Robo Home | Changes | Preferences | AllPages

TheBestBot is currently Shadow

Overall 1v1 Melee Team
Any size Shadow DrussGT Shadow Shadow
Mini Coriantumr or GlowBlowMelee WeeksOnEnd GlowBlowMelee GlowingHawks
Micro Troodon or ShizToorkild Sprout ArmyOfShiz
Nano DoctorBob NeophytePattern Lib DoctorBob


Added a few opinions/observations -- Kawigi

Looks like ShadowTeam really is the best team around. I don't know what's keeping Paul from updating DTTeam with DT's latest melee movement (and kill that nasty memory bug), I'll enjoy my 1st place in the meantime... :) -- ABC

Hey Paul, why don't you just include the team in your SandboxDT jar like Cake does? With Cake I have the bot and the team both it vuen.Cake_2.5.jar and use it in all competitions. That way when I update Cake it just updates the team with it (not that I'm working on Cake anymore, but you get the idea.) -- Vuen

I think this has been neglected a bit? -- Kuuran

@Vuen I think it's because he wants a separate 200 KB of space for Team, 1-v-1, and melee... ;-) --David Alves

Wow, Jamougha. Wow. --David Alves

i don't know how to do that cool thing that only shows up on changes. but isn't raikomx ahead by 50 points?--andrew

The cool thing is done by filling in the "Summery" field when you post a change. RaikoMX is clearly listed as the top 1-v-1 general bot. But have you seen it perform in melee? I haven't. But I bet it doesn't. -- PEZ

Curiosity got the better of me:

1st: jam.RaikoMX 0.29	79532	40700	6270	26065	3145	3294	57	57	1	4
2nd: sample.SpinBot	62835	40100	770	20235	1413	307	9	7	26	23
3rd: sample.Walls	61042	40050	3190	16448	1171	170	12	29	14	12
4th: sample.Tracker	50391	32900	220	15097	1707	421	45	2	20	17
5th: sample.Crazy	48980	40400	550	7180	180	649	19	5	30	19
6th: sample.Fire	33846	24800	0	8603	422	21	0	0	4	6
7th: sample.RamFire	32653	17000	0	10735	330	3613	974	0	0	3
8th: sample.TrackFire	31606	18250	0	12432	924	0	0	0	0	5
9th: sample.MyFirstRobot	31319	26600	0	4526	60	128	3	0	5	9
10th: sample.Corners	25007	17550	0	7081	339	36	0	0	0	0
11th: sample.Target	19411	19350	0	0	0	61	0	0	0	3
12th: sample.SittingDuck	12000	12000	0	0	0	0	0	0	0	0

Didn't do even half as badly as i thought it would...--Brainfade

Is it even programmed to know about melee? I would have thought not... in which case, it did extremely well... -- Tango

Extremely well? You mean not just spewing out exceptions? -- PEZ

Not spewing out exceptions is certainly a good start. Putting a OneOnOne bot into a Melee can result in all kinds of mess, not just exceptions. Trying to orbit all the bots at once can produce interesting movements, for example. Shooting at one bot using anothers stats is the most obvious, of course. -- Tango

Most bots spew easily fixed at zero codesize cost exceptions in melee, so I guess Raiko is one step ahead. -- Kuuran

i think we should decide if codesize restricted bots are ranked the best due to how they do on general ranking, or their specific ranking. for instance, mike zhang's adept/adept bsp is third in nano but does better in general than any other nano bot. adept should be put as best nano! --andrew

In melee, even more distinctions could be made, but we hardly have the tournaments to show it - Lib is the best melee nano against good melee bots. Infinity, Gem and Lib are really close to each other for best against good nano melee bots, Infinity probably has the edge. And I think Infinity is the best against non-melee nanos of those ones. Shiz is also really close to HawkOnFire against only melee micros, but is clearly inferior against bigger melee bots. -- Kawigi

I build nanos to fight nanos, not to fight general fields, so at least as far as I'm concerned the general rating is quite irrelevant. I think most of the other mini authors also do this just because the MiniBotChallenge conditioned us this way. -- Kuuran

At the nano level, I agree with you, but I think most people making minis now are trying to test more advanced concepts in a simplified environment - basically a mini is an excuse to put out a bot that's good at one or two things without trying to make a megabot that does everything (notice that I've been spending more time playing around with what I can fit into Coriantumr nowadays than trying to put out a more serious version of FloodHT - because it wouldn't seem worthy of a new version without some serious WaveSurfing, lots of gun testing, making the melee movement better than Coriantumr's and putting it back into the top 3 range in teams). -- Kawigi

Yeah, that's a pretty good point, I was looking at it from my angle and kind of considered some of the newer minis to be essentially in the megabot arena. I'll agree that as you get bigger it starts to change, but, definitely down low and higher up depending on the author, the majority of minis are still probably targeted at competition in their codesize. I think it's easier to think of all bots in a codesize to be competing as that size, so a nano in general is competing as a megabot and effectively a megabot then, so rank it against other megabots, same for a micro in mini, etc. Conversely, if you're planning on competing in mini or general, don't make a nano or micro, use all your mini or general bytes. Just my perspective. -- Kuuran

Totally agree with Kuuran here. There's no point with fighting megabots with a codesize constrained bot, other than it might feel nice to beat DT 1.61 with a micro =). A minis performance should be measured against its own class mainly I think. -- PEZ

So is the consensus that we should remove Adept? I'd feel bad about doing that, it's the first top nano in a long time where the author wrote an original gun instead of using a version of NanoLauLectric?'s symbolic pattern matcher. --David Alves

There is precedent for having a bot which isn't top at it's own codesize; e.g. Smoke, which was ranked as best Micro while at #2 in micro, I guess because it has really great general and PL scores. -- Jamougha

This page was created mostly to give the best bots some central attention. There is certainly room for listing more than one bot as "the best" in each category. Go ahead. I just mean that if I want to check which nano is the best I check the nano table and if one of the bots there lead with a margin I consider that bot the best. If it's a tighter race I might consider other aspects too. -- PEZ

Yeah, I'm not saying remove the bot, it's certainly a worthy contender and this new version might even be #1 nano at the rate it's going. I'm just offering a general rule of thumb to clarify the questions asked above, call it future consideration. I think the nano division is probably still too tight to pull a definite winner out of, perhaps we should put 'no one puts anything here' back for the time being? Or we can leave both, I certainly don't mind having my bot's name in lights ;) -- Kuuran

I think the closeness of the nano-1-on-1 division is there because it's difficult to make a nano that beats all the other nanos. FunkyChicken is (to this day) among the strongest or the strongest in the nano PL, but still has 9 losses (give or take, depending on the day). -- Kawigi

While I question the tactic of releasing Acero to bring down NanoSatan's rating before being able to pass him, I'll have top back before the weekend :) -- Kuuran

i was kind of kidding when i wrote that. i really just wanted to try to put wall smoothing flat movement in a nano. and it's tactic happens to beat pm guns. anyway, drbob is a just clearly a far superior bot than nanosatan.--andrew

Maybe I'm missing the sarcasm? DrBob? is hardly far superior in 1v1. -- Alcatraz

I think he's refering to the fact that it is #1 in 1 on 1 and melee. -- Mike

Ah. Yes that is impressive, especially for a nano. -- Alcatraz

So what are the requirements nowadays? I have a sudden, not so humble, interest in this :-) -- Pulsar

Yay! I think I have a best bot now. -- Alcatraz

What makes you say that Shadow is the best bot over all, David? Have you tested it in melee much? -- PEZ

Well Shadow is #1 in 1v1, #3 in Melee and #2 in Teams. The only other real contender, Aleph, is #1 in Melee / Teams but a distant #15 in 1v1 (That's what I remember, rankings are down now so I can't check). To me that says that Shadow is TheBestBot. Feel free to change it if you disagree. I added "currently" to soften the impact of that sentence somewhat :-) --David Alves

The current Shadow is untested in melee and teams. Maybe we should wake those games up. -- PEZ

I just tested, Shadow 3.49.1 throws an exception at the end of each round in melee, I don't think it affects the result, I fixed it for the next release. The only change I did that could affect it's melee score was the bullet power change. Please wake those games up, I miss them a lot. -- ABC

Yes, please wake them up if possible. My bots do significantly better in melee (as can be seen in the RobocodeLittleLeague). However they will still not be 'topbots'. -- GrubbmGait

I'd second that, I just don't have as much time to tweak my melee bots as I did before. -- Kawigi

just added my two teams GlowingHawks and HOFSwarm. -- rozu

Should HawkOnFire not be the best overall microbot? --GrubbmGait

no Troodon is right there. I mean HawkOnFire version 0.1 is not the same as the one used in HOFSwarm. -- rozu

Shiz is for some time the best micro-melee and -teambot now. Maybe even the best overall microbot. --GrubbmGait

I think Lib is clearly a better melee nano than DoctorBob, I left it as best overall because few nanos are entered in both competitions. Shadow is on top of Aleph in both team and melee, but I'm not sure it's clearly better in melee. --Alcatraz

I think that Aleph is better at melee. The reason Shadow does about as well in the ratings is because it's a much stronger 1-v-1 bot, so when it gets down to the last 2 bots alive, it prevails more than Aleph does. Another way of saying this is that if Aleph's 1-v-1 movement and gun were as good as Shadow's, it would be significantly ahead of Shadow in the melee rankings. --David Alves

That's highly debatable. The 1on1 performance is important, but at the end of a melee battle the most important factor is how well you survived the crowd, the energy you got left. More, Aleph is also a very strong 1on1 bot, there are very few bots in the melee rumble that stand a chance against him in 1on1, so Shadows slightly better 1on1 performance only affects battles where they end up fighting each other at the end. Most of the times both Aleph and Shadow end up fighting a HOT bot in the melee rumble, and both can handle that quite well ;). Anyway, in my tests (a test bed of above average melee fighters), Shadow does score better than Aleph in most of the battles. -- ABC

My gut feeling says that Aleph is the better melee-bot. Although the difference is almost neglectible, Aleph has had mostly the upperhand in the last half year. Currently the ratings are not quite reliable, see f.e. the details of StoneGhost, and those strange results influence the whole ranking. -- GrubbmGait

I'd say it is Shadow. The thing is that I tuned Aleph to perform well in those 35 rounds in the rumble, but as soon as you take some more rounds, say 1000 rounds or something, then Shadow is better. I guess if we would change the 35 rounds to 100 rounds we would see Shadow continuously at the top. Apart from that, I'm really interested to know if this change would result in a more stable ranking. -- rozu

I'm fairly sure that Komarious has the MiniBot 1v1 throne now, but both it and Pugilist are still pretty far ahead of the rest of the pack... -- Voidious

Good show! --David Alves

I just don't feel right about changing any of these myself. -- Simonton

I changed it for you. Wasn't sure about micro because WeeklongObsession leads Thorn by about 3 points, but is behind by 2 losses in the PremierLeague. Nice work with these bots. -- Alcatraz

Please know I mean no disrespect by this, but... I am curious to see how any of the ratings are changed once we resolve the tainted Robocode 1.1.5 ratings. I believe WeeksOnEnd will still be #1 MiniBot, but perhaps by a much smaller margin (and likely not 2100); WeeklongObsession's lead is so small and it also must've run with 1.1.4 or 1.1.5 (because it uses Rules and was run only by Simonton), so it may lose its lead. For now, it's fair to assume the ratings are "right", but it's something in the back of my mind until the Robocode 1.2.6 RoboRumble client is ready. -- Voidious

Yes, Voidious, you are absolutely correct. -- Simonton

Never know, it might actually boost the score more. --Chase-san

I'm coming for you Voidious... ;-) --David Alves

Changed OneOnOne micro to Waylander Melee to Shadow and micromelee to Sprout. Nanomelee is not really decided, and the overall best, well, it's just hard to tell. -- GrubbmGait

Changed OneOnOne micro to Toorkild, OneOnOne to DrussGT, nanooneonone to Neophyte. -- Nat

Robo Home | Changes | Preferences | AllPages
Edit text of this page | View other revisions
Last edited April 9, 2009 4:11 EST by Nat (diff)