suggestions

Satanus Inaximasus [310]
2015-04-01 10:53:31
[9 years, 35 days ago]

just thought its time to oppen up a suggestion as splitting category 2 as gone way off topic

so yea i think 3 fights for tourney instead of the 1 fight


 
dix17 [27]
2015-04-01 11:58:56
[9 years, 35 days ago]

how about 100 other new ideas? damm face the truth, you're weak lol


 
Gpof2 [130]
2015-04-01 17:34:46
[9 years, 34 days ago]

I dislike the idea of having anything different than 1 fight, which I stated on the other thread. But I do like the idea of bringing this to a new thread :)


 
ActiveX [275]
Head Moderator
2015-04-02 02:00:39
[9 years, 34 days ago]
Ok let's say Bot A can beat Bot B 1 out of 5 fights.

Each fight has the same probability of Bot A winning - 1/5. However the more fights that are run the more you negate the chances of luck playing a factor & Bot A wins a fifth of the fights.

With 1 fight Bot A just needs 1 lucky roll. With 5 you need to be 'very' lucky & beat the odds 3 out of 5 times. Sure that is still possible, but less likely.

So the question here is do you want an element of luck in the fights or do you just want the strongest bot to win.

Personally I find the strongest bot option stale & boring.

Same principle applies to 3 fights.

I am opposed to any increase in the 1 fight.

I think players are also overstating the 'luck' factor too. Just because a bot is lucky against 1 bot doesn't mean they will win the cat as they still have to contend with the rest of the bots in the category.

So for instance Bot A in my example may have been 'lucky' to get a win against the Bot B, but Bot A still has to beat more bots in the category than Bot B & indeed all the other entrants. Again the probability of this occurence is low.


 
Xploded [175]
2015-04-02 03:01:11
[9 years, 34 days ago]

Doesn't every bot have the "luck" factor?

If so then all bots are equally lucky? So it completely fair?

This doesn't necessarily make luck have no part in determining the winner but It makes every bot have the same equal chance of being lucky.


 
Benny [44]
2015-04-02 03:11:54
[9 years, 34 days ago]

If your bot is good, it's easier to get lucky.


 
Leader2 [85]
2015-04-02 03:40:33
[9 years, 34 days ago]

Atm the Tournament is acting like a league all players fight each other 1 time and the one with the most wins wins. Why change it?

Personally things would only get complicated if you started bringing more fights in. 3/5 is way too much maybe make it so you fight each bot 2 times.


 
Satanus Inaximasus [310]
2015-04-02 04:07:03
[9 years, 34 days ago]

2 fights sucks mare what if you win 1 and lose 1 it's a draw would be better if you had a third fight to show the winner


 
Leader2 [85]
2015-04-02 04:09:31
[9 years, 34 days ago]

Yes but if people want more then 2 is ideal exactly like a league if you draw a one off fight


 
Ender [1]
Administrator
2015-04-02 22:42:51
[9 years, 33 days ago]

I think most people understand this concept, but for those that don't, this is basically a discussion of variance: if you flip a coin 10 times and it comes up heads 3 times, that's not too surprising; if you flip a coin 1,000,000 times and it comes up heads 300,000 times...no it didn't, that's (effectively) impossible. The point is that as you continue to flip the coin, the actual results come closer and closer to the expected results (50% heads). Applying this to bots, increasing the number of battles per pair of bots brings the actual tournament outcome closer and closer to the expected tournament outcome. That is, the best build will (effectively) be guaranteed a win, given enough battles.

So the question is then: how much variance should there be? I made the conscious decision to do one battle per pair of bots because I didn't want the "best" bot to necessarily always win. I think a luck factor is interesting.

I'm mostly standing by my original decision. The variance introduced by the luck factor makes things a little bit more unpredictable, but not so much that there are bots winning that have no business winning; you almost surely still have to have a top-in-class bot to have a shot at winning. Similarly, a weaker bot getting lucky on a single battle against a stronger bot isn't going to change where they place; that's just one of many battles. I'm not closed off to change, but this is something I've thought about it and it feels like the right balance to me.

(The other practical consideration is that the code is not general in the way it treats number of battles per pair of bots. This wouldn't necessarily be too difficult of a change, but it wouldn't be a trivial change either.)


 
olli [270]
2015-04-03 00:55:19
[9 years, 33 days ago]

so good luck for all tourny hunters ^^


 
Nosferatu [278]
2015-04-03 13:26:07
[9 years, 33 days ago]

I prefer the 1 fight system.