Botters of the Galaxy multiplayer challenge discussion


#1

Feel free to discuss about the Botters of the Galaxy challenge created by @AntiSquid, @Illedan and @Wildum

Wood leagues have been added to introduce the different leagues more smoothly.


#2

Well …

I hate wood leagues. It’s not proper to BotG. I always hate wood leagues for multi player puzzles. You can win the contest if you want. When the multiplayer start, your AI will crash in wood league.

Ok, no big deal after all, just save your wood codes and copy/paste them when a multiplayer puzzle is out.

Yes, no big deal but no one want to do that. Because of that, when a multiplayer puzzle is out, we never see many players of the contest. The top 10 will never be the same because people will come on the multiplayer puzzle, copy/pasting their legendary top 3 code, “oh it crash. i’m out, bye”.

And now we have 6 wood leagues and my wood codes does not even work on them … So well, at the moment the multiplayer puzzle is far worse than the contest because of wood leagues. Can’t say for the rest.

Please, consider letting bronze AI (during the contest) skip the wood leagues. There’s no fun at all in this.


#3

HULK then ATTACK_NEAREST HERO --> Wood 1
Add Deadpool, get to Bronze and enjoy reusing your contest AI.


#4

No matter how long is it to pass wood leagues, it still no fun at all when you already beat it during the contest.


#5

I know, just trying to remove some pain for this time :slight_smile:
And by the way completely agree with your “skip woods for bronze (or better) contest finishers” idea.


#6

But you didn’t, since they are not the same :stuck_out_tongue:
Jokes apart, I understand your point. Still, for all future new players, I believe it’s better.


#7

I agree, multiplayer should be modify/change with the future players in mind. If we have to add ore wood league then it’s for the better.

I also think that the score and agro should be given in the inputs, because there are scenarios where it’s not possible to calculate it.

@AntiSquid, @Illedan and @Wildum is it possible to add these input?


#8

+1 to score in inputs. It’s hard to calculate and it will give more range to the strategies.


#9

+1 to score in inputs…
+1 to enemies alive inputs…
+1 to unit aggro inputs…
+1 to groot state inputs…

To know what is happening is more important than 6 wood leagues… Even after 6 wood leagues, new players will continue lost… Without know how to calculate the score, and the units aggro. Groot and units aggro is really hard to define, once you can’t know what the enemy is doing…


#10

Regarding input complexity, my suggestion to creators: please consider to have a beginner and expert mode for the inputs when developing the next games. Have beginner inputs by default, make the template for it, and if a bot wants advanced inputs, it can output “ADVANCED” or something and get them on next turn. Dealing with arbitrary incomplete inputs for the majority of the games gets very annoying. :slight_smile:

Regarding wood leagues: it would be really easy to fix by having a button to skip the tutorial and go straight to bronze league. That way beginners still get their incremental introduction, and experts can go straight to the real game. That’s a win-win in my book.


#11

I just entered Wood 6 league. There were 6 players there already.
My Submit wins 19 matches and loses 11 matches. Fairly good, above average.
After that it finally puts me at 6th place of 7, even though I had more wins against 4 of the 6 players. Is this working correctly?


#12

Nice idea. We’ll look into having it as a new gold rule, which would solve @BrunoFelthes / @Palmipedus request too.

Possibility:
If you print (as the first heropick) HERONAME;ADVANCED
we give you the extra values during the whole game.


#13

you could add a version number - so if the advanced inputs change, you increment the version number and still have backwards compatibilty.


#14

then it doesn’t need to start in gold only, hm


#15

Observed the same thing, I think it has something to do with players with negative score…


#16

This is the reason my code for life code is not in the arena right now for example. I was only 122th in Legend, so I can’t show off with it and I have too many other things to do right now, so it’s very low on the TODO list to hack myself out of wood.

The point that Magus is making is indeed the most important one, but I find the very existence of reduced rule sets annoying sometimes. While I fully understand the reason for it’s existence, if your goal is to make a very high level AI, then you need to write the program from the ground up with all of the rules in mind. Even small changes to the rules can potentially carry tremendous implications on the design level.

You could just add a skip wood leagues feature that is unlocked if you competed in the related contest or if you have high enough global achievement points in general (I would personally like a combination of everything, like: bronze achievement ranking OR gold league in any multi/contest OR Discipline global rank OR having competed in the related contest [or even always, tbh.]). Just have a message appear:

“You have proven yourself in the past. Would you like to skip the tutorial? y/n”


#17

We would like some feedback from those who actively play BotG

During contest melee hero teams were dominant because tower diving was a bit too easy.

So we decided to buff tower damage. Instead of making tower dive harder it seems to have completely killed it. This wasn’t our intent, we simply wanted to make it harder, while still forcing the ranged bots to find good evasive maneuvers and bush hiding spots.

Since 100 tower damage made diving too easy and 190 completely removes it from the game, we wonder if another change in tower damage would balance out the team comps again.

It’s hard to find optimal values, but 150 tower damage seems more promising.

Instead of straight up changing anything we would like feedback first. Maybe there’s a better solution. Our aim is for hero picks / team compositions to be equally appealing.


#18

I think that 150 is too low, let’s think about a 2x2 fight without Lane units, it is clear that two melee is stronger than two ranged. So, ranged need to stay safe, avoiding All In fights while poke the enemy.

If the two melle control correctly the minions wave, he can dive the tower with four Ally minions against no enemy minions. So, the Ally minions will do 140 damage (35 for each minion), while the tower will do 150… Only 10 damage more…

I think that the melles is not having success, because they are not controlling correct the wave and the own health…

Maybe 190 is too high, but 150 is too low…I think that 170 could be ok…


#19

I disagree with your calculation, even if you push the lane well, there always are minions who will spawn under the tower, so currently it’s more 140 + 190 dmg against 140 dmg.

For the good value of the tower damage I really don’t know, I feel like several things can be improved by ranged comp (if I remember well I had a low win rate against you BrunoFelthes even with a 100-damage tower).

Otherwise, maybe the way bushes work can be changed, I find them really op as the enemy is basically invincible for one turn when he is in a bush you aren’t. For example, it could be possible to Move_Attack a hero in a bush if he goes out, or if you go in the bush.


#20

So, lets do the math, and I will explain what think that is a good wave control.

Each 15 turns, there are a wave spawn. First spawn is at turn 3.
The wave reach the center of the map and start to fight at turn 7.
There is 3 melee minion with 25 damage (My number before was wrong) and 400 health, and 1 ranged with 35 damage and 250 health.
If you do not attack any minion, at turn 14, the first minion will die (1 minion receive 60 damage every turn, others 2 minion receive 25 damage),
At turn 18, another wave will respaw, and another minion will die, you still have 2 minions. One with 125 health, and another with, 250.
You can kill this 2 minions at next two turns. So you will have 6 minions against 4 at turn 20. You can push the aggro of this 4 minions, with hulk shield, kill this minions as fast as possible, with 5 or 6 turns. So, this is turn 26 now, and you have 6 minions, the next wave spawn is at turn 33, so, you have 7 turns to fight without enemy minions. And you will have 2 ranged (70 dmg) + 5 melee (125 dmg), a total of 195 dmg.

So, you have 195 dmg from minions, plus 50 more base damage from your hero (if you are using hulk + deadpool), plus 1.055 health (against doctor + iron man).

But, this is a bad wave control. It is too simple
Who play MOBA, know about the slow push concept. The strategy is simple, first, you deny your own minions, push the enemy minions to your side, and then, start to push again against the enemy tower, slowly, so, you can stack 2 waves. But no one is doing it… You can use hulk shield and deadpool invisibility, to push enemy minions aggro, so they will not damage your minions when you start to push again…

So, with a simple wave control, you will have, for 5 or 6 turns, 55 more damage, 1.055 more health, more and better shields and stuns…

Finally, what I think is that who is using melee, just need a better long term strategy to win… not just 10 “ifs” to do an easy “All In”, like a lot of people did during the contest…

PS.: I was very happy when I read my name in your post mortem. I spent a lot of time creating an effective defense strategy, It was not easy, and probably I have more strategy to defend my self, than a lot of people that was writing simple dive strategies that was very effective.

Link to a replay with WAIT bots:

And a basic sample stacking first 2 waves: