@kezzamachineII
Howdy.
I like your line of thinking.
I would be very interested in seeing a SWMinis "autoplay mode"/"AI mode" develop,
it would probably be as popular as Dynamic Duos, maybe moreso even!
a SWMinis equivalent of the other table top games as
"Killteam/W40Kin40mins", XWing, Tannhauser, Imperial Assault etc would be fun for table-top.
though, an AGIXML or AI handler for a videogame approach/ "digital" approach would be fun too.
I would like to see such a project eventually develop.
SW:Killteam is already a SW40K port that could run in 'ai mode"/"single player mode".
learning from other such games,
SWMinis would benefit from "stochastic mechanisms" and pseudorandom stuff to ensure the different AI's
don't all have the same attack pattern. some folks love tabletennis against the wall - other folks get bored at that format hehe.
Auto-chess and chess-compositions are a very different beast to Chess itself.
dumping a small bag of 80+ tokens that are color coded, and attacking only the face-up areas is an example of "raider-AI".
having color-banded bases movement trays/'base rings', and having Red Group attack Blue targets, Blue Group attack Yellow and Green targets, Yellow group attack Red etc... that's another way that produces some great variations in attack strategy.
a lazer pointer ball (fresnel lens with very small holes in grating around it, and a random generator to cycle the MED/light and location) also works.
a "digital map" as some folks now have, either via Overhead Projector or a monitor in the tabletop, can also be used to make the random locations. this can randomly highlight a square, or overlay shapes as 'attack patterns' for that round.
the shape generator can be a hyperdimensional polychora which is then taking a flat section - Conway or Coxeter's work becomes very interesting...
I digress;
SWPlay had a similar idea for making a stand-alone SWMinis App,
with a view to eventually being able to map the turn structure from live games (say, for recreating tournament moves)
or being able to port squads from your squad-list
I would very much support an app project and digital approach, be it this project,
or the "boardgame-sim App" SWMinis app project or similar.
Electronic Journal of Combinatorics online has several approaches for mapping complex game states such as SWMinis "chess compositions".
- after mapping the meta-combinations of the game, and then testing those across the known maps (or, the variable maps in Tile Wars even),
an AGI routine could be developed to emulate some of the playstyle from the tournament pros.
This would be EXCELLENT for recreating and training.
I'd like to recreate several of the 2011,2012 and 2013 regionals scenarios that went to time, or the nationals last couple of rounds - those were very close games.
|it's not really n^X, rather, it's 'babushka limits'...
it's n!/(n-r)r'! - where those placeholders could be further cancel-able expressions...
or so, or the cumulative series of nPr;
the 'large number' of squares and indefinite number of distinct pieces are what make evaluating
the superset of all possible gamestates and indexing those tricky.|
Anywho -
I'm keen to hear what folks are thinking,
and seeing these new inroads for SWMinis form.
Do folks want these projects to be more "physical - table-top AI" or,
'digital, videogame AI", and why can't we shoot for the moon and aim for both?
What did folks think of VASSAL AI and SWMPlays 'simple' AI -
specifically, why are certain squad combinations easier to defeat than others, given that
your squad and play style did not change?
is this something to do with how the AI forms a get.that.|gridsquare.floatingpointvalue| update routine?
how many values would each gridsquare need to hold to rectify that?
the present 5 per square mightn't be enough... but is 11 or so too much?
is a Wilber model possible?