Home » Stars! 2.6/7 » The Academy » Primary trait of AI races?
Primary trait of AI races? |
Tue, 06 February 2007 01:09 |
|
RobO | | Crewman 2nd Class | Messages: 17
Registered: February 2007 Location: Denmark | |
|
Where can I fidn a description of which AI race has which primary trait, so that I can pick a proper selecton for my testbed?
I've been looking a fair number of places, no luck. I do remember seeing it somewhere, listing the Macinti as CA but a search in these forums gave nothing.
Report message to a moderator
|
|
| | | | |
Re: Primary trait of AI races? |
Mon, 26 February 2007 18:38 |
|
|
Are the LRTs, habs and other settings known for the AI races? I mean, we know the HE is tri-immune, and the AR has ISB ... but do we have full or at least better details?
Report message to a moderator
|
|
| | | | | |
Re: Primary trait of AI races? |
Fri, 02 March 2007 05:29 |
|
Carn | | Officer Cadet 4th Year | Messages: 284
Registered: May 2003 | |
|
Kotk wrote on Wed, 28 February 2007 20:38 | Yes ... that sounds cool idea! I mean to hack the AI-s in Stars! executable to be just slightly better race designs than they are now.
Of course AI anyway plays them weakly ... but still.
|
I also have a game idea for that, if you realy manage to hack it:
The Monster
7 players in medium packed or so. accbbs, no alliance victory, victory by vote or defeat.
And yes the Monster - 8th player is AI with robotoid coding but JOAT, 3immune 20% PGR, 1in700/15/5/25/checked/25/2/25/ 6 cheap/ ISB,OBRM,IFE,RS(?),MA(its an AI after all). Maybe set up starting positions so AI is a bit distance and equal distant from all players, probably in the middle.
Could be fun, probably all people have to ally to survive, but everybody will think about how to avoid wasting to much to stop the AI and when the right time for backstabbing is and whom to ally with for the time the backstabbing starts.
Edit to add:
More fun for the host and more headache for the players by not telling them what the game is about and see how long it takes them to figure out, that there is a real, though stupid monster around the corner.
[Updated on: Fri, 02 March 2007 05:31] Report message to a moderator
|
|
|
Re: Primary trait of AI races? |
Fri, 02 March 2007 05:35 |
|
|
Ooo I like this idea!
I think it'd be fun to be able to tweak the coding a bit too if that's in any means possible - say, have them build decently large fleets or use chaff or something. Probably impossible, but one can dream eh?
Report message to a moderator
|
|
| | |
Re: Primary trait of AI races? |
Fri, 02 March 2007 07:08 |
|
Carn | | Officer Cadet 4th Year | Messages: 284
Registered: May 2003 | |
|
[email | m.a@stars[/email] wrote on Fri, 02 March 2007 12:44]
Provided it has the *minerals* for all those factories and ships. Good mines are still required.
|
25/2/25 mines should do the job even for the AI on most planets. Even with con1 this produces 62.5 kt G per year, so 21 factories per year. 3000 facs so with conc>10 the AI can mine enough germ in 15 years and the needed mines are built with 1in700 and cost 2 in 4 turns. So for 90% of the colonized planets, the AI will have enough factories for the pop in less than 20 years.
With 30 planets colonized(which is by the way probably a lower limit, with such amounts of pop, resources and mierals on HW), that means 27*5500+=148500+ resources in 2440(+ since the AI probably fails to max pop, he will just let it grow, therefore 80% filed worlds are realistic).
[email | m.a@stars[/email] wrote on Fri, 02 March 2007 12:44]
Even so, I wouldn't expect the 100 AMP nubs before the 60s at the very least, certainly not all fused into a single unbeatable fleet.
|
As he has 150k at 2440 and all techs cheap, i'm still certain he will get nub tech before 2450, as with cost 5 facs and cost 2 mines he will put 60% of resources generated before 2440 in research. And research is cheap so getting we26 will need 300k and nubs 400k.
And 100 nubs cost just 60k resources, the minerals will be avaible, so 100 nubs in 2450 might be possible, 2460 is probably the latest, by that time AI will have tech maxed.
And that those nubs fly around in packs of 5 will make life easier, but still it will be disturbing.
After all, if the AI would even remotely know how to balance pop and minerals, how to gather fleets and how to build chaff and how to design remotely sensible nubians(and avoid all other hulls), he would mop the floor with 15 allied players, simply due to having more minerals.
[Updated on: Fri, 02 March 2007 07:09] Report message to a moderator
|
|
|
Re: Primary trait of AI races? |
Fri, 02 March 2007 07:42 |
|
m.a@stars | | Commander | Messages: 2765
Registered: October 2004 Location: Third star to the left | |
|
Carn wrote on Fri, 02 March 2007 13:08 | 25/2/25 mines should do the job even for the AI on most planets.
|
Your earlier post (which I quoted) seemed to disregard improved mines.
Quote: | With 30 planets colonized(which is by the way probably a lower limit, with such amounts of pop, resources and mierals on HW), that means 27*5500+=148500+ resources in 2440(+ since the AI probably fails to max pop, he will just let it grow, therefore 80% filed worlds are realistic).
|
Remember, we're talking about the AIs here, and the only ones capable of substantial spread WILL spread like mad, regardless of other considerations for as long as there's available planets. That means a significant portion of their econ devoted to colonizers, escorts for such and *battling* the opposition, instead of pure growth.
The AI will also launch a lot of packets, regardless of results, which will certainly eat resources as well as minerals. It will also build tons of worthless ships, even if it doesn't need them.
So yeah, its econ could reach 150k by 2450, but I seriously doubt it would actually do so. I'd love to test such an AI-hack, tho.
So many Stars, so few Missiles!
In space no one can hear you scheme! Report message to a moderator
|
|
|
Re: Primary trait of AI races? |
Fri, 02 March 2007 07:57 |
|
Carn | | Officer Cadet 4th Year | Messages: 284
Registered: May 2003 | |
|
[email | m.a@stars[/email] wrote on Fri, 02 March 2007 13:42]Carn wrote on Fri, 02 March 2007 13:08 | 25/2/25 mines should do the job even for the AI on most planets.
|
Your earlier post (which I quoted) seemed to disregard improved mines.
|
I wrote in intial post about game idea:
The Monster
7 players in medium packed or so. accbbs, no alliance victory, victory by vote or defeat.
And yes the Monster - 8th player is AI with robotoid coding but JOAT, 3immune 20% PGR, 1in700/15/5/25/checked/25/2/25/ 6 cheap/ ISB,OBRM,IFE,RS(?),MA(its an AI after all). Maybe set up starting positions so AI is a bit distance and equal distant from all players, probably in the middle.
[email | m.a@stars[/email] wrote on Fri, 02 March 2007 13:42]
Quote: | With 30 planets colonized(which is by the way probably a lower limit, with such amounts of pop, resources and mierals on HW), that means 27*5500+=148500+ resources in 2440(+ since the AI probably fails to max pop, he will just let it grow, therefore 80% filed worlds are realistic).
|
Remember, we're talking about the AIs here, and the only ones capable of substantial spread WILL spread like mad, regardless of other considerations for as long as there's available planets. That means a significant portion of their econ devoted to colonizers, escorts for such and *battling* the opposition, instead of pure growth.
|
In what way it will be an advantage for human players if the AI colonizes and pop drops like mad?
And if he invest a lot of early resources and mins in DDs with we6 phaser, that still could be disturbing, because early could mean 2415 build.
[email | m.a@stars[/email] wrote on Fri, 02 March 2007 13:42]
The AI will also launch a lot of packets, regardless of results, which will certainly eat resources as well as minerals. It will also build tons of worthless ships, even if it doesn't need them.
|
I've never seen the robotoid AI launching packets, therefore i specified, that the AI should use the robotoid strategy. That is probably simple, just hack the robotoid race to the intended values and the AI will play along normal - which probably means using no gates, but they are probably anyway to stupid for that.
[email | m.a@stars[/email] wrote on Fri, 02 March 2007 13:42]
So yeah, its econ could reach 150k by 2450, but I seriously doubt it would actually do so. I'd love to test such an AI-hack, tho.
|
Whether it does certainly depends on what the humans will do, but in a testbed he would probably reach 300 k in 2450.
I think greatest problem for AI will be 512 fleet limit.
...
[Updated on: Fri, 02 March 2007 07:58] Report message to a moderator
|
|
|
Re: Primary trait of AI races? |
Fri, 02 March 2007 08:16 |
|
m.a@stars | | Commander | Messages: 2765
Registered: October 2004 Location: Third star to the left | |
|
Carn wrote on Fri, 02 March 2007 13:57 | I wrote in intial post about game idea:
The Monster
7 players in medium packed or so. accbbs, no alliance victory, victory by vote or defeat.
And yes the Monster - 8th player is AI with robotoid coding but JOAT, 3immune 20% PGR, 1in700/15/5/25/checked/25/2/25/ 6 cheap/ ISB,OBRM,IFE,RS(?),MA(its an AI after all).
|
I guess your later post was just too terse, then, at least for uncaffeinated me.
Quote: | In what way it will be an advantage for human players if the AI colonizes and pop drops like mad?
|
Obviously, extreme waste of resources and pop. True, 20% PRG JOAT will have a lot of pop, but it still needs to learn to send it in decent-sized chunks, with adequate protection, or is still wasted.
And I'm not even considering the easy "tech transfer" popdropping an AI usually gives.
Quote: | And if he invest a lot of early resources and mins in DDs with we6 phaser, that still could be disturbing, because early could mean 2415 build.
|
Well, if the AI could come up with a decent DD design, yeah, but it will never launch a serious attack so early, even if it could.
Quote: | I've never seen the robotoid AI launching packets, therefore i specified, that the AI should use the robotoid strategy.
|
Lucky you. The HE AI is a far deadlier packeteer than the PP AI simply due to its better econ and minerals. It will packet anything in sight, whether it is near or far or has full defenses. Even the AR AI will act the same, if perhaps a bit less intensely.
Quote: | That is probably simple, just hack the robotoid race to the intended values and the AI will play along normal - which probably means using no gates, but they are probably anyway to stupid for that.
|
No AI uses gates, even the AR or PP.
Quote: | I think greatest problem for AI will be 512 fleet limit.
|
I don't remember the AI going full-bore on shipbuilding even in Huge universes when it was under serious attack. If memory serves, one fleet per planet seems about their usual norm. That was of course the HE or AR AI, with less than impressive resources-per-planet, so a "JOAT" might push things a bit farther...
Now if only we knew the relevant bits to hack and give the poor beastie a try...
So many Stars, so few Missiles!
In space no one can hear you scheme! Report message to a moderator
|
|
|
Re: Primary trait of AI races? |
Fri, 02 March 2007 09:33 |
|
Carn | | Officer Cadet 4th Year | Messages: 284
Registered: May 2003 | |
|
m.a@stars wrote on Fri, 02 March 2007 14:16 |
Quote: | In what way it will be an advantage for human players if the AI colonizes and pop drops like mad?
|
Obviously, extreme waste of resources and pop. True, 20% PRG JOAT will have a lot of pop, but it still needs to learn to send it in decent-sized chunks, with adequate protection, or is still wasted.
And I'm not even considering the easy "tech transfer" popdropping an AI usually gives.
|
With 20% and 3i i'm not worried, that AI performance will be hindered by wasting pop. And pop dropping a planet, which doubles pop every 3 years is not a cheap way to transfer tech.
Tech will be anyway geined fast due to fast forming alliances.
m.a@stars wrote on Fri, 02 March 2007 14:16 |
Quote: | And if he invest a lot of early resources and mins in DDs with we6 phaser, that still could be disturbing, because early could mean 2415 build.
|
Well, if the AI could come up with a decent DD design, yeah, but it will never launch a serious attack so early, even if it could.
|
Quantity has a quality of its own.
Either the AI "wastes" a lot on research and buildup, then Nubs come 2450 and thats early enough to cause problems or AI "wastes" a lot of resources by sending DDs earlier than 20.(if he starts later, it will not affect his economy.)
m.a@stars wrote on Fri, 02 March 2007 14:16 |
Now if only we knew the relevant bits to hack and give the poor beastie a try...
|
Yes.
Wasn't there some guy who found in 5 minutes the code piece for instaforming?
Report message to a moderator
|
|
| | |
Re: Primary trait of AI races? |
Mon, 05 March 2007 14:22 |
|
iztok | | Commander | Messages: 1210
Registered: April 2003 Location: Slovenia, Europe | |
|
Hi!
Question wrote on Mon, 05 March 2007 19:06 | Robototoid = Absurd, colonizes over 3/4th of the galaxy by itself, if you cant kill it early on, dont try.
Turnindrones = Spams minefields everywhere and destroyer fleets, colonizes a decent amount of planets, but does little else.
|
Both are good opponents if you testbed an AR. Turnindrones (SS)force you to start building defensive ships quite early, and are with its 75% cloaked fleets of Beta DDs a real nuisance for an early game AR, esp. a 3-immune 6-7% one.
For killing Robotoids (HE) use cap-ship missiles, as they use LOTS of armor, only modest shielding and virtualy no jammers. Early jihad CCs will have a long lifespan against that AI. Only after it starts fielding MetaMorphs with 2k armor you'll need jugg/doom BBs. Destroying their planets is also way to easy, as they're constantly sending only small fleets after your attack fleet(s), instead grouping them in two-three large fleets.
BR, Iztok
Report message to a moderator
|
|
|
Goto Forum:
Current Time: Mon Jun 03 16:31:55 EDT 2024
|