There was some talk last month about 'random' behavior from factories.  I feel like it ought to be possible to have a set of territories (let's say 8, although maybe you would need more), with a bunch of factories on them which although not truly random, would appear randomish.  But how do you figure out that set of factories?  I have a feeling that in most cases, you will end up in a cycle or steady state, but maybe there are some cases with chaotic patterns, that would be hard to predict from a user perspective.

Intuitively I feel like it ought to behave something like Elementary Cellular automaton:

http://en.wikipedia.org/wiki/Elementary_cellular_automaton#Single_1_histories

Any ideas on how we could determine sets of factories that would behave in interesting ways (Rule 105 is a good example)?

==============================================

I created a javascript simulator so that it is (relatively) easy to see the effect of factories:

http://prestopnik.com/wargear/RandomFactories/

The UI is still pretty rough around the edges, and please look out for bugs, as I have not tested it exhaustively.  I have some ideas for improvements, but I wanted to see what interest level their was in the community before I invested more time.

http://mathworld.wolfram.com/ElementaryCellularAutomaton.html

http://plato.stanford.edu/entries/cellular-automata/index.html

I haven't done much with the "programming" of factories.  Can you do Cellular Automation (CA) on, say, 8 of them?  If you hide them off screen can the player "see" them in any way?

A random number generator like this would depend on a seed - you could accomplish this by assigning randomly assigning neutral territories to the "cells territories".

I can't find the "rule" for 105, but one page I link to says Rule 30 and rule 90 are "rules leading to seemingly chaotic, non-periodic behavior".

This stuff is fascinating. I haven't been keeping up with features of factories. Do you have a way to actually simulate the elementary CAs (or any CA) with factories?

There's a lot of open questions about these things. Rule 110 is capable of supporting universal computation; it can simulate a Turing machine. So too can Conway's game of life: http://www.youtube.com/watch?v=My8AsV7bA94

So in principle 110 can do anything. On more or less random inputs, my guess is that 110 acts random/complex. Rule 90 just XORs a cell's two neighbors, yet it leads to interesting behavior!

Think of that example, where your students are wearing hats. (rule 90)

I've always seen everything in the world working like that, except that every once in a while, one of your students chooses to wear a hat and then sit next to another. Breaking the rule.   That Creates "Breaks" or "resets" in the pattern, where it will start-over at another point in the line.

And that's life, a fairly predictable set of patterns, all broken up by random events. 

Here, if you created one set of factories that contained whichever one of those rules for each terriotry on the board, you could get all the territories to follow the patterns. which would seem really chaotic on a turn by turn basis. Unless you were a ridiculously clever player, and could work out the the underlying rule based on the behavior of the territories, and then predict it.  In which case, good for you, you deserve to win that everytime. 

I think you could simulate CAs with factories, although it would take a lot of factories.  I was hoping we could find something with similar behavior, but simpler.   In my Finite State Machine board, I create factories to handle transitions based upon a 3 bit set, and it takes about 20 factories to translate 3 bits into 8 different actions.

To see why, think of a case with just two bits A and B.  You want four different situations 00, 01, 10, and 11.  Well 00, is a default state, so you make that a factory tied to a member that is always on.  But then to implement the 01 state changes, you need two factories (both with just B as the member) - one to undo whatever the default (i.e.  00 ) factory did, and one to do what you actually want.  Same thing with the 10 state.  For the 11 state, you need three factories, two to undo the 10 and 01 factories and one to do what you actually want for 11. 

Anyway, you can see that with ~20 factories you could implement an application of a ECA rule at a single point.  So you just need 20 factories * however many ECA elements you have, and you could create an ECA in wargear factories.

If that's the best way to do something like this, I could make a tool so that it is available to everyone, but I was wondering if it there would be a simpler way to make it happen.

I'm thinking that I should add the ability to import/export a factory set from the factory simulator to make it a bit easier to talk about and share factory sets. 

Ratsy/Hugh - Since Rule 90 is just an xor, that might be simple enough to implement with a ton of factories.  Certainly a good place to start, and once I have some more time to work on this I'll let you know if I get anywhere.

Hugh - I've been thinking for years of a 'Game of Life' based wargear map.  I picture two players on opposite sides of a grid, creating glider guns, etc to shoot at each other.  The only problem was even the fastest objects only move a space every two turns, which would make games extremely long and slow moving. If anyone has suggestions for how to make Game of Life dynamics a bit more interesting in wargear, I'd love to hear it.

My instinctual problem with the complex use of factories is that they take at least one turn to actualize, and probably more than one in most cases.  If we had RTF's, different story.