Keep the wheels turning

October 25, 2005 11:29 AM

Set the Arcs in Motion [Flash game]
posted by Gyan (42 comments total)

1486, that was kinda cool.
posted by djseafood at 11:33 AM on October 25, 2005

Curse you, djseafood, I only got 1484?

By essentially picking an arc at random. Is it just me, or is this game so complex as to be essentially random? Perhaps I'm just too simple for it.
posted by gurple at 11:39 AM on October 25, 2005

kinda cool.. but too bad there's not more strategy.. like a way to move a block without setting off a chain reaction..
posted by pez_LPhiE at 11:40 AM on October 25, 2005

1486 on my third try. Neat, but is it just a game of chance?
posted by you just lost the game at 11:43 AM on October 25, 2005

Seems to complex to be solvable, but it's still fun randomly clicking and watching the chain reaction.
posted by cloeburner at 11:47 AM on October 25, 2005

1881.

I agree that it mostly seems random, though I'm sure that with enough time one could discern a pattern or two. My only strategy is to start with an arc relatively near the middle that will trigger multiple-pathed reactions. I figure this maximizes the chance of a catastrophic arcbuster, which is the key to the big scores.
posted by jdroth at 11:49 AM on October 25, 2005

2538. Booyah.
posted by Faint of Butt at 11:54 AM on October 25, 2005

2551.

Although my first couple of tries got 6 and 8.
posted by jamesonandwater at 11:54 AM on October 25, 2005

This game is sooooo addicting. I played it for an hour last night. Max score was 1887.
posted by SirOmega at 11:56 AM on October 25, 2005

It's not just random -- the best moves I've found so far only work once the board has "self organized itself" (which seems to be, most frequently, two side of the square across a diagnoal with all the faces in that half already facing the same direction.) 2248 so far ... hmm, wonder what the theortical max would be?
posted by bclark at 11:58 AM on October 25, 2005

Oddly entertaining. Note there's a bug that allows you to set off several chains at the same time by repeatedly clicking a different cog while there is already action going on elsewhere; the score reported is since the last cog you set into action though.
posted by fvw at 11:58 AM on October 25, 2005

Yes, compelling and random.
posted by OmieWise at 11:59 AM on October 25, 2005

1713. :|
posted by jenovus at 12:02 PM on October 25, 2005

Whoa... 1,000,256. Took over 4 hours to complete.
posted by Witty at 12:03 PM on October 25, 2005

I got 2600-ish this weekend. I was pretty fascinated. Is it possible to set it into perpetual motion? Witty, was there a plan to your approach?
posted by sohcahtoa at 12:13 PM on October 25, 2005

Disinfect the Core works on a similar principle, but pits you against a computer player, and seems a little more strategy-susceptible. (via jay)
posted by gorillawarfare at 12:44 PM on October 25, 2005

It's a cellular automaton, you can work out sequences that will at the very least run for a long time. If you spend a lot of time setting up your initial starting condition you should be able to get horrendous scores.
posted by substrate at 12:45 PM on October 25, 2005

I am experimenting with making patterns. I have not achieved any great score so far, but am enjoying this slightly more than working. I have managed to get them all pointed in the same direction and am working on a pattern that will result in a good, long run.
posted by beetsuits at 1:19 PM on October 25, 2005

2613. Damn you Gyan!
posted by medium format at 1:53 PM on October 25, 2005

sohcahtoa, look at the time stamp on the FPP and the time stamp on Witty's post. As of my post, it hasn't even been four hours yet.
posted by emelenjr at 2:48 PM on October 25, 2005

gorillaware's link's incredibly fun once you figure out the trick!
posted by fnerg at 4:43 PM on October 25, 2005

I hate you all and will not post my score. Until I get something respectable, at which time I will again love all of you and all mankind. For the moment, though. Go 'way.
posted by mmahaffie at 4:47 PM on October 25, 2005

Hey, that was fun.

1885 by the way.
posted by Relay at 5:36 PM on October 25, 2005

3097 without really trying. I got lucky I guess. It's mesmerizing to watch.
posted by mai at 6:02 PM on October 25, 2005

This took some effort. I'm not sure it was worth it.
posted by MrMoonPie at 6:38 PM on October 25, 2005

Well done MrMoonPie. I was playing with a similar approach. Now I'm trying to create alternating rows that will trigger each other. So far I've reached 2464 with various combinations. Of course, I've blown off all the work I brought home tonight....
posted by mmahaffie at 7:14 PM on October 25, 2005

1658, awwyeah. I'm sure I'll be here all night, too....
posted by kalimac at 7:20 PM on October 25, 2005

Anyone have an patterns that generate high scores? I've done a two different alternating row designs that yeild ~1000 and ~1400.
posted by betaray at 7:24 PM on October 25, 2005

got 2348 on my first try. started doing patterns - squares of the pointed in the same direction doesn't really do much.
posted by muddgirl at 7:39 PM on October 25, 2005

Try all _| and then in the last column have 7. It certainly wins for time.
posted by betaray at 7:53 PM on October 25, 2005

emelenjr: I saw it on Memepool over the weekend. I assumed I wasn't the only one.
posted by sohcahtoa at 8:00 PM on October 25, 2005

This reminds me of a 4 state Potts model.
posted by Nelson at 2:03 AM on October 26, 2005

I should think it's possible to set up a self-sustaining pattern.
posted by five fresh fish at 1:09 PM on October 26, 2005

There is a (relatively) simple inductive proof of the inability to achieve perpetual motion on a finite board. There is no perpetual motion possible on a 2x2 board, assume there is no perpetual motion possible on a convex (with respect to a square lattice) board of n circles. We want to show there is no perpetual motion possible on any n+1-board we create from the current n-board by attaching another circle. If we could have perpetual motion, it would have to involve turning the new circle an infinite number of times, or else perpetual motion would have been possible on the n-board. However, due to convexity, the new circle can touch at most 2 of the other circles. Thus, in turning the new circle it will eventually become inactive when it points away from all of the other circles. This contradicts our statement that it would have to turn an infinite number of times to achieve perpetual motion. Thus, no perpetual motion on the n-board implies no perpetual motion on the n+1-board, and by induction on n, there is no perpetual motion possible on any finite board.

Clearly perpetual motion is possible on an infinite board. Interesting questions involve a) the possibility of perpetual motion on a torus or cylinder, b) calculating the maximal number of iterations for a given board of size n.

Apologies for the amount of math in this post/shattering anyones hopes of being the first to infinity..
posted by shoesandships at 3:55 PM on October 26, 2005

Er, yah. What he said.
posted by five fresh fish at 5:17 PM on October 26, 2005

Shoesandships, I'm interested in your proof but not quite following it.

In particular, why does your inductive step involve "attaching another circle"? The way I'm understanding "circle", it seems like this would create a board of size n+2, making your proof only applicable to even-sided boards.

Also, this part is completely obscure to me: "However, due to convexity, the new circle can touch at most 2 of the other circles. Thus, in turning the new circle it will eventually become inactive when it points away from all of the other circles." Maybe it was meant to be obscure, and I'm missing the joke.
posted by gorillawarfare at 11:18 PM on October 26, 2005

A circle means one of just one of the playing pieces. The board is not required to maintain a rectangular shape, however (the convexity part) restricts us from forming odd shapes involving holes and notches and the like. As long as we have these restrictions, and we add another circle to the outside of the board, it will only be touching either one or two of the other circles. If you play with the board in the link, the circles in the corners of the board are examples of circles that touch only two others. Once you have set the game in motion, the circle in the top right corner will only be able to move at most three times. Eventually it will reach a state where it points away from the board and nothing the board does will be able to activate it again. This is the fate of any circle we add to a board to attempt to achieve perpetual motion... it will be an outsider and eventually become inactive.

The idea behind the induction is to show that if you have an arbitrarily large board that doesn't have perpetual possibilities, adding one more circle isn't going to help it any. Since this was shown true for any size, and we know the 4-board doesn't work (no perpetual motion), we then know that any 5-board we can make from it won't work either. Since we know that any of the five boards won't work, we know that any 6-board we can make won't work as well. Then we get that any boards of size 7, 8, 9, ... won't work, ad infinitum.
posted by shoesandships at 8:21 AM on October 27, 2005

The eventual outcome of any sequence on the circle game is a pattern with all units oriented away from each other because the ends of the circles act as repulsers and eventually they all come to rest away from each other. No math necessary.

I wouldn't characterize this as cellular autonoma. It doesn't possess the the requisite complexity. Perpetual states can be achieved on a finite board. There are plenty of stable configurations in the Game of Life. Thats one of the things that keeps it artificial.
posted by Mr T at 9:01 AM on October 27, 2005

Thanks a lot for that extra clarification, shoesanships. I get what you're after now, and it seems like your proof is sound.

And Mr T, it's true that you get the same idea across more succinctly. But we wouldn't have math if nobody wanted to formalize things like this! And where would we be without math, man?
posted by gorillawarfare at 9:23 AM on October 27, 2005

2997
posted by crunchland at 11:31 AM on October 27, 2005

I always think of this line when talking about math for math's sake:

Where the Baddelaries partisans are still out to mathmaster Malachus Micgranes and the Verdons catapelting the camibalistics out of the Whoyteboyce of Hoodie Head.

Put that in your pipe and smoke it!
posted by Mr T at 1:03 PM on October 27, 2005

I got the highest score so far … 3193!!

Witty claims he got 1,000,256 and it took 4 hours but I don’t believe him. I've been obsessively playing this everyday for the past 18 days and my highest scores have been 2047, 2346 and 2527. I get the feeling that a score of one million is impossible.

Here are some observations:

Circles, edge hoops and any pattern that prevents a connection are bad.

To get a high score I hit reset over and over until I find a board relatively free of these structures. Then I tweak the grids by rotating them so they can connect with other grids. I keep tweaking until I accidentally set it off.

I also look for barriers that can inhibit growth. This example shows an area that can never expand because it cannot make a connection. You have to remove such a barrier or start behind it.

I've coded a “grid game” simulator which calculates the score for each grid given a certain board configuration. It still has some bugs to work out before I can post a link here.

It can generate a random board or allow you to hand design any board or you can tell it to generate any pattern. I'm very curious what the absolute high score is. I'm hoping that I can convince the author of “grid game” to allow you to preset the grids before starting off a run.
posted by patcoston at 10:08 AM on November 11, 2005