to a point -- the point at which an *actually* hard puzzle *does* require more advanced tactics, and I'm forced to put as much information onto paper as I can fit in order to discover some convoluted logical deduction that may not even have a name.

Here's one that gave me trouble. It doesn't require X-wings, Swordfish or anything worse, but I couldn't make much progress without a lot of pencil marks. If you can solve this without doing the same, post your technique here.

This forum may be due for some discussion on advanced 'bareback' solving tactics -- that use no (or few) pencil marks.

Code: Select all

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


By the way:
The solver at (http://www.scanraid.com/sudoku.htm) could not solve it.
The solver at (http://www.paulspages.co.uk/sudoku/) rates it as "Outlandish".
The solver at (http://act365.com/sudoku/) verifies that it does not require X-wings, Swordfish, Nishio or "guess" and has a unique solution.

(If it turns out to be easy for any of you to solve, then I may have not made myself clear. What I *really* meant was: "I agree with you completely. Here's one that's *supposed* to be hard that I solved in 5 minutes -- blindfolded.")

[simes]

...and for completeness, my solver can also solve it, using simple block/col/row interactions and a couple of pairs.

[PaulIQ164]

Yes, that one's evil, isn't it. In the end, being sat in front of a computer was just too tempting, so I cheated a bit. The trick is with row 7, and the values 3, 4, 5 and 8. They all appear in column 2, and in box nine, so that the 3, 4, 5 and 8 in row 7 must be in r7c1, r7c3, r7c4, and r7c6 in some order. Now there's only one place left for the 9 in box 8 to go, then only one place left for the 4 in the same box. That gives you the 4 in box 2, which gives you the 1 in r2c8. Then a little bit of logic gets you a 4 in r3c8, and it starts to fall into place.

I guess it would be possible to make that first leap on paper - once the computer told me it it seemed so obvious.

Anyway, to clarify, I was never trying to claim that I don't use pencilmarks because I am some kind of SUDOKU GOD!, merely that I don't like using them if I don't have to. If the puzzles the program generates go harder than ones in the papers, I guess I'd probably have to use them then, but I've not really tried.

[petef]

My solver rates it as "easy". It needs only the first two of the five algorithms that I employ.
--
Pete Forman

[tso]

...The trick is with row 7, and the values 3, 4, 5 and 8. They all appear in column 2, and in box nine, so that the 3, 4, 5 and 8 in row 7 must be in r7c1, r7c3, r7c4, and r7c6 in some order. ...

I didn't see that -- now it's obvious. Still, that's the kind of thing we're talking about. Though I missed it this time, most of the time I would catch something like this. If I had filled the grid with pencil marks, it would be a "hidden quad" -- which I would probably never find or even look for.

By the way, this was a Japanese hand-composed puzzle. In retrospect, the '3458' construction you found seems *obviously* contructed -- it may well have been the *key* s/he set, not unlike chess problem composers set. Find the key, and there's an 'ah-ha' moment, the rest is easy. I'll go out on a limb an guess that you *won't* find similar constructions within randomly generated puzzles and I *will* come across this very same key again in a hand-made puzzle.

[kat]

I am only on moderate puzzles and i have to note down an awful lot of the possibilities

[Karyobin]

Aaahh yes, but how do you get to Carnegie Hall? [/xb (purple splodge)] {ziffty-squib}] {\bx]