• The forum software have been upgraded to the latest version.

    If you notice anything that looks off, or does not work, please let us know.

    For more information, click here.

AoS Army Harmonizer

GreenyRepublic said:
make that shit gluten-free as I'm intolerant, thanks
Click to expand...
lol

ILKAIN said:
do you know anything about helicopters? literally impossible to fly without air lmao....*SMH*
Click to expand...
Clearly he's ascended beyond our reality and has no need for such simply things as physics.

Aginor said:
I'd like to remind everyone once more that personal attacks are not OK in this forum.
Click to expand...
Minor sidenote, are there any rules for this kinda eh, chaos?
 
First off, I would like to make it clear that personal attacks are not allowed on the forum, being direct or indirect. I have sent a warning to those in question.

I expect that everyone behaves from now, else we will be forced to close the thread.

samheim said:
I don't care about Dan or Greeny and what they are telling me they don't know how to do. That's nice for them. But they need to keep their opinions to themselves really.
Click to expand...
If you do not want to receive feedback, why did you create the post in the first place?

Moving back to the topic of the thread, creating a system that automatically balances board game armies is possible, with the technologies we have available. However, no company that creates board games wants to spend the money on developing this (to my knowledge).

By using "AI" (or more accurate, a set of rules) you can have the computer play millions of games while collecting data, and then tweak the army rules before doing it again. After a few times, you will get a balanced set of rules.

To do this, you will need the ability to simulate an actual board game, so the "AI" is able to play the game, which makes collecting data possible.

Secondly, you need to have the ability to pass along the rule set in such a way that it can easily handle changes or specific unique features an army might have.

In short, while it is possible it requires a large amount of work initially, and then after you have the cost of the processing power/hardware used while the "AI" plays/balances the game.

However, if you feel like giving it a try, I would strongly recommend basing the army rules in a language like Lua, and then have the main software import the rules into a base model according to the rule imported. If you don't go this direction, you will quickly have more models than you can easily manage, and for each army added, you will significantly increase the complexity of the software, making every change/refactor from that point a task no one wants to undertake.
 
ILKAIN said:
you have to ban us all....every single one of us has violated rule 1 lol
Click to expand...
mwha, I think most of those would fall under fair use, especially the rules and pointcosts. And AoS obviously doens't have that problem. So I think most people are probably fine.
 
Edit: sorry @The Red Devil our posts overlapped. You are the boss.

ILKAIN said:
you have to ban us all....every single one of us has violated rule 1 lol
Click to expand...
....and not only rule #1 to be honest.
The key part is to _try_ to not violate them.

We are not the police and we don't want to act like them. There is some leeway for everything.

NIGHTBRINGER said:
"3 - Don't link to or post any pictures of a macabre/pornographic nature."

So no Khorne or Slaanesh then? :p
Click to expand...

I know you are joking, but just to help someone who would ask the same in earnest:

The rule mainly refers to real life stuff, but yeah.

Violence is more tolerated than pornography in most parts of the world (and in the Warhammer universe) so the threshold is a bit different, but yes, this is neither PornHub.com nor Rotten.com and we expect users to respect that.

So far this works pretty well most of the time and we would like to keep it that way.
 
Aginor said:
The rule mainly refers to real life stuff, but yeah.
Click to expand...
In short,
  • real boob = not allowed
  • Juan Diaz sculpted daemonette boob = allowed
:cool:

Aginor said:
Violence is more tolerated than pornography in most parts of the world (and in the Warhammer universe)
Click to expand...
So it would appear as though Khorne has a better PR team than Slaanesh.

Aginor said:
this is neither PornHub.com nor Rotten.com and we expect users to respect that.
Click to expand...
Dare I even ask what kind of site Rotten.com is? :confused: I've never even heard of it.

Aginor said:
So far this works pretty well most of the time and we would like to keep it that way.
Click to expand...
That's true. This place is happily pretty squeaky clean of that sort of stuff.
 
NIGHTBRINGER said:
Dare I even ask what kind of site Rotten.com is? :confused: I've never even heard of it.
Click to expand...

A very disgusting one from way back, I didn't check if it still exists. Should probably not have mentioned it...

Edit: phew, seems it doesn't exist anymore, or something is blocking it.
 
The Red Devil said:
Moving back to the topic of the thread, creating a system that automatically balances board game armies is possible, with the technologies we have available.
Click to expand...
Although it's technically possible I don't think it'l result in a good game though. The AI does not care about things like anti-patterns and toxic gameplay. And "explaining" those concepts to an algorithm seems far too difficult. You can of course try to limit it by putting in expert knowledge, but at that point you've already finished writing most of the game yourself. And of course then we haven't even looked at the problem of the simulation yet. Not only do you need to simulate your game, you need to simulate human players (for clarity, the playstyle of human players, not the level of human players). Otherwise you'l just get a game for AI's, which again isn't necesarly fun as even with abstract games it'l rapidly lead to very unhuman playstyles. And that's something that hasn't really be solved yet. Though I think one of those starcraft bots came close-ish a year or two ago. And even if you did manage to create a human-like player you'd run into the annoying fact that there's billions of humans with different playstyles. You won't be able to build that many bots, so you can probably count on some player showing up with some broken combination that you simply haven't covered.

On top of that, even if you'd allow it to learn from other games. You'd run into issues as most games don't speak the same language. So you'd have a massive problem trying to translate everything into a common "language" so the AI can actually learn from them. Even within the same genre.

All in all I'd say it's just nowhere near worth the effort. Even in the case of videogames, where you already have most of what you need, it doesn't really seem worth it until making human-like players becomes trivial. And even then I'd question it's value beyond just being a fun technical excercise.
 
I think that - by far - the hardest part to judge is positioning.
IMO movement is what separates the mediocre players from the very best ones.

Seraphon is a great example. If you look at the Saurus Oldblood on foot: if you play large units of Saurus his command ability can be anything between completely useless and game-winning.
Same for Skinks. How do you judge their "Wary Fighters" ability? If you know what you are doing it will win you games. It certainly has won games for me. In literally everything else Skinks are just bad, but this one makes them viable.
 
Aginor said:
I think that - by far - the hardest part to judge is positioning.
IMO movement is what separates the mediocre players from the very best ones.

Seraphon is a great example. If you look at the Saurus Oldblood on foot: if you play large units of Saurus his command ability can be anything between completely useless and game-winning.
Same for Skinks. How do you judge their "Wary Fighters" ability? If you know what you are doing it will win you games. It certainly has won games for me. In literally everything else Skinks are just bad, but this one makes them viable.
Click to expand...
Mm, well at least the AI can relativly easily learn how to move. However, judging how much value that adds to the game will be amazingly difficult. Especially as the AI can move with pinpointprecision. A human might screw up and move his skinks just too far (or too close) an AI will not. And as long as it can move that well, any movement, no matter how small, will be amazingly powerfull, especially when combined with the AI's ability to look ahead till the end of the game. So it'l basicly move perfectly the entire game. Resulting in vastly different results compared to the value an average human player will get out of it.
 
Canas said:
Mm, well at least the AI can relativly easily learn how to move. However, judging how much value that adds to the game will be amazingly difficult. Especially as the AI can move with pinpointprecision. A human might screw up and move his skinks just too far (or too close) an AI will not. And as long as it can move that well, any movement, no matter how small, will be amazingly powerfull, especially when combined with the AI's ability to look ahead till the end of the game. So it'l basicly move perfectly the entire game. Resulting in vastly different results compared to the value an average human player will get out of it.
Click to expand...
Eventually yes, that's the other end of the spectrum.
But in the beginning (and probably for a long time) the AI would most likely be spectacularly bad.
 
Aginor said:
Eventually yes, that's the other end of the spectrum.
But in the beginning (and probably for a long time) the AI would most likely be spectacularly bad.
Click to expand...
taking 27 turnes to run around the board and breaking cohesion untill it kills it's self from attrition
 
Another issue is that game is played via battleplans. So you not only have to teach the ai to play army vs army in a "realistic manner", but you also have to program all the battleplans.
 
LizardWizard said:
Another issue is that game is played via battleplans. So you not only have to teach the ai to play army vs army in a "realistic manner", but you also have to program all the battleplans.
Click to expand...
and they keep changing
 
Aginor said:
Eventually yes, that's the other end of the spectrum.
But in the beginning (and probably for a long time) the AI would most likely be spectacularly bad.
Click to expand...
You'd be amazed at how fast that goes. There's some cool videos about a car learning to park by drifting. If I remember correctly it only takes like 50 tries before it parks perfectly.
 
Canas said:
Although it's technically possible I don't think it'l result in a good game though. The AI does not care about things like anti-patterns and toxic gameplay. And "explaining" those concepts to an algorithm seems far too difficult. You can of course try to limit it by putting in expert knowledge, but at that point you've already finished writing most of the game yourself. And of course then we haven't even looked at the problem of the simulation yet. Not only do you need to simulate your game, you need to simulate human players (for clarity, the playstyle of human players, not the level of human players). Otherwise you'l just get a game for AI's, which again isn't necesarly fun as even with abstract games it'l rapidly lead to very unhuman playstyles. And that's something that hasn't really be solved yet. Though I think one of those starcraft bots came close-ish a year or two ago. And even if you did manage to create a human-like player you'd run into the annoying fact that there's billions of humans with different playstyles. You won't be able to build that many bots, so you can probably count on some player showing up with some broken combination that you simply haven't covered.
Click to expand...
I think you underestimate the power of machine learning, with the proper system in the back and "AI" can evolve quickly "skill" wise due to the fact it can learn as quick as you throw money at it (i.e. hardware). Basically as long as you just throw enough resources on it, it could play millions of games in a matter of days, and through that, you get data you can use to improve the army rule set for human players.

After all, we are not looking for someone to challenge a human player per say, but to optimize the game rules and make them as fair as possible.
Being able to run X model, Y army against Z over and over again would be faster to do with a system like this in place, than through traditional means. However, the initial cost would be massive, which is why it is not utilized.

This means there is a potential market for software like this, as long as it's modular enough to support other board games easily. However, the market is very niche, and the risk of not earning the investment back is higher than the chance for you earning on the software.

Canas said:
Especially as the AI can move with pinpointprecision. A human might screw up and move his skinks just too far (or too close) an AI will not. And as long as it can move that well, any movement, no matter how small, will be amazingly powerfull, especially when combined with the AI's ability to look ahead till the end of the game. So it'l basicly move perfectly the entire game. Resulting in vastly different results compared to the value an average human player will get out of it.
Click to expand...
This all depends on what you allow the "AI" to do and know, it would be very simple to add a rule on a specific "AI" model where it randomly does a slight misplacement like a human player could do.

LizardWizard said:
Another issue is that game is played via battleplans. So you not only have to teach the ai to play army vs army in a "realistic manner", but you also have to program all the battleplans.
Click to expand...
In the scenario/case we are discussing, this would be a rather trivial matter compared to the other parts. These should be created similar like army rules in Lua (or another side language/format), both to remove complexity away from the main software, but also to make it possible for someone with no knowledge of the actual software to add content for the simulations.
 
You must log in or register to reply here.
Back
Top