[projects] Random Projects

[jrising]

This has been a week of random projects. Recently or newly on my plate...

• Dialogue Grapher: A friend and I have long wanted to create a collaborative idea framework-- think message boards, except each message is a single line and connected to several others in a graph, and the whole structure is publicly editable (see an example, or the diagram in my LJ bio). It could be used to communicate complex ideas, as a new kind of wikipedia, for refining Lakoffian arguments, and even to explore the unconscious structure of society.
  
  
• General Sudoku-Maker: I have no interest in Sudoku puzzles, but the current 6.001 project is an automated Sudoku solver, and I think I've figured out a method for automatically constructing Sudoku puzzles of arbitrary difficulty. Anyone have ideas for Sudoku variations that would make such a program worthwhile?
  
  
• "Lips" Trixie: I was the lips, with two possessed hands that stripped off clothes that no one could see anyway because I was covered in black. It was fun, and I think went well as choreographed-goofing-off April Fools experiments go. The weirdest part was how differently people treated me when my face was black.
  
  
• Rocky Party: I didn't know it was asavitzk's birthday until too late (*pout*), so I hosted anyway. I hope the other half of the split party had fun, because we certainly did. And for once, no casualties! AND, at girlygothic's insistence, some of us finally played...
  
  
• Rocky Monopoly: The Rocky Monopoly game was a great success! It was easy to play, and got lots of people paying and receiving (minor) sexual favors. I'm going to incorporate some new innovations over the next week. Tell me if you want me to bring the game to your party.
  
  
• Mental Reorganization: I've been reorganizing my mind to relegate my subvocal "I", increase my sensory awareness, become more conscious of the collective unconscious. Good fun.
  

Comments

[catullus-5.livejournal.com]

I was once trapped in a lab with no Internet access, so I wrote a sudoku solver and then a sudoku creator.

Creating valid sudoku puzzles, at least with the method I chose, is much harder than solving them. What I did was randomly place numbers one by one into random squares on the 9x9 grid, making sure each newly placed number was legal to put there. Eventually I would get to a point where I couldn't place any numbers legally. Then I would randomly remove one number and resume trying. Eventually the program would successfully place the 81st number to create a validly filled-in grid. The next step was to randomly pick "givens," the filled-in squares that are supplied to the player. This isn't trivial either, because often you wind up with multiple solutions. So I repeatedly picked sets of givens and then solved the puzzle to make sure the filled-in grid I'd generated was still the only solution that fit those givens. The whole thing took several seconds per puzzle, and the fewer the givens, the longer it took. There has got to be a more efficient way of generating sudoku games than my quick-and-dirty monte carlo placement method, but I just implemented the first method that came to mind. My ultimate goal, in the end, was to generate a sudoku puzzle so I could solve it myself and pass the time that way. Needless to say, the programming ended up keeping me much more busy.

[jrising.livejournal.com]

I have something more sophisticated that I'm working with, as part of this class project. Each sudoku space is represented as a set of possible values, which is refined as other values are put in. So when I'm constructing a random filled-in square, I randomly specify one of the possible values, and determine all of the logical consequences of that before choosing my next value. I can still get inconsistent grids and have to backtrack, but it isn't so bad.

For generating interesting puzzles, I can basically say that the difficulty level corresponds to the number of times a player is forced to guess a value because there aren't any immediately logically constrained spaces. Most sudoku puzzles, by that measure, have difficulty 0, so if I want to generate a puzzle of difficulty 3, the tree of possibilities should only have about 81^3 branches, which is fairly doable.

[revolos55.livejournal.com]

*blink*

*blinkblink*

Umm, sure. Good luck all that... stuff.

[jrising.livejournal.com]

Thanks-- I'll need it.

*smiles and nods*

[revolos55.livejournal.com]

Meanwhile, the rest of we monkeys will be figuring out exactly how to turn bones into weapons.

[jducoeur]

Dialogue Grapher: I may have to pull you into the Querki project once that gets off the ground. It's a cross between an OO database and a wiki; I suspect this would be a rather interesting side-project for it at some point. (I'm already planning on having true comment threading in the wiki, with arbitrary linkages, but this goes to a whole new level.)

Anyone have ideas for Sudoku variations that would make such a program worthwhile?

Well, there are a bunch of great Sudoku variants out there -- I tend to play multi-Sudoku more often than I do the base game, for example. If you want ideas, there's a good book of Sudoku Variations that I've got at home...

[jrising.livejournal.com]

Querki sounds interesting, but I don't have a clear sense of it from your the pages I pulled up on ProWiki. We should talk sometime (and I may have to cross-pollinate with one of my way-back-burner projects-- sort of an all-properties-based content system).

Thanks for the pointer on multi-Sudoku! I'm not currently motivated enough to want to see a book-full of variations, but now I have a place to start.

[jducoeur]

Querki sounds interesting, but I don't have a clear sense of it from your the pages I pulled up on ProWiki. We should talk sometime (and I may have to cross-pollinate with one of my way-back-burner projects-- sort of an all-properties-based content system).

Sounds like a similar idea, yes.

Basically, Querki is the written-from-scratch rewrite of ProWiki, the wiki system that I've been working on for several years now. That grows out of my LARP writing -- after many games, I came to understand that it's basically a world-building problem, and I know from experience that the best way to manage world-building is using object-oriented databases. So I took UseMod (one of the primordial wiki bases), and rewrote it extensively, winding up with ProWiki ("property-based wiki"), which I've used for writing my past couple of games. I generally describe ProWiki as well-suited for "semi-structured data": a mix of highly structured objects with more loosely structured connective tissue, which seems to be ideal for LARP.

At this point, though, I've pretty much decided that ProWiki is at the end of its life. It desperately wants a far fuller feature set, it's hideously unscalable as currently constructed, and frankly I'm sick of trying to deal with something this complex in Perl (a language I don't care for much). So the plan, probably starting sometime this summer, is to begin writing a new system from scratch, based on a properly designed database. Language is as yet undecided (I'm enamored of Ruby, but not sure it's capable of the job), but whatever I do, it's going to be an open-source project, intended to run on multiple platforms, with professional-grade architecture and code. Once I've got the basic architecture and bootstrap functionality in place, I'm going to be looking for people to join in and play.

Thanks for the pointer on multi-Sudoku! I'm not currently motivated enough to want to see a book-full of variations, but now I have a place to start.

You're welcome. I've only dabbled slightly with the other variations. Multi-sudoku doesn't add a lot of strategic depth to the game (indeed, it's actually slightly easier to solve), but I find it a lot of fun...