Thanks for liking the third puzzle. It was my first real go at an interesting one. Still a lot to learn.
I'm not sure all the other ideas are one-offs. I think just changing numbers to letters and keeping rules as is can make the presentation more fresh at times. Not all letters have to perfectly spell a message. Still, you ask, "Should I write hitori puzzles with sudoku kind of steps?"
or,
SHOULD IWRITE HITORI PUZZLE SWITHS UDOKUK
and that works, if the message is shortened. That kind of thing made the MIT meta fun, as a crossword-based logic puzzle actually felt word puzzle-like. The numbers feel forced to me, and are very hard to scan when they become double digit numbers (my same problem with 16x16 sudoku that use 1-16 instead of 0-F or something else).
Still, I'm not totally convinced. If you were to give me a whole book of wordy hitori in larger sizes I'm not sure I'd like it any more than a good book of standard hitori (which is not to say I'd dislike either by the way); although there's certainly an excellent novelty factor in the word puzzles.
Amen! What I've always found frustrating about Hitori (or as I fondly refer to it, "HATEori") is that tedious scanning up and down a row/column every time I mark a square white. The problem is that there is so much information because all the cells start out with digits in them. The other thing that bugs me is that it's usually an advantage (for me anyway) to take some time to use uniqueness (if an 11, say, is the only non-white 11 in its row and column, then it must be white) to eliminate "clues" that have no effect on the solving of the puzzle just so I can concentrate on things that might be helpful. When the style of the puzzle actively encourages you to cheat, that's a problem.
I'd go one step further on the "encouraging you to cheat" meta nature of Hitori. If you see a cell with no duplicates in its row and column, nothing will ever force it to be black, so it must be white (as noted by stigant). But more strongly, something must FORCE it to be white in order for the solution to be unique, and you can sometimes actually exploit that to figure out significant bits of logic. "If the puzzle goes this way, then really it doesn't matter whether this square is black or white ... so it must go this other way where it must absolutely be white...".
And amen to the tedious scanning aspect of Hitori. I do actually like some of the logic in the better Nikoli puzzles, but it's very similar to the logic in Heyawake and Heyawake is way less annoying.
I've got no way to eliminate the "cheating" aspect from Hitori with meta-solution thinking. Excluding the extra digits makes some of the spots easy to see (like the upper-left of puzzle A forces the 4 in row 2, column 2 to be black as otherwise the whole grid has 2 solutions. But this is true in all presentation of the form. If anything, it encourages a constructor to put extra digits (you sometimes have extra constrained digits to place) into such spots to block immediate realizations. Still, its one of the things that has always felt "off" in hitori.
I have a fondness for the concept of Hitori, and there's a lot of wonderful things that can be done with it. On nikoli.com the Hitori time trials and Botsu Bako puzzles are some of my favorite puzzles on the site. Yet I do find the same problems with the type as you and everyone else does. I think it's mostly the fact that getting unstuck in a Hitori is often a disgusting facepalm moment, when in other nikoli puzzles it's usually an enjoyable "Aha!" Really good Hitori have the latter too, but they often get shadowed by the gruntwork in between
( ... )
Yes, color can help. And in the vein of Hitori variants, I meant to point to your Out of Sight puzzles. They have some of the "minimal" feel I think works well, with some extra elements that give it a more complete logical set. If only we could get more Out of Sight authors - or other fresh versions - and get the regular Hitori out of sight. Maybe I'm being too harsh, but your point on the facepalm and not the "Aha" is a well crystallized summary of how this puzzle fails with the solver's general experience.
I agree with MellowMelon; I think you're being too harsh on Nikoli's brand of Hitori. The rules of Hitori are rich enough for some very elegant and creative logic (unlike, say, Kakuro). I don't judge Hitori by all its shovelware incarnations; I judge it by how elegant the best examples can get. Which is, after all, the theme of your blog! :)
Yes, you need to trust that the puzzle designer isn't just relying on a boring, cheap, computer-friendly and human-unfriendly trick... but I have that same trust issue with other puzzle types, including Sudoku.
And yes, there are some powerful, dubious meta-solving techniques, but it's hardly the only puzzle type with those. Heck, in Dominos and Numberlink puzzles you're pretty much expected to use uniqueness constraints while solving.
I think I'm arguing that both computer-generated ones and Nikoli ones fall into different regimes of being unsatisfying. I'm proposing a change to the puzzle that would make either form much more solvable and allow the interesting logic to spring forward, instead of most often running into a puzzle where my bookkeeping was the sticking point instead of something fundamentally interesting.
Still, while my opinions may not match yours, Hitori is far and away my least favorite Nikoli type. (25x25 Sudoku wins the next spot, but that's just a needlessly unfun variant of a type that I avoid twice a year, not a type that appears much more frequently.) I prefer the mechanical nature of their Kakuro much more, maybe as a test of pattern recognition more than math, maybe for the "race" aspect to it. I've bought maybe 100 nikoli books - 10 of which being kakuro pencil puzzle books. I've bought 0 Hitori books. So for me, it has problems. And I should be its audience as an observational logic puzzle solver.
I remember playing those awhile ago. The sparse grid in the second one worked well as I recalled, and certainly maintains more of the Heyawake feel where fewer numbers are wanted but also serving this "enough digits to do Hitori but not every digit" goal I seem to be offering.
And, I think the full grids can work as well because (depending on the Heyawake-ness) the clues that would be noise in a plain Hitori puzzle might be key to the hybrid solution.
Re: #2, I remember wondering at the time why the full set of numbers was necessary. The sparse grid of course enables one to have a nice theme or symmetry to the clue set (as well as the room layout).
I agree that your first puzzle felt very much like a computer-generated puzzle -- I wouldn't have known it wasn't made by a computer had you not told me. The other two were better than the first one, and showed signs of being handmade. I noticed the 121/212, 34/43/34, etc. blocks in the third one -- that was nice. :)
I can't say Hitori's my favorite Nikoli puzzle, but I give it much more love than a lot of people do, and actually enjoy solving and making them with no empty cells (not to say that the puzzles with empty cells automatically suck). In fact, I have purchased and completed all three Hitori volumes in the Pencil Puzzle Book series -- and thoroughly enjoyed them. Every time I see someone bash Hitori, I'm rather tempted to pull a Chris Crocker and make a YouTube video of me sobbing and crying, "LEAVE HITORI ALONE!" Feel free to ignore my opinion, though -- I like all kinds of things that no intelligent person should like. :P
Comments 25
Reply
Tom.C
Reply
I'm not sure all the other ideas are one-offs. I think just changing numbers to letters and keeping rules as is can make the presentation more fresh at times. Not all letters have to perfectly spell a message. Still, you ask, "Should I write hitori puzzles with sudoku kind of steps?"
or,
SHOULD
IWRITE
HITORI
PUZZLE
SWITHS
UDOKUK
and that works, if the message is shortened. That kind of thing made the MIT meta fun, as a crossword-based logic puzzle actually felt word puzzle-like. The numbers feel forced to me, and are very hard to scan when they become double digit numbers (my same problem with 16x16 sudoku that use 1-16 instead of 0-F or something else).
Reply
Still, I'm not totally convinced. If you were to give me a whole book of wordy hitori in larger sizes I'm not sure I'd like it any more than a good book of standard hitori (which is not to say I'd dislike either by the way); although there's certainly an excellent novelty factor in the word puzzles.
Tom.C
Reply
Reply
And amen to the tedious scanning aspect of Hitori. I do actually like some of the logic in the better Nikoli puzzles, but it's very similar to the logic in Heyawake and Heyawake is way less annoying.
- Jack
Reply
Reply
Reply
Reply
Yes, you need to trust that the puzzle designer isn't just relying on a boring, cheap, computer-friendly and human-unfriendly trick... but I have that same trust issue with other puzzle types, including Sudoku.
And yes, there are some powerful, dubious meta-solving techniques, but it's hardly the only puzzle type with those. Heck, in Dominos and Numberlink puzzles you're pretty much expected to use uniqueness constraints while solving.
- Derek
Reply
Still, while my opinions may not match yours, Hitori is far and away my least favorite Nikoli type. (25x25 Sudoku wins the next spot, but that's just a needlessly unfun variant of a type that I avoid twice a year, not a type that appears much more frequently.) I prefer the mechanical nature of their Kakuro much more, maybe as a test of pattern recognition more than math, maybe for the "race" aspect to it. I've bought maybe 100 nikoli books - 10 of which being kakuro pencil puzzle books. I've bought 0 Hitori books. So for me, it has problems. And I should be its audience as an observational logic puzzle solver.
Reply
Heyawake + Hitori = Heyatori:
http://grandpascorpion.livejournal.com/4872.html
The Heyawake constraints help to solve the noise issue.
Reply
Reply
Re: #2, I remember wondering at the time why the full set of numbers was necessary. The sparse grid of course enables one to have a nice theme or symmetry to the clue set (as well as the room layout).
Interesting post.
Reply
Reply
I can't say Hitori's my favorite Nikoli puzzle, but I give it much more love than a lot of people do, and actually enjoy solving and making them with no empty cells (not to say that the puzzles with empty cells automatically suck). In fact, I have purchased and completed all three Hitori volumes in the Pencil Puzzle Book series -- and thoroughly enjoyed them. Every time I see someone bash Hitori, I'm rather tempted to pull a Chris Crocker and make a YouTube video of me sobbing and crying, "LEAVE HITORI ALONE!" Feel free to ignore my opinion, though -- I like all kinds of things that no intelligent person should like. :P
Reply
Leave a comment