@GNiklasch Yet another approach. Rebuilt the vampire again with a new Don Quixote methodology. Vampire can't bounce out, but you know what...I don't care. I don't care one whit. Windmill can just be a fat cow for Vampire to suck dry the whole game.
I found a way in that doesn't cause the queen or her helper to bump into Windmill's workers, which can get all the way to the queen. The only time this could possibly fail (although I haven't done a wide range of tests yet) that I can see, would be if a windmill worker is arriving/departing the queen at just the right moment when Vampire arrives.
Ah, one more rail detection bug that causes the follow-rail code to never engage. Even so, using Dave's controller I see that what I have now is 40% effective (i.e. in a Vampire v Windmill only tournament, Vampire wins 40% of the time)
Just finished reading back through your messages. Intensive work... I wonder if this will affect the position of Vampire's other victims, now that Windmill is there to hog its attention?
@trichoplax I've been running a lot of 50 game tourneys with Dave's controller. MoaR rises a little, but really, there's not ENOUGH impact on Windmill (the two pair off pretty rarely to begin with, and Vampire only strikes home one in every 2 games, give or take). No vampire, Windmill scores the most food, every time.
I'm actually fiddling with some of the "is it worth leeching?" code, as Vampire can get stuck on impossible black holes (effectively killing itself). Looks like it was almost in a good spot already, but I've had to remove the "or > 60 food" condition, as it's WAY better to go after anyone else.
Vampire depends on a last minute decision I made, that it would be better to allow workers to steal from enemy queens than to allow queens to steal from enemy workers. I was in two minds at the time, but it seems to have worked out really well. It's great to see what's possible.
Oh god, letting queens steal from workers would've been terrible. No one COULD take advantage of it.
leaders atm are Glider/Lightspeed (they change off), Miners/Windmill, Single Queen/Lonewolf/Steamroller
but yeah, if Vampire gets its teeth into Windmill, its pretty much game-over for Windmill. I have seen a game where Steamroller or Trail Eraser came by and managed to knock Vampire off (and windmill had enough laden workers close enough by to reboot the system, scoring 10 points)
It's going to a very different Hill, indeed, but probably not because of my contributions - the only things I have in mind (or am currently working on) will have negligible impact on the arena in ~90% of the games...
I've discovered a pattern that allows for an increased food collection rate by 6.67% (compared to the simple straight diagonal line approach) with just a Queen. More details coming soon-ish.
@Alion it's not a but, it's laziness (anything of ants wasn't checked for equality. I even remember thinking "oh, I'll implement it when I'll need it")
...it just works? kind of? I'm currently not doing anything to make it collect food (so it's only getting the food right in front of it, so 2 food / a pair of miners / move), but it just does and is working quite okay-ly
Oh @trichoplax You may also want to confirm expected food theft behavior in Dave's controller. I think it may not match your own controller when 2 queens have 1 worker adjacent to themselves and the other queen, particularly when in motion (e.g. when a queen moves next to an enemy worker, should that worker steal food from her?)
I managed to catch vampire stealing food from Glider, despite Glider always having 1 worker adjacent to Vampire's queen at some point during the simstep (Glider moved first, workers moved away, Vampire caught up, steals 1 food).
May also be a sitjuation where a static arrangement also causes a net flow of food, but I'm less confident on if it is actually occurring and even less so as to why (Vampire lost food to Windmill during one test, but because I can't see which ants belong to who, I am not confident in the observation).
Ran a tournament with that new bot (still exactly as unfinished as before). first game - only Windmill beat it - looked promising. But in the end, the average score was.. 2.3..
@Draco18s Currently it only ever saves food when it can't spawn miners, and really it could be twice as effective. And I'm currently trying to fix that sometimes 2 pairs of miners are sent out
empty map, have fixed that single bug (really it was a mistake of my idea of deploying the pairs but ¯_(ツ)_/¯), empty map, 923 food, 257 workers, though I made it stop making workers past 300 food
btw, Vampire updated about an hour ago. Did another edit a few minutes ago to improve the description of what I'd done to make it work (I'd run out of time prior to needing to leave for an appointment and did a quick 3-line entry).
Glider also got a small update recentish, too. I forget when (last night?) resolveing some deadlock scenarios
@dzaima That is better than any other ant's "on an empty field" result
17,000 moves and Windmill has 321 food (~70 workers)
@dzaima You're using a counting system, yeah? 3 cells worth of information would be enough to count out to 512 (two gets you to 64). If you counted to 64, and then reset, counting to 64 again...you might be able to count out 128 workers (the second time you hit 64 you'd have some threshold of food, acting as the 4th bit of storage). I wonder if that would work