[Home]The2100Club

Robo Home | Changes | Preferences | AllPages
People who have written a bot rated 2100 or higher in the RoboRumble@Home general 1-v-1 category. The bot must have participated in at least 1000 battles.

Using pre-packaged data can provide a huge leap in rating points. There are separate listings here for with and without pre-saved data.

Members without pre-loaded data:
Name Flag Bot Date of admission Screenshot
1. Voidious Dookious 1.16 August 30, 2006 [screenshot]
2. David Alves Phoenix 0.857 November 22, 2006 [screenshot]
3. Skilgannon DrussGT 0.7.4 November 8, 2007 [screenshot]
4. Kev Hydra 0.201 November 15, 2007 [screenshot]
5. ABC Shadow 3.77d January 16, 2008 [screenshot]

Members with pre-loaded data:
Name Flag Bot Date of admission Screenshot
1. Voidious Dookious 1.132W August 22, 2006 [screenshot]

Chat

[1] If only it would last until 1000 battles... --David Alves

[longing_for_2100.png] I hear ya, bro... =) It won't be long. -- Voidious

Well, I still want to do it without dirty dirty pre-saved data... but I don't think a rating should be disqualified from anything for using pre-saved data, so there it is. Back to the MovementLaboratory for me. -- Voidious

David, are you going to try to crack 2100 with pre-saved data? I'll gladly remove my entry above if we want to make a truce on that. I'm interested to hear what you guys think, as (like I said) I have mixed feelings about all this. -- Voidious

How about two categories, 1 for 2100 with and one for 2100 without? like the TC --David Alves

Yeah, that seems like a good idea. I've made separate listings above. -- Voidious

Wow, i have been away from robocoding for two weeks and the 2100 barierre has been crumbled. Even though it is with datasaving ;) Congratulations! --Loki

I see absolutely no problem with preloaded data in a bot. Ugluk even has classes representing specific opponents and the most efficient techniques used to beat them. There is nothing unethical about it. It's just a philosophical debate. -- Martin

I agree that it's "acceptable" ... Any competition has an arbitrary set of rules, and it's been agreed previously that it is allowed in the rumble, and there are bots that do it. Still, I personally think it's more fun to compete with learning speed and on equal footing with every other bot when you enter the competition, and if I had a vote for RR@Home 2.0, I'd vote against pre-loaded enemy-specific data being allowed. -- Voidious

It's official, Dookious 1.16 is at 2102.81 with 1073 battles. Congratulations man, very well done. --David Alves

Holy cow, thanks man. You definitely earn some of the credit in the motivation department. =) -- Voidious

Wow, congratulations from me too. Thats definitely very impressive. --Mue

Amazing, and those TC scores! Congratulations Voidious. -- ABC

One week I go on vacation (actually two times 5 days) and then the first member of The2100Club is honoured. Congrats on this once impossible achievement. -- GrubbmGait

Just re-reading some comments on the Dookious page: And btw, the 2100 barrier will not be broken with improvements to DookiSaber, I'd put money on that! I'm not sure even Mue could squeze many more points out of the current top Rumble guns. I sure lost that bet... =) -- Voidious

I'm adding myself quickly before Phoenix drops back below 2100. =) --David Alves

Congrats David! Getting drunk with company is a lot better than getting drunk alone! (even if the other is Voidious ;D -- GrubbmGait

Congratulations!! --Krabb

Congrats from me, too! --Bayen

Amazing job!!! -- Kev

Finally! =) Awesome job, David, you da man. So now you can get start sleeping again? ;) -- Voidious

Thanks everyone. =) The weird thing is it was only 2096 when it reached 1000 battles, I almost pulled it. --David Alves

Phoenix was at .721 when I entered, and Dookious was at 1.27 when you entered? We've had some weird versioning stuff happening lately, man. =) -- Voidious

Whoa, I just saw, Phoenix is hovering above the 2100 mark, with more then enough battles. Its already in the club, but its steady there. 

2  davidalves.Phoenix 0.860  2100.56  2694  6-1-2007:9:4
--Chase-san

DrussiousGT is really close...2098.56... --Starrynte

(Edit conflict) Yeah, it's pretty amazing that Skil's movement is so solid considering how recently he started Robocoding. But I'd expect he would want to break 2100 using his own gun as well, no? --David Alves

I guess that DrussGT really says something about how to write your movement code: if you can completely trash the simple targeting bots, you are guaranteed at least a very good ranking. Of course, many people test with top bots anyway, since, intuitively, it doesn't make a lot of sense to think that doing very well against a bad bot is better than doing somewhat well against a good bot. -- AaronR

If you can't get high rankings against the weak bots, then it means your movement is buggy. Just look at how high BasicGFSurfer ranks, and it doesn't even segment it's movement. It's because it can get high scores against the HOT bots. As you approach 100%, your score increases exponentially. It works like this: if you gain 4% against a bot that you were scoring 95% against, you have decreased the hits it got against you by 80%. If you gain that 4% against a 50% bot, you decrease the hits he got against you by 8%. Thus it follows, your score gain will go up 80%, or go up 8%. This is why being strong against the bottom bots is more important. That's where the points are =). -- Skilgannon

Not sure I quite agree with your math there, Skilgannon. Your rating is adjusted in proportion to how much you beat your expected score against a given bot. If your expected score is 94.9% (based on the difference between your rating and the other bots, and you get 99.9% against it, then you have beaten the expected score by 5% and you get the same benefit as you would if your expected score was 50% and you got 55%. At least that's my understanding from talking to Albert a long time ago. --David Alves

Ok, I've asked this question before, but now I really want to know. Where do we find out *exactly* how the scoring is done in the rumble? -- Simonton

Don't think it's documented anywhere... Read the source code? =) --David Alves

Do we have the source code? -- Simonton

... And this is why we need to rewrite RoboRumble in the RoboWikiRenovation. -- AaronR

WAIT! I saw it somewhere here on the wiki! If I remember or find it again, i'll let you know. --Chase-san

Is this it? RoboRumble/RankingSystem --Chase-san

There is better info here: [ELO]. The roborumble uses an exponential function to calculate the expected score, so the closer to 100 you get, the higher your score goes. According to Kawigi here your score approaches positive and negative infinity as you approach 100 %or 0% of the score, respectively. So what I said before was valid =) -- Skilgannon

OK, now after more than 3 years of robocoding i understand this ranking system... And I don't like it ;/ :) --Krabb

Maybe we should move this discussion to a place where it is not so hard to find... I would hate if, a year from now, everyone is once again going "I know I saw the link for how the scoring system works around here somewhere!" =) -- AaronR

It's interesting that despite this "exponential increase" stuff, especially near 100, the rankings are nearly (or completely) identical to what they'd be if your score was a straight average of your score %'s against each bot. -- Voidious

I'm pretty skeptical of the idea that at 100% score would get you infinity points. Look at all the 99.9% and 0.1% scores in the rumble ... I'd like to see the source, or hear from the person that wrote the source. -- Simonton

(Edit conflict) So either you have to get ridiculously high scores against every bot to notice the effect or... the rumble server code is buggy and never really worked as intended? I don't know what to make of that. -- AaronR

@Skilgannon I agree with you that that is how the ELO system works. About two years ago, I mentioned that on the wiki, and Albert (who wrote the scoring code) told me that that is not how *his* system works. So unless you know otherwise from reading the code, I'm pretty sure it's not exponential. --David Alves

WaveSerpent may be headed for membership... good luck Kev! --David Alves

This gun seems to be helping a bit =). Thanks for providing the tree, Simonton! Another 500 battles.... Nervous times! -- Skilgannon

Yippie!!!!! That makes 3! Now the race is between ABC and Kev...and AaronR if he hurries up! -- Skilgannon

Wow! Great work, Skilgannon! -- Voidious

Number 3. Ever. Impressive. Man, it's going to take a lot of work for me to be first to 2200. -- Simonton

Wowza, very impressive. Great job! --Chase-san

I think getting two bots over 2095 should count :P. Anyway, congratulations! -- Kev.

Whoa, where'd all those points come from? Congratulations! --David Alves

Thanks guys =). For the first 400 battles it was tied with Phoenix, then it dropped to about 2112, then right around 900 battles it dropped down to 2106, where it stayed. And I just sat there, hitting reload every 5 minutes =). And regarding your remark (Voidious) about 2200 being impossible, what IS the score a bot would get if it got 99% against everybody? -- Skilgannon

Great work Skilgannon, congrats! I'll see what I can do to join you, but probably only in a couple of weeks, I'm very busy at work right now, no time for serious robocoding... :| -- ABC

Yeah, same here, too much work to do a lot of Robocode right now. Just wait until the weekend, though... =) -- AaronR

Well done David, you got Firebird in aswell, first 2100 DC bot! --Chase-san

Welcome, Kev! -- Skilgannon

Congrats, Kev! Add yourself to the list. Half the 2100 club is now teenagers, but perhaps ABC will tip the scales again soon. =) -- Voidious

Thanks guys! Would anyone mind putting a screenshot up for me? -- Kev

The once untouchable boundry is now starting to be filled in. Wow, I wonder if we're gonna have to strive for a 2200 club soon. (and to think I haven't even made 2000 yet, me and my laziness) --Chase-san

Prediction: Shadow 3.77d will be above 2100. Congrats in advance. ;-) --David Alves

I wish I was as optimistic, 2102 and falling with 733 battles... -- ABC

I share David's assessment; I predict 2103. Now go play some Wii Sports to take your mind off it while I fire up my client. =) Good luck! -- Voidious

I still have 4 stars left to collect in Super Mario Galaxy, but nothing beats refreshing the rankings page every few seconds. ;) -- ABC

Looks like I did it, cool! Maybe I can get (back) to the top 3 with this kind of small tweaks... -- ABC

Big congrats, dude! So now you can just relax and stop working on Shadow, right??? =) Added that screenshot link for Hydra, too - sorry I forgot to add that ever, Kev. -- Voidious

Thanks Voidious and congrats ABC! -- Kev

Hey, well done! I need to get back to work on DrussGT and Stormrider, it seems ;-) -- Skilgannon

Way to go ABC! I need to make the 2000 then i'll work on catching up. --Chase-san

Thanks all. I still have a few more small tweaks to try, but I think I would need some major change to become a threat to the top 2. -- ABC

Congratulations! Now stop working on it before you threaten my #2 spot! --David Alves

Welcome, Mue! ;-) -- Skilgannon

Robo Home | Changes | Preferences | AllPages
Edit text of this page | View other revisions
Last edited February 28, 2008 12:44 EST by Skilgannon (diff)
Search: