Anti stall algorithm
Moderators: hoijui, Moderators
Ah, you haven't added or done anything innovative, you've just progressively tweaked and rebuilt the existing systems to do the same thing + 1 thing, things which should really have been designed for int he first release, things like water support etc... You aint really discussed anything interesting either.
Krogothe though has experimented with strange things and had unusual ideas, for example the groupAI with predictive targeting.
As a result regardless of how much better your AI gets, there will be a point where the returns you receive will get smaller and smaller and you'll have to change your base design, from the one you use now, which is similar to what it was in the last source release in terms of classes and what not, though i dont know about the innards of those functions...
Soon NTai will be at it's last release, and Epic will be started, and it'll lave you all behind. Krogothe knows this, and he's keeping up, that's why he decided to totally scrap KAI and rewrite from the ground up rather than add to the existing KAI. I expect that when Epic is released the only decent competition ti will have is KAI because of this.
At your current rate I expect AAI 0.9 to be the upper boundary if you continue as you have, and OTAI 1.25.
But since NTai releases have a lot more in them than OTAI and AAI, and they occur much more frequently, I myself have ran into such diminishing returns in XE8 development and have recently had to scrap my entire construction system as a result and rebuild form the ground up, and I expect I'll need todo the same to my attack and scouting systems for XE10.
Krogothe though has experimented with strange things and had unusual ideas, for example the groupAI with predictive targeting.
As a result regardless of how much better your AI gets, there will be a point where the returns you receive will get smaller and smaller and you'll have to change your base design, from the one you use now, which is similar to what it was in the last source release in terms of classes and what not, though i dont know about the innards of those functions...
Soon NTai will be at it's last release, and Epic will be started, and it'll lave you all behind. Krogothe knows this, and he's keeping up, that's why he decided to totally scrap KAI and rewrite from the ground up rather than add to the existing KAI. I expect that when Epic is released the only decent competition ti will have is KAI because of this.
At your current rate I expect AAI 0.9 to be the upper boundary if you continue as you have, and OTAI 1.25.
But since NTai releases have a lot more in them than OTAI and AAI, and they occur much more frequently, I myself have ran into such diminishing returns in XE8 development and have recently had to scrap my entire construction system as a result and rebuild form the ground up, and I expect I'll need todo the same to my attack and scouting systems for XE10.
-
Lindir The Green
- Posts: 815
- Joined: 04 May 2005, 15:09
-
Lindir The Green
- Posts: 815
- Joined: 04 May 2005, 15:09
Well, I'm sorta helping to make NTAI, but I don't really think that it is way ahead of the competition or anything. My goal is just to make it way better than the competition at XTA.
And, with my extremely limited knowledge of C++, I don't really have any idea which AI has the most "innovative" structure.
So... I'm pretty much neutral.
edit:
I am kinda annoyed by AAI's closed source-ness; it seems pointlessly competitive and bad for Spring. But I can see why submarine could be annoyed by AF constantly talking about his AI in other AI threads, and always going on about how great and innovative Epic is going to be.
Maybe it will be great, but so far it isn't coded...
And, with my extremely limited knowledge of C++, I don't really have any idea which AI has the most "innovative" structure.
So... I'm pretty much neutral.
edit:
I am kinda annoyed by AAI's closed source-ness; it seems pointlessly competitive and bad for Spring. But I can see why submarine could be annoyed by AF constantly talking about his AI in other AI threads, and always going on about how great and innovative Epic is going to be.
Maybe it will be great, but so far it isn't coded...
Last edited by Lindir The Green on 27 Apr 2006, 15:01, edited 2 times in total.
What side is there to choose since AAI is closed source and done solely by sub? He's not exactly calling for help that way. 
I'm trying to help AF with NTAI currently, but I haven't managed much yet. (Sorry, will go working on it ASAP.)
But I like AAI 0.63 at the moment best, it doesn't use that much processor power and it plays somewhat ok and is somewhat flexible. It still crashes, but not all the time.
Some competition is good, we get different approaches to things and don't suffer group think where some good alternative is shut off by prejudices.
Don't take it so seriously, guys.
I'm trying to help AF with NTAI currently, but I haven't managed much yet. (Sorry, will go working on it ASAP.)
But I like AAI 0.63 at the moment best, it doesn't use that much processor power and it plays somewhat ok and is somewhat flexible. It still crashes, but not all the time.
Some competition is good, we get different approaches to things and don't suffer group think where some good alternative is shut off by prejudices.
Don't take it so seriously, guys.
hmm, Those arent innovative things that you added, since they where already there when you released the first version which totally muddles me as to what you thought it was that I said.
Aside from that, a custom build placement algorithm isnt innovative, and even the build palcement algorithm in the first AAI source release showed heavy influence from JCAI as did many of your systems.
Since then you have not added anything innovative (which is what I said in the first place), but rather you've stabilized, tweaked, rewritten, and fixed support (Mods, Water support, etc). My point wasnt to completely dismiss AAI as nothign special but rather to say that since the first version new releases dont attempt to add anything innovative or creative, just bugfix, tweak, fix support, add one or two minor requests, release, over and over again.
I suppose next you'll lay claim to threat matrixes, or accuse me of saying it (I did suggest ti first, but it was in use long before spring existed).
I think instead that your closed source is simply because your in it for the competition, you havent experimented and released new systems and changed the way AAI behaves drastically for the better, you have discussed anything on the forum save "how do i do xxx with a unit", and I know you've said to the contrary but if you're not going to experiment test out thgisn like that and se ewhat we think then why are you here? You've already got a working skirmish AI, you've even then released numerous more versions of it.
Aside from that, a custom build placement algorithm isnt innovative, and even the build palcement algorithm in the first AAI source release showed heavy influence from JCAI as did many of your systems.
Since then you have not added anything innovative (which is what I said in the first place), but rather you've stabilized, tweaked, rewritten, and fixed support (Mods, Water support, etc). My point wasnt to completely dismiss AAI as nothign special but rather to say that since the first version new releases dont attempt to add anything innovative or creative, just bugfix, tweak, fix support, add one or two minor requests, release, over and over again.
I suppose next you'll lay claim to threat matrixes, or accuse me of saying it (I did suggest ti first, but it was in use long before spring existed).
I think instead that your closed source is simply because your in it for the competition, you havent experimented and released new systems and changed the way AAI behaves drastically for the better, you have discussed anything on the forum save "how do i do xxx with a unit", and I know you've said to the contrary but if you're not going to experiment test out thgisn like that and se ewhat we think then why are you here? You've already got a working skirmish AI, you've even then released numerous more versions of it.
i thought aai was the first ai to do this, but im not sure...AF wrote:hmm, Those arent innovative things that you added, since they where already there when you released the first version
i never had a look at jcai building placement algo - instead of spending hours on figuring out how it works i rather tried to solve the problem myself with pen & paper :)AF wrote: Aside from that, a custom build placement algorithm isnt innovative, and even the build palcement algorithm in the first AAI source release showed heavy influence from JCAI as did many of your systems.
thats why i really like you: you think know everything better than the developer itself. how do you know how much time i spend on experimenting with different solutions for a problem?AF wrote: you havent experimented and released new systems
one suggestion: try running aai v0.2 against aai v0.63 and tell me if there is any difference :)AF wrote: and changed the way AAI behaves drastically for the better
imho there're still plenty of possibilities to improve gaming experience when fighting aai - as long as my studies allow me to spend some free time on coding i try to improve aai as much as possibleAF wrote: but if you're not going to experiment test out thgisn like that and se ewhat we think then why are you here?
I dotn actually.
Submarine, you still dont udnerstand what i said:
What submarine thinks I mean:
Also
submarine thinks I said:
Submarine, you still dont udnerstand what i said:
What submarine thinks I mean:
What i actually mean:Nothing you did was original, ti was in existign AI's before you created AAI
Thus I am nto referring to AAI, but rather every version fo AAI AFTER THE FIRST RELEASE. Now that I've said that I really hope I dont ahve to re-explain myself a third time.You ahvent added anything innovative in the AAI releases AFTER THE FIRST VERSION BUT NOT INCLUDING THE FIRST VERSION
Also
submarine thinks I said:
What i actually said:Nothing has changed to improve AAI only bugfixes tweaks and added support for essential things
Whereas?No new innovative systems have been added to AAI, rather the original systems have been rewritten and changed to work better but they're still there essentially in some essential form, working udner the same architecture, whereas KAI and NTai have repeatedly changed structure to push forward new systems and expand
It took me 20 minutes to fully understand the JCAI building algorithm, write my won class and port ti to NTai then playtest, tweak try again,a ttempt to fix, then discard it, and come to a conclusion as to why it didnt work for NTai but did for JCAI.i never had a look at jcai building placement algo - instead of spending hours on figuring out how it works i rather tried to solve the problem myself with pen & paper :)
Because when I see your releases I dont notice any stark changes save how AAI builds things, which with the system I saw in the last source release could be done without a total rewrite. However you have experimented, how else would you have gotten say LRPC support? Or water support? My statement was in reference tot he type of systems I was mentionign earlier, not brand new systems to add support for other things, not things that where based on I wodner what'd happen fi I did that....thats why i really like you: you think know everything better than the developer itself. how do you know how much time i spend on experimenting with different solutions for a problem?
