Talk:Griezel/Version History

0.2

Wow, 3rd place and probably a tweak or two away from the #2 spot - congrats! Looking at the survival score vs Coriantumr, maybe you could gain some points with more aggressive bullet power selection? That's generally my conclusion if I'm outranked by a bot with lower survival. Good luck chasing the MiniMelee throne. =) --Voidious 19:19, 3 September 2009 (UTC)

Thanks, curse those minibattles as they pushed me down to third place. This update was ment as a warm-up before I started with a new meleebot inspired by Portia and Diamond, but I think I'll do an update of the megaversion first to (try to) lift it into the top-10. The gap doesn't seem so big anymore after the ...Hawk experiments. --GrubbmGait 21:17, 3 September 2009 (UTC)

No need for tweaks to get to #2, due to the activity in megamelee it passed Coriantumr and is breathing Griezel 0.4 in the neck (until I update that one) --GrubbmGait 23:15, 23 September 2009 (UTC)

Fear of the Centerfield

Hmm, the recent Griezel confirms the hunches I've when developing with MiniSurreptitious (which Glacier's minrisk factors are loosely based on). When watching battles, I noticed what while MiniSurreptitious ended up in the centerfield it never seemed to take many hits, at least not significantly diferent than any other time. Thus I never used a centerfield fear myself. I think it's because so many other bots end up around the edges/coners quickly, that makes the center less dangerous than traditional wisdom advised. --Rednaxela 00:38, 4 October 2009 (UTC)

I had similar conclusions myself, with no real testing though. I think perpendicularity and keeping distance makes it relatively safe to be near the center (not for long though) after 3 or 4 bots die the center becomes extremely dangerous I think. And when there are only 5 bots in the field is safer to ram Shadow(which is pretty dangerous) than staying in the center and be targeted by all 4 bots. --zyx 00:43, 4 October 2009 (UTC)

Being in the center during the first quarter of the fight is even beneficial to your inflicted damage. Most of the time you will end up near the edge anyway, due to staying away from all others. And when there are only 5 bots left, ram anyone but Shadow, but preferrably a xxxHawk, as its MR-(flee)-behaviour is very predictable. ;-) Just staying far, far away, shooting peas and prey not to be hit is probably the best 'strategy'. --GrubbmGait 00:57, 4 October 2009 (UTC)

xxxHawk won't match HawkOnFire since it is HawkXXX =P --Nat Pavasant 02:25, 4 October 2009 (UTC)

Not literally, I mean that any 'pure MR' movement bot is predictable when fleeing. But ramming will be only successfull if there are 2 bots left AND you have considerably more energy. --GrubbmGait 14:06, 4 October 2009 (UTC)

Note the smiley --Nat Pavasant 14:41, 4 October 2009 (UTC)

And now for the logic of MR movement in meleebattles:

• put risk at corners and centerfield => rank 13
• put NO risk at corners and centerfield => rank 13
• put risk at corners but none at centerfield => rank 16
• put risk at centerfield but none at corners => rank 16

As these differences (0.3% APS) are very hard to test at home, I'm affraid I will continue with some more 0.5.x versions before I release the 0.6 with the real added functionality. Sorry. --GrubbmGait 21:56, 5 October 2009 (UTC)

I believe I see what's going on here... Performance is best with either both risks, or neither, and the cause for this would be that: 1) If you have centerfield risk only, the bot is unnaturally biased towards the edges, and 2) If you have corner risk only, the bot is unnaturally biased towards the center, 3) if you have both, then they cancel out to an extent, 4) If you have neither, the bot is more free to move to the locations which seem good in relation to walls and other bots. I think it makes a lot of sense really :) --Rednaxela 19:12, 7 October 2009 (UTC)