[Split from "Battlesim with arbitrary tech" by 'some scoundrel' masquerading as Ashlyn (-- Wumpus)]

m.a@stars wrote on Thu, 02 March 2006 10:11

As a spinoff of the effort, I'd bet some "automation" scripts can be evolved too, to help ppl skirt heavy MM which puts them off the game.

IE a player-assisting AI. This is actually a long way from trivial. Even something as "simple" as automating scouting gets pretty hairy pretty fast - there's a lot of the element of judgement in working out whether you want to send the scouts outwards from your homeworld fast, or cover everything, or when it makes sense to build more scouts, or what areas are important to scout now (which gets confused by diplomatic issues), etc etc etc.

About the only other thing that came to mind which could conceivably be "simple" is something like mineral transferring between planets - and here, the repeating orders stars! already has cover most of the things that I can see being readily automatable.

It might be interesting to have a thread on what kind of stars! tasks might be automatable, and how one could go about doing that. Maybe my imagination just isn't up to the task ,p As is probably generally known, I don't actually *play* much at all, so... (Maybe move this and m.a@stars's post to a separate thread if this goes somewhere interesting?)

[EDIT 5.3.2006 - oops, micha != m.a@stars ]

{Mod edit: fixed quote}

[Updated on: Tue, 07 November 2006 15:04] by Moderator

Report message to a moderator

wumpus wrote on Sat, 04 March 2006 22:54

m.a@stars wrote on Thu, 02 March 2006 10:11 As a spinoff of the effort, I'd bet some "automation" scripts can be evolved too, to help ppl skirt heavy MM which puts them off the game. IE a player-assisting AI. This is actually a long way from trivial. Even something as "simple" as automating scouting gets pretty hairy pretty fast - there's a lot of the element of judgement in working out whether you want to send the scouts outwards from your homeworld fast, or cover everything, or when it makes sense to build more scouts, or what areas are important to scout now (which gets confused by diplomatic issues), etc etc etc.

Um, no. Scouting is actually fun, not just drudgery. Even so, some automation is possible (i.e: checking that your proposed routes cover all possible planets within a 250 ly radius, or sending fresh scouts to planets too long ago explored)

Main automatable/boring tasks:

- Filling all your planetary Qs with your latest designs and/or removing old ones.

- Mineral/pop balancing.

- Ensuring all your fleets get to their assigned destination at the correct speed and/or time. Routing newly-built fleets... Targeting all those 200 pesky skirmishers near your HW...

- Upgrading starbases, calculating packets, buiding minelayers at the borders...

[Updated on: Tue, 07 March 2006 04:13]

Report message to a moderator

m.a@stars wrote on Sun, 05 March 2006 10:40

Main automatable/boring tasks:

I think you underestimate the "human element" that goes into doing a stars! turn well Plus, I think a lot of these problems are fairly difficult AI problems; consider that stars! has been around, more or less unchanged in all but (comparatively) minor details, for over a decade, and there are still a fair few different viable race designs and play styles - testimony, IMO, that the vast majority of situations in stars! have no simple, always-correct solution.

What I have been pondering is an X file analyzer which will warn of *possibly* strange orders or make suggestions; it might spot stuff like sending damaged ships through gates such that they have a vanishingly small chance of surviving (because the player forgot to take the existing damage into account), or targetting an enemy fleet 1LY away from a planet when you may have meant to target the planet itself, or chasing the MT at a speed which isn't going to work out, or setting a fleet to improbable orders (I've slipped up and set layers to "patrol" rather than "lay" a few times, ouch ;P). The scope for actually tampering with the players turn is fairly limited I fear :-/

Another thing that could be useful: an analyzer for finding out things about enemy races:
- Collects data on all the ship designs you've seen and tries to extrapolate tech levels from that
- Tries to guess enemy ship designs based on their weight and the guessed-at tech of their first production yea
- Tries to determine hab, PRT, LRTs, ...
- etc etc
There'd be a few potential snags to deal with - gifted ships being the most obvious - but this strikes me as a managable task.


But back to the matter at hand:

The MM in stars! is time consuming, but doing it "just right", taking all factors into account, is IMO one of the handful of points which can separate a so-so player from a really great one... emulating a so-so player might be doable (but far from trivial), emulating a really good one is probably very hard, if it is even viable at all without a direct tap to the player's brain .

Having said all that... my AI programming experience is extremely limited; if someone feels they have can formally express algorithms for doing some of this stuff well, then I'm all ears You know what kind of information is available to the AI (basically: anything visible normally in stars!, and very little else), so...

In the mid-term, it may actually be easier to write a complete stars! AI, hopefully better than the existing ones, than to write a player assistant - since the AI then has complete control, and can plan and so on without having to take an unknown element - IE the player - into account.


Some specific thoughts on your suggestions:

Quote:

- Filling all your planetary Qs with your latest designs and/or removing old ones.

How does the proposed AI recognise an "old" design that you don't want anymore? How do you prevent it from removing intentionally "old tech" ships you're building? (EG contrast chaff, a cheap sweeper, and a state of the art weapons scrapper; depending on your general tech, these may all be scout/FFs with cheap engine and a single beam weapon...)

Quote:

- Mineral/pop balancing.

I can't even begin to see how to automate that without a lot of configuration possibility by the player. Having an AI doing it for itself could definitely be viable, but doing it for a human whose plans the AI can't evaluate easily at all, sounds Hard. Setting orders and letting the player change them *might* be viable, but I'm not entirely convinced this won't make more work then it saves.

Quote:

- Ensuring all your fleets get to their assigned destination at the correct speed and/or time.

How does the AI know what is "correct" other than by looking at your orders? Which stars! seems (to me) to be perfectly capable of showing and highlighting problems with all on its own. Granted, you could catch some of the "off-b

Report message to a moderator

wumpus wrote on Sun, 05 March 2006 13:18

In the mid-term, it may actually be easier to write a complete stars! AI, hopefully better than the existing ones, than to write a player assistant - since the AI then has complete control, and can plan and so on without having to take an unknown element - IE the player - into account.

With assistant there will be always dumbuser problems, that you dont have with AI. AI has its predictable dumbness. User can be endlessly dumb and usually assume that it is detected and he is warned about his dumbness.
Example:
I made such a 1 in 10 hab IT HG and then did order fleets 1-5 as "automatic scouts". Your AI started to scout with these, seemingly OK but Why it did not say that
"Mayflower #2" got no scanners?
"Stalwart Defender #3" and "Swashbuckler #4" run out of fuel after scouting 5-7 planets?
5 scouts are probably too few for 1 in 10 IT HG?

None of these problems occur with full AI.

Report message to a moderator

Kotk wrote on Mon, 06 March 2006 07:25

Your AI started to scout with these, seemingly OK but Why it did not say that "Mayflower #2" got no scanners? "Stalwart Defender #3" and "Swashbuckler #4" run out of fuel after scouting 5-7 planets? 5 scouts are probably too few for 1 in 10 IT HG? None of these problems occur with full AI.

As I envision it, a decent "assistant" wouldn't have these problems either.

1st, you'd state these were scout ships, hence, automated checks for their scanning capabilities.

2nd, you'd state area to cover, and time to cover it. The assistant would plainly tell you if that was possible or not.

3rd, fuel usage calculations are trivial, even for long trips.

Having the assistant arrange for refuelings or guessing what use a design should have, or even the optimal # of scouts to use is a very different proposition.

Report message to a moderator

Quote:

How does the proposed AI recognise an "old" design that you don't want anymore? How do you prevent it from removing intentionally "old tech" ships you're building?

AI needs not evaluate designs and consider their goodness instead it does MM.
I imagine a "design management wizard" where there are 16 designs as list. Each has orders on side.
Scenario "scrap":
1) user selects a design, choose "scrap" from listbox and presses "manage".
2) AI splits ships out from fleets, renames all these split fleets "AI Scraps it #x", tries to get them to closest orbital, if there are multiple chooses one with least minerals and sets WP order to scrap.
3) Player gets "Done!" messagebox and list of warnings if some such ships have cargo or have more than 2-3 years travel to closest orbitals.
Scenario "stop building":
1) user selects a design, choose "stop building" from listbox and presses "manage".
2) AI removes all such ships from queues only partially built ones are left.
3) User gets "Done!" messagebox and list of warnings where some such ships are partially built.
And so on?

Quote:

- Mineral/pop balancing.

Yes its good task for AI. I have observed that complete lack of pop or mineral management loses game but perfect does not give so major edge.
1) User marks "breeders" from planet list,
2) User Choose "breeder cap" (25%,33%,50%... something else),
3) User Choose algorithm how to decide where to carry pop like (best_germ_first, least_resource_first, worst_hab_first).
4) User marks "AI pop management" fleets from fleet list
5) done.
AI does what it does ... it is always better than nothing. Once AI result is not lot worse than manual management then big part of MM is gone from game.

Quote:

Routing newly-built fleets...

User wants to have 100 new "Woosh Fluffers" 50 new "Marmon Creeners" and 70 new "Bleed Surkers" at "Stinky Socks" in 4 turns. AI does it or replies "impossible".

Quote:

Upgrading starbases

Heh such little something may be left to user too.

Quote:

calculating packets

Maybe just a calculator how many pop and defenses it likely kills?

Quote:

buiding minelayers at the borders...

I vote against any aid to SD! Whoever wants to be that painful opponent deserves to be last to submit and have no personal life.

Report message to a moderator

m.a@stars wrote on Mon, 06 March 2006 11:10

As I envision it, a decent "assistant" wouldn't have these problems either.

Right, but difference is that decent assistant solves such problems while standalone AI has them not.
All i tried to say there that wumpus is right that standalone AI is simpler to make than AI assistant therefore.

From assistant one expects warnings about direct foolishness and statistics + prognoses for discovering long term faults in his own wishes. The checks/statistics/prognoses take code and interaction that is not needed for standalone AI.

Standalone AI just builds N scout fleets of designs X (and fleet composition Y) per turn up to M scouts. All these preset by AI maker, fit with AI races PRTs and LRTs or are calculated from game size/density/playercount/HWminerals. Then it scouts what it can, no regions, no time. No prognose asked "what happens if i use that "change of heart" or that "cotton picker" as scout".

Report message to a moderator

Wow! I was not expecting such interest!

wumpus wrote on Sun, 05 March 2006 12:18

m.a@stars wrote on Sun, 05 March 2006 10:40 Main automatable/boring tasks: I think you underestimate the "human element" that goes into doing a stars! turn well Plus, I think a lot of these problems are fairly difficult AI problems; consider that stars! has been around, more or less unchanged in all but (comparatively) minor details, for over a decade, and there are still a fair few different viable race designs and play styles - testimony, IMO, that the vast majority of situations in stars! have no simple, always-correct solution.

I'm afraid you overestimate the scope of my concept of "automation". It would be closer to your own of "optimization" than to any true "Intelligent Agent"

I don't want to code a full AI (yet). What I want is to automate some checks and actions, the ones the player would/should be doing anyway. A "smart" script could yield a host of warnings, perhaps with proposed solutions. Another kind of script could automate, for example, the introduction on all planetary Qs of your latest "nubian-killer" design, taking care of, say, placing it after autobuild defenses at border planets with less than 100% defenses, but before anything else for the rest of your big worlds, while leaving your small colonies untouched. Weeding out all non-partially-built obsolete designs (as stated by the player) at the same time should be easy, too.

Quote:

What I have been pondering is an X file analyzer which will warn of *possibly* strange orders or make suggestions; it might spot stuff like sending damaged ships through gates such that they have a vanishingly small chance of surviving (because the player forgot to take the existing damage into account)

Or because the player forgot that some designs are heavier than others, yes. That would be one of the main roles of my envisioned "routing" helper.

Also, if we knew about fleet composition, it could suggest fuel xports for damaged or fuel-challenged fleets.

Quote:

or targetting an enemy fleet 1LY away from a planet when you may have meant to target the planet itself,

Trickier, that, but essential, too. I guess you'd need some kind of "popularity" counter for destinations assigned to fleets in the vicinity. Should leave room for allowing skirmishers and such, tho.

Quote:

or chasing the MT at a speed which isn't going to work out,

That's not that frequent (or boring) an event. Still, it would be interesting, as well as interception calculations for common fleets, such as: Do I have the fuel? Will I catch them in one turn? Two? If they change course?

Quote:

or setting a fleet to improbable orders (I've slipped up and set layers to "patrol" rather than "lay" a few times, ouch ;P).

Heh. What about having your freighters "unload" all their cargo in open space because you loaded them manually and forgot to check the intermediate waypoints? I think much of that could be accomplished by assigning a role to each of your designs and letting a dumb beancounter script pore over the whole fleet listing.

Quote:

The scope for actually tampering with the players turn is fairly limited I fear :-/

Limited, hah. My current scripts are just outputting reams of "suggestions" for the player to heed (or not) and manually set up.

Pls explain, if you dare , what kinds of things can be done at the x file level: Changing planetary Qs? Speed adjustments for fleets? Routing destinations for planets? Packet warp speeds or destinations (at launch)? Changing waypoint orders?

Quote:

Another thing that could be useful: an analyzer for finding out things about enemy races: - Collects data on all the ship designs you've seen and tries to extrapolate tech levels from that

That would be cool, yes. A small database on every race of the galaxy.

As a starter, it would be nice to display exactly *when* you noticed for the first time an unknown

Report message to a moderator

Kotk wrote on Mon, 06 March 2006 10:43

Quote: How does the proposed AI recognise an "old" design that you don't want anymore? How do you prevent it from removing intentionally "old tech" ships you're building? AI needs not evaluate designs and consider their goodness instead it does MM. I imagine a "design management wizard" where there are 16 designs as list. Each has orders on side. Scenario "scrap": 1) user selects a design, choose "scrap" from listbox and presses "manage". 2) AI splits ships out from fleets, renames all these split fleets "AI Scraps it #x", tries to get them to closest orbital, if there are multiple chooses one with least minerals and sets WP order to scrap. 3) Player gets "Done!" messagebox and list of warnings if some such ships have cargo or have more than 2-3 years travel to closest orbitals. Scenario "stop building": 1) user selects a design, choose "stop building" from listbox and presses "manage". 2) AI removes all such ships from queues only partially built ones are left. 3) User gets "Done!" messagebox and list of warnings where some such ships are partially built. And so on?

I wouldn't have expressed the concept better myself.

Quote:

Quote: - Mineral/pop balancing. Yes its good task for AI. I have observed that complete lack of pop or mineral management loses game but perfect does not give so major edge. 1) User marks "breeders" from planet list, 2) User Choose "breeder cap" (25%,33%,50%... something else), 3) User Choose algorithm how to decide where to carry pop like (best_germ_first, least_resource_first, worst_hab_first). 4) User marks "AI pop management" fleets from fleet list 5) done. AI does what it does ... it is always better than nothing. Once AI result is not lot worse than manual management then big part of MM is gone from game.

Actually, my current scripts do find the "breeders" by themselves, based on pop growth. They also allow for a "shared" definition of said "breeder cap", i.e: the script suggests and the player vetoes (or not).

Indeed, having the player set up a list of "breeders" and a list of "needers" would make things simpler for the AI.

My algorithm for pop balancing doesn't yet include potential minerals, just expected pop growth. Should be possible to enhance it, of course.

Quote:

Quote: Routing newly-built fleets... User wants to have 100 new "Woosh Fluffers" 50 new "Marmon Creeners" and 70 new "Bleed Surkers" at "Stinky Socks" in 4 turns. AI does it or replies "impossible".

Yesyesyesyes. It would allow you to change plans until you get a "possible".

Quote:

Quote: Upgrading starbases Heh such little something may be left to user too.

The actual upgrade *design*, yes, of course. The actual task of adding it to 100+ planetary Qs is a work for the AI.

As would be warning you that so-and-so hapless colony still has delta-torps instead of the latest armageddons, or an undersized gate instead of your newest any/any.

Quote:

Quote: calculating packets Maybe just a calculator how many pop and defenses it likely kills?

My aim would be to ascertain if packeting is desirable at all. And if so, how best carry it on.

Also, it could be possible to position "defensive" freighters in advance!

Quote:

Quote: buiding minelayers at the borders... I vote against any aid to SD! Whoever wants to be that painful opponent deserves to be last to submit and have no personal life.

Heh. It would be even better to *help* SD and PP becoming more popular, while also giving their rivals the adequate tools to fight them.

Report message to a moderator

Kotk wrote on Mon, 06 March 2006 11:24

[email m.a@stars[/email] wrote on Mon, 06 March 2006 11:10] As I envision it, a decent "assistant" wouldn't have these problems either. Right, but difference is that decent assistant solves such problems while standalone AI has them not. All i tried to say there that wumpus is right that standalone AI is simpler to make than AI assistant therefore. From assistant one expects warnings about direct foolishness and statistics + prognoses for discovering long term faults in his own wishes. The checks/statistics/prognoses take code and interaction that is not needed for standalone AI. Standalone AI just builds N scout fleets of designs X (and fleet composition Y) per turn up to M scouts. All these preset by AI maker, fit with AI races PRTs and LRTs or are calculated from game size/density/playercount/HWminerals. Then it scouts what it can, no regions, no time. No prognose asked "what happens if i use that "change of heart" or that "cotton picker" as scout".

Awww, well, if you can actually build an AI who knows how to calculate these kinds of things.

Me, I can only envision how I could build something ressembling an AI which would use its assistants to gauge if any semi-random wish/goal/method is doable or not or even beneficial enough for it to semi-randomly plunge ahead on it.

But, even your "all-knowing" AI would need some of my "assistants" to carry its wishes out.

Report message to a moderator

m.a@stars wrote on Mon, 06 March 2006 13:22

Actually, my current scripts do find the "breeders" by themselves, based on pop growth. They also allow for a "shared" definition of said "breeder cap", i.e: the script suggests and the player vetoes (or not). Indeed, having the player set up a list of "breeders" and a list of "needers" would make things simpler for the AI.

With real AI (that plays for itself) it is simple to toss together some formula and let to calculate answer to each of the questions. Formulas can never be "ultimate" but "close enough" is OK for AI opponent. OTOH if assistant AI decides lots of things on its own in players turn then player starts to object ("why it tries to carry pop away from my sweet ironium candy world?"). It is simpler to write few formulas into code than to make intuitive UIs for each aspect of marking and prioritizing the breeders/needers. Most simple i can imagine is to select some pop management status for each planet:
"i havent decided"
"it is breeder"
"it is needer until its full"
"it leave alone"
"it is full"

Quote:

The actual upgrade *design*, yes, of course. The actual task of adding it to 100+ planetary Qs is a work for the AI.

For AR following upgrade path of his orbitals is a priority. Orbital means capacity of planet and all "planetary" defense AR got, no much else to build on planet. There the problem is to have minerals. For rest of the PRTs orbital is source of ability to gate forces around and/or to build ships. If it cant build orbital then its too weak planet anyway and only its strategic location may somewhat matter. Now if AI starts to bugger some late game player about his pesky stations with slot of gatlings and slot of wolverine shields ... then it is annoying for player. He does not care. Only 3% of his opponents nubians can kill any theoretically possible station, so why to build it full and waste minerals.

Quote:

Heh. It would be even better to *help* SD and PP becoming more popular.

SD is actually a very popular PRT among experienced players. It takes up to 2 times power to compete with a good SD. Only thing ... the turn file that SD submits is usually ~2 times bigger than that of others. If some major part of that SDs MM can be at least semi-properly automated then games will turn into SD slugfests where most play SD and minority play useful sidekicks to SD.

Report message to a moderator

Kotk wrote on Mon, 06 March 2006 17:10

m.a@stars wrote on Mon, 06 March 2006 13:22 Actually, my current scripts do find the "breeders" by themselves, based on pop growth. They also allow for a "shared" definition of said "breeder cap", i.e: the script suggests and the player vetoes (or not). Indeed, having the player set up a list of "breeders" and a list of "needers" would make things simpler for the AI. With real AI (that plays for itself) it is simple to toss together some formula and let to calculate answer to each of the questions. Formulas can never be "ultimate" but "close enough" is OK for AI opponent. OTOH if assistant AI decides lots of things on its own in players turn then player starts to object ("why it tries to carry pop away from my sweet ironium candy world?").

Because your "Iron candy" world is growing 55k pop a year and you colonised a strategic outpost less than 200ly away and you have absolutely no other breeder at hand, so skimming 30k pop from said "Iron candy" is actually best option?

Pls note than my "production" script could actually place a veto on moving pop and/or minerals away from the best shipbuilding sites.

Quote:

"it is needer until its full" "it is full"

Both of those are automatically handled using pop growth as deciding factor.

Quote:

For AR following upgrade path of his orbitals is a priority. Orbital means capacity of planet and all "planetary" defense AR got, no much else to build on planet. There the problem is to have minerals. For rest of the PRTs orbital is source of ability to gate forces around and/or to build ships. If it cant build orbital then its too weak planet anyway and only its strategic location may somewhat matter. Now if AI starts to bugger some late game player about his pesky stations with slot of gatlings and slot of wolverine shields ... then it is annoying for player. He does not care. Only 3% of his opponents nubians can kill any theoretically possible station, so why to build it full and waste minerals.

You must be talking about actual AI. My proposed "production helper" would indeed confine itself to making sure SB upgrades (if any was proposed) went the smoothest possible way in all suitable worlds, or warn of any problem.

In fact, one of the reasons for suggesting some SB to be upgraded would be relative popularity of the old and new designs.

Quote:

SD is actually a very popular PRT among experienced players. It takes up to 2 times power to compete with a good SD.

All the more reason to evolve tools to at least reduce the MM in doing so.

Quote:

Only thing ... the turn file that SD submits is usually ~2 times bigger than that of others.

You should see some of my JoaT turns.

Quote:

If some major part of that SDs MM can be at least semi-properly automated then games will turn into SD slugfests where most play SD and minority play useful sidekicks to SD.

I don't think it would be easy to evolve an automaton to actually improve an SD handling of offensive/defensive/multilayered minefields. And if it was indeed doable, perhaps some other PRTs would acquire at least some of the benefits of expert SDs?

Alternatively, send SD to the same ban limbo as CA.

By the way, someone should tell the Admiral that my name is not an email address.

Report message to a moderator

Quote:

By the way, someone should tell the Admiral that my name is not an email address.

Or someone should tell you not to use email characters in your names?

Report message to a moderator

BlueTurbit wrote on Mon, 06 March 2006 17:52

Quote: By the way, someone should tell the Admiral that my name is not an email address. Or someone should tell you not to use email characters in your names?

Of course it is not an "email" character!

M.A. @ Stars!BattleBoard

Report message to a moderator

[email

m.a@stars[/email] wrote on Mon, 06 March 2006 11:03] BlueTurbit wrote on Mon, 06 March 2006 17:52 Quote: By the way, someone should tell the Admiral that my name is not an email address. Or someone should tell you not to use email characters in your names? Of course it is not an "email" character! M.A. @ Stars!BattleBoard

m.a@stars
m a@stars
ma@stars
m.a_stars

The program obviously disagrees with you.
So "get with the program"

[Updated on: Mon, 06 March 2006 12:30]

Report message to a moderator

m.a@stars wrote on Mon, 06 March 2006 12:03

Pls explain, if you dare , what kinds of things can be done at the x file level: Changing planetary Qs? Speed adjustments for fleets? Routing destinations for planets? Packet warp speeds or destinations (at launch)? Changing waypoint orders?

Anything the player can do in the UI, and (beyond cheats), no more There are still some fuzzy "unknown" bytes in my analysis of the files, but that list is pretty short now, and would be easy to eliminate entirely once attempts are made to *synthesise* X (or other) files rather than just analyzing them.

m.a@stars wrote on Mon, 06 March 2006 12:03

Perhaps you are in a position to release a "full info" dumper, for strictly "kosher" uses, or at least know if such a marvel can one day exist?

I've been meaning to have a look at freestars! file format with regards to this, haven't gotten around to it yet though Not hard per se, but not exactly much fun either

Report message to a moderator

I want scouts that you click on a distant planet with "Scout To" orders and it automatically fills in the route based on if planet in +-20 degrees that is unscouted and less than 1 year travel go at colonise warp. Else if planet ...less than 2 year etc
All the way to the final planet - and while I am wishing - to draw a red line from the point it runs out of fuel and a red/blue line on any fuel expensive jumps (say over 2 warps above normal speed)

Thats the kind of AI I want - less MM but me still in charge.

Also I want to set 2 waypoint orders (pop/min balancing), one when the ship arrives, and another before it leaves.
I do admit I need a master class in waypoints as others seem to automate far more than I can.

Oh and a sort by "destructive force" tab on fleets view. So I can swiftly see what fleets I should be worried about. (sort by weight isnt really good enough).

See I really dont want an AI to do things for me - just tell me what I need quickly, and remove some of the "make work" clicks.

Oh and a sweep minefields within Xly order, with a sub order saying - if damaged leave it to other ships at you location with same orders unless you are the least damaged in which case do it but send me a message saying you are about to do it.

Oh Oh and a big shiny thing, to stick on the wall

Report message to a moderator

Maybe we could consider submitting this whole issue to some real guys who are actually top-notch investigators on Artificial Intelligence issues?

I mean, there are certainly a few universities out there with professors and lots of alumni waiting for the right challenge to use in their classes as an exercise in Artificial Intelligence programming.

I think our case is complex enough to be a real challenge, and if their combined skills are worth anything, they might even produce something really useful we could integrate into freestars.

Think about it, it might even prove useful to repopularize Stars! among Computer Science students (and professors). If they take the challenge, we might even end up with several different AI competing against each other, and even beating Micha in some Duel Championships

Report message to a moderator

BlueTurbit wrote on Mon, 06 March 2006 18:18

m.a@stars wrote on Mon, 06 March 2006 11:03 BlueTurbit wrote on Mon, 06 March 2006 17:52 Quote: By the way, someone should tell the Admiral that my name is not an email address. Or someone should tell you not to use email characters in your names? Of course it is not an "email" character! M.A. @ Stars!BattleBoard m.a@stars m a@stars ma@stars m.a_stars The program obviously disagrees with you. So "get with the program"

Heh. Go ahead, then. Send me an email.

Report message to a moderator

wumpus wrote on Mon, 06 March 2006 18:43

m.a@stars wrote on Mon, 06 March 2006 12:03 Pls explain, if you dare , what kinds of things can be done at the x file level: Changing planetary Qs? Speed adjustments for fleets? Routing destinations for planets? Packet warp speeds or destinations (at launch)? Changing waypoint orders? Anything the player can do in the UI, and (beyond cheats), no more There are still some fuzzy "unknown" bytes in my analysis of the files, but that list is pretty short now, and would be easy to eliminate entirely once attempts are made to *synthesise* X (or other) files rather than just analyzing them.

Excellent! Any idea when (or how) will that stage be reached?

Quote:

m.a@stars wrote on Mon, 06 March 2006 12:03 Perhaps you are in a position to release a "full info" dumper, for strictly "kosher" uses, or at least know if such a marvel can one day exist? I've been meaning to have a look at freestars! file format with regards to this, haven't gotten around to it yet though Not hard per se, but not exactly much fun either

The fun would come once some "assistant" was written to analyze the extra info and some mass actions could be added to the x file.

About file formats: you could start with something similar to Stars! own excel-like dumps. Fleet and Q and perhaps packet/salvage details could be dumped that way, I hope.

Report message to a moderator

joseph wrote on Mon, 06 March 2006 23:06

I want scouts that you click on a distant planet with "Scout To" orders and it automatically fills in the route based on if planet in +-20 degrees that is unscouted and less than 1 year travel go at colonise warp. Else if planet ...less than 2 year etc All the way to the final planet - and while I am wishing - to draw a red line from the point it runs out of fuel and a red/blue line on any fuel expensive jumps (say over 2 warps above normal speed) Thats the kind of AI I want - less MM but me still in charge.

Sounds good. I've been meaning for a long time to at least try creating optimal scout routes out of nil. Not exactly easy, tho, even if you don't have penscanning.

Could you explain a little bit more about these conditions of yours? Why 20 degrees? Any others?

Quote:

Also I want to set 2 waypoint orders (pop/min balancing), one when the ship arrives, and another before it leaves. I do admit I need a master class in waypoints as others seem to automate far more than I can.

It would depend on what you want to accomplish. Please give some more details.

Quote:

Oh and a sort by "destructive force" tab on fleets view. So I can swiftly see what fleets I should be worried about. (sort by weight isnt really good enough).

That would need extracting the fleet details from the m file. I've been petitioning for those, too.

Quote:

See I really dont want an AI to do things for me - just tell me what I need quickly, and remove some of the "make work" clicks.

That's what I want to accomplish with my "assistants".

Quote:

Oh and a sweep minefields within Xly order, with a sub order saying - if damaged leave it to other ships at you location with same orders unless you are the least damaged in which case do it but send me a message saying you are about to do it.

Hhmm. Tricky, that. Sweeping is more of an art than a science. Everybody has his/her own particular way to do it, even when no skirmishing is taken into account.

Quote:

Oh Oh and a big shiny thing, to stick on the wall

Pls ask someone in the eyecandy department. Here is mostly math and reams of bitpacked ship orders.

Report message to a moderator

NingunOtro wrote on Tue, 07 March 2006 03:22

I mean, there are certainly a few universities out there with professors and lots of alumni waiting for the right challenge to use in their classes as an exercise in Artificial Intelligence programming. I think our case is complex enough to be a real challenge, and if their combined skills are worth anything, they might even produce something really useful we could integrate into freestars. Think about it, it might even prove useful to repopularize Stars! among Computer Science students (and professors). If they take the challenge, we might even end up with several different AI competing against each other, and even beating Micha in some Duel Championships

Heh. Wouldn't that be fun.

Seriously, though, I've always wondered why Stars! (or an Stars!-like clone) is not swamped by AI researchers doing moddings and such. Perhaps the cryptic input/output interface has something to do with it. But once there's a way to reliably feed such an "academic" AI with the full info from the m file, and also to integrate its orders with the x file, I would say Stars!' popularity among university students should soar.

Report message to a moderator

m.a@stars wrote on Tue, 07 March 2006 11:44

Seriously, though, I've always wondered why Stars! (or an Stars!-like clone) is not swamped by AI researchers doing moddings and such.

Some of the Jeff-made AI-s in game play at newbie level, lot of games made significally later have far worse AIs. But still i doubt there is any reason to be interesting for some AI researcher.

Almost all games have "AI players" and "Non-player characters". These parts of software are far from "Artfical Inteligence".
They have no goals. Getting to goals has no choices of strategies. Strategies do not raise subgoals. Getting to subgoals have no choices of tactics. No generated decision trees, no logics, no heuristics. Vegetables not AI-s.
Instead the game AI-s base on simple decisions: IF some-situation THEN some-activity.
Example:
1) IF enemy in hitting distance THEN hit enemy
2) IF enemy outside hitting distance THEN go into hitting distance from enemy
3) IF no known enemy THEN wander in random direction

To compensate such utter sillyness game AI-s often cheat by having knownledge that they should not have, by outnumbering the human players, by having unusual features or skills or by needing less or no time and resources for some expensive activities compared to "Human players" or "player characters".

Stars got low cheating AI. Two cheats it does ...
1)"Expert" AI race has negative RW points,
2) MT trade can be done without meeting MT.

Report message to a moderator

m.a@stars wrote on Tue, 07 March 2006 11:37

Could you explain a little bit more about these conditions of yours? Why 20 degrees? Any others?

+-20 degrees makes 40 degrees. That means joseph uses 9 scouts to scout 360 degrees. each scouts 40 degree sector away from HW, the areas of interest beween scouts do not overlap and so its very simple to program such AI. Joseph wants that user interface probably because he wants to pick what sector first (free) scout scouts, what sectors 2 next scouts (built at turn 1) scout and so on.

Report message to a moderator

Report message to a moderator