8611c's rage thread

[8611c]

Split from viewtopic.php?f=15&t=33534. (abma)

This will be probably a stupid question, but what other goal (except a pathfinding and creating multi agent system) could I possibly have?

http://www.aiwisdom.com/ai_genrerts.html
Full articles appearently not readable but the titles+descriptions are alone give some ideas.

How hard is it? Do I need to know how to develop simple games in Spring or it is possible to program without this knowledge?

The AI-interfaces use the same terminology as used in game modding. People are only able to write AIs relatively quick when they previously spent years slowly getting familiar with "spring stuff."
Many questions in AI forum are more related to modding (or playing) than about AI. (the rest of dicussion is about bugs*)
I think without some knowledge in spring game modding it is impossible to make an AI.

Questions look trivial but then suddendly without knowing about "spring stuff" or "tricks" of the game are hard to answer.
In this thread someone wants his AI to check if a unit produces resouces. Seems simple enough?
viewtopic.php?f=15&t=32732#p563016
Deciding whether a unit is 'factory', also not as easy:
viewtopic.php?f=15&t=32157#p558064
That is simple examples, hope the idea is clear.

Can I choose any game from Games section or is someone better for my goals?

IMO there are only 2 games worth considering for AI:

1) Kernel Panic
+ simple eco
+ few, simple types of units
+ predictable, reproduceable outcome when units fight.
- It has a "few issues" when used with current engine version. (98.0)
viewtopic.php?f=43&t=33512#p569854
It is not actively played: it would be up to you to find out how much those issues matter or not.
Using old engine (95.0) is imo not an option because it is 2 years old and there have been bugfixes to AI interfaces.*


2) Balanced Annhilation
+ known to work, because played regularly
+ uses current stable engine
+ gameplay balance stays fixed
- "spring stuff" needed
- when units fight the outcome is influenced by lots of mechanics/physics: which way a turret is turning, projectiles can miss, splash damage, units can heal others, etc.
This influence of many small factors makes it somewhat unpredictable.

Every other game you would not just be AI developer but also a beta tester for the game. Or the game is too complex. (Techa Annihaltion)

3) zero-K (because others mentioned it)
- similiar to BA but even more complex
- uses testversions of engine
- much game-side scripting that is hard for the AI to 'understand', another barrier
After BA it is the second "big" game but if you have no preference either way then for AI imo BA is simpler.


Of all games you can relatively easy make your own version.
For example to remove all the base-building disable the spawning of the builder units, by commenting out one line:
https://github.com/spring/spring/blob/d ... wn.lua#L68
--Spring.CreateUnit(startUnit, x, y, z, facing, teamID)
With the startUnit constructor removed, players/AIs have no way to build stuff.
(Imagine Age of Empires but you start without towncenter, without peasant)

Instead spawn some dozen military units:
for i=1,12 do
Spring.CreateUnit("name of some tank", x+(i*20), y, z, facing, teamID)
end
and you have simple deathmatch, without building.
Seems cumbersome, and it is, but modifying BA in such way is imo still the best way.


* On AI-interface bugfixes see threads such as: "Java AI Interface broken?"
viewtopic.php?f=15&t=27471
A bug that was "fixed for the second time", and then from this comment it does not seem so robust still:
https://github.com/spring/spring/commit ... nt-6834063
Another thread about bugs:
viewtopic.php?f=15&t=32468
What happens if bug is in your AI: viewtopic.php?f=15&t=32439


tl;dr:
I would not be willing to bet something important like school/uni projects or grades on it, because of the many traps.

[PepeAmpere]

I would not be willing to bet something important like school/uni projects or grades on it, because of the many traps.

Do not scare people. Its possible and I encourage everyone to do that. It is just for people with motivation, not for random dudes.

[gajop]

The AI-interfaces use the same terminology as used in game modding. People are only able to write AIs relatively quick when they previously spent years slowly getting familiar with "spring stuff."
Many questions in AI forum are more related to modding (or playing) than about AI. (the rest of dicussion is about bugs*)
I think without some knowledge in spring game modding it is impossible to make an AI.

Factually incorrect.
There are many examples of people like me who made AIs first without any considerable knowledge of modding/making games.
In fact most AI developers haven't really done much modding.

Deciding whether a unit is 'factory', also not as easy:
viewtopic.php?f=15&t=32157#p558064

Read below. It's a one liner with 3 checks. It is easy.

After BA it is the second "big" game but if you have no preference either way then for AI imo BA is simpler.

Although I disagree with some of your other points for ZK, you're correct on this.
BA uses less Lua which is often hard to read from your Java/C++ AI and is therefor easier to understand and predict the game state.

Of all games you can relatively simple make your own version.
For example to remove all the base-building disable the spawning of the builder units, by commenting out one line.
Now instead spawn some squad military units and you have simple deathmatch, without building.

Don't do this. Just simply limit yourself to certain units by deciding what you build. There's no need to modify the actual game itself - it's probably too daunting of a task.

* On AI-interface bugfixes see threads such as: "Java AI Interface broken?"
viewtopic.php?f=15&t=27471
A bug that was "fixed for the second time", and then from this comment it does not seem so robust still:
https://github.com/spring/spring/commit ... nt-6834063
Another thread about bugs:
viewtopic.php?f=15&t=32468
What happens if bug is in your AI: viewtopic.php?f=15&t=32439

Long story short AI more or less works and there are people actively developing working AIs for ZK which you can read here: http://zero-k.info/Forum/Thread/14261

tl;dr:
I would not be willing to bet something important like school/uni projects or grades on it, because of the many traps.

Would also like to note that some people did exactly that and it worked fine - I used it for a minor undergrad class many years before. I think even AF did something AI-related with Shard for his Bachelor/Master thesis, but you would have to check that.

Also +1 on both Pepe's posts.

[8611c]

I would not be willing to bet something important like school/uni projects or grades on it, because of the many traps.

Do not scare people. Its possible and I encourage everyone to do that. It is just for people with motivation, not for random dudes.

My post explained why that was my conclusion.
Imo if somebody wants to use it in real world one should mention the problems too. Otherwise you risk someone's money or grades over the biased view of some interweb project, that is not nice. For hobby however, just do whatever you want.

Deciding whether a unit is 'factory', also not as easy:
viewtopic.php?f=15&t=32157#p558064

Read below. It's a one liner with 3 checks. It is easy.

The point is that these three checks are not as obvious to figure out, for someone not familiar with spring and modding.
To anyone else the difference "builder" VS "factory" or what a "yardmap" has to do with any of this, is a mystery.
(Of course it is easy, once figured out.)
There are hundreds of such things, people slowly picks them up while modding or by playing the games or by carried-over knowledge about the original Total Annhilation.
(the line between playing and modding/debugging is very thin)

:Of all games you can relatively simple make your own version.
For example to remove all the base-building disable the spawning of the builder units, by commenting out one line.
Now instead spawn some squad military units and you have simple deathmatch, without building.

Don't do this. Just simply limit yourself to certain units by deciding what you build. There's no need to modify the actual game itself - it's probably too daunting of a task.

Read below. It's a one liner.
And in this case it really is a simple straight-foward thing: there is no "strange spring stuff knowledge" required. As shown, just swapping out one unit's name for another.
Any other AIs/test opponents will not play by your own special rules, if they only exist in your head.

Long story short AI more or less works and there are people actively developing working AIs for ZK which you can read here: http://zero-k.info/Forum/Thread/14261

I am aware of this.
But I also noticed AI authors often first have to implent/fix something in the interface, or investigate some bug before they can work on their AI.
The author you linked is perfect example of that.
https://github.com/spring/spring/pull/182
https://github.com/spring/spring/pull/167
viewtopic.php?f=15&t=32827&p=563842#p563842
https://github.com/spring/spring/pull/124
viewtopic.php?f=12&t=32450&p=560575#p560575
viewtopic.php?f=15&t=32365 -> https://github.com/spring/spring/pull/118
...
Many cases where truely new people have similiar question about AIs, they just accept it as "bug" or "spring strangeness" and then they simply never post again.
Not everyone is willing or able to do such "side projects" or has time for it, that has little to do with "motivation."
Can you guarantee not to encounter some bug/missing feature at worst moment before your uni deadline?

[8611c]

Pro tip: If you get told to "chat to people" at so early stage then it that is disguised way to say that tutorials & documentation are insufficient.
Applies irl too.
So one should avoid that because it might scare people away.

warned user for this post. its not constructive just agressive. Felony 1 + 2 -- abma

User was banned - mandatory minumum for multiple warnings in a short time period.

[Anarchid]

I think it's safe to conclude that "8611c" is not one of those people you should be talking to if you want to achieve anything productive.

(And is in fact someone with zero experience in AI)

[abma]

i've split to a new topic, imo its really just a rage/trolling.

also side note: i've made an ai which sucked basicly worked without ANY MODDING KNOWLEDGE.

knorke, you've made so many cool stuff, why do you linger with such many senseless posts?

[8611c]

(And is in fact someone with zero experience in AI)

I did make various Lua AI things and dabbled with the Java AI crap enough that it "basically worked."

Did anyone actually try to read the posts? Appearently not.
And yes, the "advice" to "ask someone" is imo just a bandaid for lack of documentation. To me that this behaviour has become a red flag alarm signal in OS.

[FLOZi]

A shame that it came to this, but seemingly inevitable with his current attitude.

[lamer]

Some dev wars going on?
I didn't see trolling (except too many links to solved threads?). It was just another point of view - safe one.

Can you guarantee not to encounter some bug/missing feature at worst moment before your uni deadline?

[hokomoko]

Can you guarantee not to encounter some bug/missing feature at worst moment before your uni deadline?

If you can guarantee nothing bad can happen while working on your thesis, you're not researching anything interesting enough.

Obviously this is mitigated by schedules, you start with a small proof-of-concept:
Can I make an AI that produces random units and gives them fight commands to wherever in 1-2 weeks?

Yes - I stay with spring and continue to work with schedules and goals.
No - I find an alternative subject for research.

[Forboding Angel]

i've split to a new topic, imo its really just a rage/trolling.

also side note: i've made an ai which sucked basicly worked without ANY MODDING KNOWLEDGE.

knorke, you've made so many cool stuff, why do you linger with such many senseless posts?

He spent a year trying to get me and smoth permabanned and making us look like the bad guys. Then he finally fucked up and showed his true colors for everyone to see. He's terribly talented and in some cases has some really good ideas, and when he's not being a jerk he's even nice to talk to.

It seems that Das Bruce's joke (which I didn't even know about till months after the fact... I didn't even know why he made the 8611 account in the first place until a LOOOONG time later) caused him to sort of self implode.

I wish he would stop with the nonsense, cause he has tons to offer the community, and while he is a jerk, he's not stupid. Even tho atm he's kinda acting like it :-/

[smoth]

Eh, he more or less succeeded with his efforts. I cannot really come to this forum without anxiety of him lurking archiving then showing up to crush me. However, that anxiety is between my own ears and as the constant moron that I am, I feel like one day we can be friends again. Try not to be too hard on the guy forb, he isn't like us, we say our peace get pissed then go play descent together(when I get back in august... we need to finish that) . I don't know what his deal is but honestly, I don't know what to think. I feel like he killed my motivation last year in his little crusade and also feel that it wasn't only his crusade, that was just the last straw for someone who tried to maintain a very high level of optimism. So again, it may be MY responsibility as in the end, I choose to not get over it but to some degree I do dread having that persona rear it's ugly head and start a bunch of drama again. Probably the reason that I have just been doing story stuff and playing warmachine IRL.

I still believe he could return to his old self or even let his grudge die. Then again, as I said above, that is my perpetual idiotic masochism. The same shit that makes me hope one day argh and I will get along.

[gajop]

This offtopic thread is turning more offtopic.
I don't want to encourage knorke or anyone elses bashing, so locking this one up.

If you have valid concerns about the state of the AI (and hopefully decent knowledge to back it up), make posts in the AI/engine forum.
There is probably some refactoring required at the skirmish AI interface that kloot mentioned, and it's OK to bring that topic up if it'll lead to something productive, but please refrain from just posting complaints with no meaningful ways of resolving them as we're already low on manpower with a lot of known issues out there. Keep it positive.