Bots and Simulations > Bot Tavern

The F1 League lives again!

<< < (10/32) > >>

Moonfisher:
I just took a look at the list you added in the zip file, very nice to have a larger list of bots BTW, but I noticed the numbers on the bots... is that the order you're running the bots in ?
The numbers don't match the dates where the bots where submitted. When a bot is submitted everyone can read it's code, making it very easy to take bot nr 1 in the league and make a bot capable of beating nr 1 but nothing else, then run the league with your bot first and follow up with the rest... that's not realy much fun, and I think the instructions for running the leagues also state that you need to follow the order in wich the bots where submitted (Just order the topics by the date the topic was created and the order should be clear).
I realy think any new bot has to beat all bots created before it, alteast the 30 in the league, in order to get first place. Otherwise the bot is only holding it's place because it's being protected by a stronger bot, and not because it made it through on it's own. The point is also that people aren't suposed to start off trying to find a way to beat Guardian (Or Etch II now), you gotta work your way up through the league from the bottom, which is a softer learning curve and a lot of fun.

bacillus:
Don't worry about the numbers; it's just my way of keeping track.
What I think you think is that I'm running the league in a KO fashion; I have a BIG grid, and whenever a bot wins a contest, I enter the result into the grid. If the two species form a stable ecosystem, which I have seen once only so far, each contestant gets 1/2 a point. Every bot has 76 contests, that is, faces each other bot in the league. No bias. Was this what you were worried about, or have I misinterpreted?

DAY 12
--------
Finished running Bubbles;  final score 63/76, putting it one point behind the current leader Blue on Blue.
Status: 12.8%

Damn, Teakwondo training is in town today, I'll have to work hard to get anything done today.

Moonfisher:
Yeah was just worried you followed the order of those numbers. But I don't think you need to run all bots against all others, as far as I understand it you can just add the first bot, run the second bot against the first, then run the third against the first and second, asf asf... that's less fights, and when you think about it the results from running all vs all won't nesesarily be better than just running them cronologicaly. It'll take less time and be consistent with the way the leagues have been run up to now, I think if you want to change the way the leagues are run you should realy run it by one of the bot gods first (Since in the end it's their call). It may seem more fair to run all vs all, but in reality I don't think the results would be better, beating bots that aren't in the league is rarely going to be the issue and who's to say that a strong league bot should drop in position because theres some bot outside the league able to beat it, and you would also have a loophole since people can create a bot that won't make it in the league but can beat that one bot they're having problems with then make several bots exploiting the same weakness and pull the oponents bot out of the league (Although I doubt anyone whould be so lame).

Anyway just saying I think you're doing more work than you need to, and you realy need to run any changes to the way you're running the league by a bot god to make sure your results are valid. Would be a shame to spend all that time running the league if the results aren't going to get aproved. Theres a reason someone (Think it was EricL) insisted that you used the features in DB to run the league automaticaly, people like consistent results, if you're going to change the way the leagues are run, I strongly sugest you get it aproved by someone "official" if you expect the results to count as a valid league update. Even running your method you should still use the automated setup for running F1 with the league table and all that, to make sure all settings are correct (You can't trust the regular 5 vs 5 restart match thing, it won't always have the right settings).

abyaly:
These are my worries:
- You're leaving bots out that should be run
- You're doing it manually
- You're scrapping the old league system in favor of yours (it's a challenge league, not a tournament league)
- You're not using the latest version

bacillus:
Adressing those problems:
-I'm not leaving out any bot except those that fit in other leagues; my understanding is that Commander Keen is running F2, where we will take the top 30 of each and condense them into the final result.
-I'm doing it manually because the automatic one has never worked for me
-I only based the system as an educated guess; I don't care if this will never be official, I am ranking the bots to create a platform which can then be challenged
-I'm not using the latest version for consistency; at the rate the new versions are coming out, every bot will have had its contests spread throughout between two and five versions, some such as lionfish about a dozen.
FYI, I was actually doing it the way Moonfisher suggested, which adds up to 2926x5=14630 rounds. It may seem like a lot, but I cover an average of 125 per day.

Besides, if this defies the rules of Formula 1, why not make this the F1.2 league?

DAY 13
--------
Ran first 10 rounds of Cadillus.
Status: 13%

Navigation

[0] Message Index

[#] Next page

[*] Previous page