Author Topic: The AI 9 through 10 run  (Read 71815 times)

Offline Eternaly_Lost

  • Sr. Member
  • ****
  • Posts: 336
Re: The AI 9 through 10 run
« Reply #90 on: February 29, 2012, 12:23:02 pm »
Yea, I keep trying to figure out a way of not having those massive waiting periods.  I've considered and looked at implementing a "fast forward mode" that would just do a really coarse simulation (like an extreme version of what you can do now with performance-profile="Extremely Low" and +10 speed) but making that more than say 50% faster than Extremely-low (I'm looking for like 500% faster than that or more) would involve the equivalent of a nervous-system-transplant for the game.  We've done those before, but now's not the time.

Maybe some kind of device you could build that when it's not in low power mode it increases all resource gain by 10x (or 100x, whatever) but also increases AIP-over-time, AI reinforcements, wave timer countdown, cpa countdown, exo countup, hybrid spawning and maturation, etc by the same factor.  And maybe have it auto-disable if you have over X ships attacking you or are attacking an AI planet or something like that.  The key would be making sure the speedup applies to all the things that "time passage" is supposed to "cost" (other than minor things like how far AI ships move; if the Avenger is up or something it should probably just suppress this device).  It would probably take a number of releases for you folks to find all the stragglers, but does that sound useful/tenable from the player side?  It'd be involved code-wise but way more feasible than literally speeding up the game that far.


Well that would be useful, it also needs to make the shards a lot faster or something. I play normally on Epic, as the time is perfect for me for everything but when it time to bring shards home. For every shard in my 10/10 game, I basically put my entire fleet and turrets in one system, waited for the traders to fly in (several hours) to build the Armor Inhibitor I needed to hold off the hunter killers that fly in under several hundred shield bearers.  (Mainly so that I could kill them without having to go though all those shields first.). After that is done, it scan, order shard home. Leave desktop running, turn on Tivo with a movie and grab my laptop to have access to my email. It takes that long for a Shard to fly home, and that is on +10 extremely low.

I normally have to throw on a TV show in a normal game, and those I don't have to wait for the traders, but it still is where almost all of my waiting time on epic are, so I don't think I should change my speed setting, as most of the time I am doing stuff somewhere when I am not waiting on those shards.


Offline keith.lamothe

  • Arcen Games Staff
  • Arcen Staff
  • Zenith Council Member Mark III
  • *****
  • Posts: 19,505
Re: The AI 9 through 10 run
« Reply #91 on: February 29, 2012, 12:33:11 pm »
You could also make each home world have no more than one "Really Bad(tm)" spawn.  Like double Raid, AI Eye, or whatever.  Of course 10 should be allowed two such spawns :) .
I don't think I want to put a lot of constraints on what it does there; those are basically the only things keeping GUDare from walking all over 9+ ;)

But stacking an AI Eye on top of that big pile of pain so that you can't even bring in the fleet without an unending spawn of mkV ships...
Have ideas or bug reports for one of our games? Mantis for Suggestions and Bug Reports. Thanks for helping to make our games better!

Offline TechSY730

  • Core Member Mark V
  • *****
  • Posts: 4,570
Re: The AI 9 through 10 run
« Reply #92 on: February 29, 2012, 01:04:37 pm »
Oh, and I almost forgot.

Congratulations on beating these (intentionally) absurdly unfair AI difficulties. Quite an accomplishment, and a testament to your skill.

Dang that was overly ornate, but in any case, congrats!  :)

Offline Mánagarmr

  • Core Member Mark V
  • *****
  • Posts: 4,272
  • if (isInRange(target)) { kill(target); }
Re: The AI 9 through 10 run
« Reply #93 on: February 29, 2012, 01:21:14 pm »
I'm impressed. GUDare is a true master. Or just insanely patient. can't tell. :D
Click here to get started with Mantis for Suggestions and Bug Reports.

Thank you for contributing to making the game better!

Offline Hearteater

  • Core Member
  • *****
  • Posts: 2,334
Re: The AI 9 through 10 run
« Reply #94 on: February 29, 2012, 01:35:27 pm »
I have a feeling the AI may target Netflix in the near future.

Offline keith.lamothe

  • Arcen Games Staff
  • Arcen Staff
  • Zenith Council Member Mark III
  • *****
  • Posts: 19,505
Re: The AI 9 through 10 run
« Reply #95 on: February 29, 2012, 02:04:45 pm »
I have a feeling the AI may target Netflix in the near future.
Whether to prevent its use or to make viewers unable to focus on strategy games, it's difficult to say.
Have ideas or bug reports for one of our games? Mantis for Suggestions and Bug Reports. Thanks for helping to make our games better!

Offline Wanderer

  • Master Member Mark II
  • *****
  • Posts: 1,579
  • If you're not drunk you're doing it wrong.
Re: The AI 9 through 10 run
« Reply #96 on: February 29, 2012, 04:01:36 pm »
You could also make each home world have no more than one "Really Bad(tm)" spawn.  Like double Raid, AI Eye, or whatever.  Of course 10 should be allowed two such spawns :) .
I don't think I want to put a lot of constraints on what it does there; those are basically the only things keeping GUDare from walking all over 9+ ;)

But stacking an AI Eye on top of that big pile of pain so that you can't even bring in the fleet without an unending spawn of mkV ships...
Thanks for all the pats on the back folks, they feel nice, but I personally don't feel I'm up to level yet.  The 9.0 game would have given me the same pain if the mapgen had spawned the same way, it really wasn't the level difference except in early game trying to argue with the Tech II, and one whipping boy malfunction on the cross to III.  Admittedly as I crank up the level the results of the mapgen are worse, but it's the same net effect.

Keith, regarding the AI Eye.  Most of my strategy involved getting underneath the raid engine.  Since the Raid engine spawned locally as well it was a moot point.  I just didn't put all the pieces together when I developed the strategy.  It's frustrating to not be able to fully smash the system in one go though when you're fighting double raids, which is why I'd probably just bail if I faced it again.  There's almost literally no way to handle Eye spawn + Raid spawn if it takes too long to get at them.  It'll flush your game.  Personally I'd merely like to see (generically) if Raid engine on planet no Eye.  Ship mass is just too much.  That it's a MK V Eye doesn't help any.  Early game Eye II + Raid would be enough to shutdown a parsec or two.

For waiting on resources, I don't have a good answer.  It's part of the process for the game I'm playing.  I was over-extending constantly against foes with better marks and higher numbers.  The fact that I was able to use a whittling technique at all was pretty powerful.  I might beach-head Cyborg next game with those fortresses to abuse inbounds instead of trying to wait on Kerensky.  Since they don't get the Mil III bonuses, there's no point being that contained with them.  Admittedly they'll die a lot but oh well.

Any option that involves better resources for a higher permanent result (IE: AIP) is not something I'd ever use.  You'll notice I don't ever pop Zenith ship storages, I almost never pop distribution nodes except in VERY early game to get things going.  I usually, near midgame, have about 200+ of each converter somewhere in the back for when I end up in heavy rebuilds of certain items (Raid Starships comes to mind with their crystal costs).  I wouldn't mind something that could reduce wall clock time but I also want to reduce game time.  I rarely play at +10 time, because things get away from me too easily, and I tend to micromanage a bunch of things or use that time to clean up minor problems.  Also, game time is an issue too.

It cost me an hour to have to go deal with the second Rebel Colony, the extra CPA was just icing on the cake.  I could have spent resources on the metal/crystal boosters, but they're just sooooo overpriced on a non multi-homeworld game.  I'd rather have more of just about anything else that will help keep me from having to spend the resources in the first place by being bigger and meaner.  Of course cheapening them basically is a gimme for a multi-homeworld.  Right now they're unfeasible though.  6500 K for what's usually a VERY small gain (what, MAYBE an Econ III's worth?) is not worth a MK III ship.  Ever.  I rarely take Fighters to II and I'd rather upgrade THEM to III on non-multi-homeworld.

For boosts of a non-permanently pain inducing nature, it'll be driven by the result.  What's the point of rebuilding my fleet twice as quick if I have to nearly suicide them on CPAs twice as often?  I know, everything needs a tradeoff, but that's not necessarily one I'd bother with.

Resources are a huge part of the game.  I don't have anything I could recommend as a wise fix. 

On a side note, one of the most frustrating things I ran into was the need for the baiting mechanic.  Oh, I understand not sending 100 ships into a whipping boy to die.  But it's just confusing.  I had 2200 ships sitting on the other side of a wormhole playing poker, MK IV and V, and not entering.  Their firepower was MASSIVE.  They wouldn't budge.  Yet later, with only 500 of those same ships sitting against the whipping boy, they flooded in with a tiny little 400 ship MK III wave.  Against MUCH stronger firepower.  I don't get it.

K-Raiding to insane volumes really allowed me to test a few different strategies, one of which was the Neinzul Factory ships.  That was 18k (6 worlds) worth of knowledge for something that basically just 'came along for the ride' and didn't change the real game outcome, just sped it up a bit, and required an investment to Engineer III to use well.  They're only truly worth it to avoid MK IV factory captures, which you can't avoid with CSGs on anyway, and you're unlikely to HAVE MK IV ships if you spent that K on the builders.

It also let me put up weaker defenses at different points which still were utterly consumed by minor components of the CPAs, to the point where I'm not sure if it was worth it to spend K on that.  I'm still refining my strategy, to be sure, but there were aspects of this game that were significant wastes of resources.

For the Dyson... once you know where it is, just don't keep a scout on it if you're not ready to use it.  It's quieted down and will just happily sit there, as far as I know.
... and then we'll have cake.

Offline Hearteater

  • Core Member
  • *****
  • Posts: 2,334
Re: The AI 9 through 10 run
« Reply #97 on: February 29, 2012, 04:18:25 pm »
I'm really surprised you don't upgrade Fighters to II that often.  I find them excellent cannon fodder.  On an M+C per HP, only the Armor Ship and Shield Bearer beat them (plus the Younglings, Autobombs and Mirrors, which are all special cases).  For the K-cost, Fighter II is a no-brainer for me.  They just absorb too much damage.  Especially early game when your economy is strapped, they can blunt waves without your docks even feeling it.

Offline Wanderer

  • Master Member Mark II
  • *****
  • Posts: 1,579
  • If you're not drunk you're doing it wrong.
Re: The AI 9 through 10 run
« Reply #98 on: February 29, 2012, 04:29:29 pm »
I'm really surprised you don't upgrade Fighters to II that often.  I find them excellent cannon fodder.  On an M+C per HP, only the Armor Ship and Shield Bearer beat them (plus the Younglings, Autobombs and Mirrors, which are all special cases).  For the K-cost, Fighter II is a no-brainer for me.  They just absorb too much damage.  Especially early game when your economy is strapped, they can blunt waves without your docks even feeling it.

Oh, I will occassionally, but a lot of times it'll depend on my extra ships.  In the early game I usually need bombers.  It never fails that some wise guy decides to drop a fortress on my head.  That last game I won with Fighters at Mk I only.  I disagree with damage absorbtion, though.  I know pound for pound you're correct, intellectually.  What I see on screen though never makes me feel like they're worthy, particularly for my extremely limited early game K.  There's just too many other things for my K to do in the early game, like bolster my Economy for stronger ships.  At 2500 K that's half my Econ III.  It's an entire world worth of K to upgrade to MK II.  I could upgrade to Flagships and get a 1.7 boost for my better ships for the same K price.

If my enemy has unlocked something early and vicious with polycrystal armor, I'll unlock them.  But they're definately a seasonal thing.  I don't want cannon fodder.  I want staying power.  In particular on offense when I can't replace them, the K feels significantly wasted.
... and then we'll have cake.

Offline Wanderer

  • Master Member Mark II
  • *****
  • Posts: 1,579
  • If you're not drunk you're doing it wrong.
Re: The AI 9 through 10 run
« Reply #99 on: February 29, 2012, 04:40:21 pm »
So, 9.6/9.6 game started up.  Mapgen did something different again.



Fraggle doesn't have my ARS, but Huge and Dyson do.  Dyson's also pretty well dug in.  There's a fortress on Shawshank making life into the backfield annoying.  There's also an attritioner on Roulette but it's off the Shawshank gate so no biggie, but there's another Fort (II this time) on Roulette.  Janeway has a troop accelerator that I'm going to need to take out the system relatively early if I don't want to be stalled up in my early game.

Looking east, Twycross is nothing but there's an alarm post on Macross, and Macross is heavily built with posts and ions... and a Co-P, killing any chance of using that as a Dyson connected whipping boy unless I want to soak 20 AIP.  Dyson's also pretty heavily built, and is the only system I've scouted with more than 4 resource nodes on it.  Most of them are 1/1s or so.  I can't use that however.  There's an ARS on Dyson, too, though.  So, two ARS's in easy reach, but I'll have to take down Macross to get to it.  Looks like I'll have to soak that Co-P.

Research so far has been Warp Sensor, Grav Turret, Maw II, Econ II.  Space Planes haven't been enough of a nuisance yet to bother getting Tachyon Turrets up, but they'll be shortly once the economy stablilizes.  After building off the MK I fleet I've started on Starship and Raid lines.  Deflected the first two waves rather easily, though Twycross sent in some 'help' for the second wave.

I'm curious to see what the sphere does now that I've scouted it but don't have a scout on it.  It'll take it a goodly amount of time to dig out of Dyson and onto Macross, so I'll work the back.  Eventually I'll drop a few scouts on it and see what it's done for the last 30 minutes or so.

Assuming the homeworlds are still Earth and Riker, but that might be different.  Up till now the ARS's have always been in the same location.  This should be interesting.

Immediate priorities: Clear the Fortress off Shawshank, take Shawshank, Janeway, and blow Roulette to pieces.  Nerf Fraggle Rock and Twycross (I don't have the firepower to full neuter yet) and then work my way into Huge for the first ARS.  Then look east.
... and then we'll have cake.

Offline keith.lamothe

  • Arcen Games Staff
  • Arcen Staff
  • Zenith Council Member Mark III
  • *****
  • Posts: 19,505
Re: The AI 9 through 10 run
« Reply #100 on: February 29, 2012, 04:54:13 pm »
Personally I'd merely like to see (generically) if Raid engine on planet no Eye.
Yea, that would probably be a positive step.  I need to look at the core guard post seeding in general just to make sure I understand what kind of variance we're looking at.  The raid/cpa posts are supposed to be way more brutal than the other ones, but it's like the other ones don't even matter compared to those ;)  Which may be as it should be

Quote
Any option that involves better resources for a higher permanent result (IE: AIP) is not something I'd ever use.
Yea; for instance taking planets for resource spots isn't something one sees in successful 9+ play, generally :)

What I was talking about was basically saying that you would get the same amount of resources produced for the same amount of "bad stuff happening elsewhere" that you would without the device/option/whatever, you would just get it without having to wait it out in real-time.  In theory the actual game result is exactly the same, and if so I don't know why you'd not use it simply because there happened to be AIP involved (because it would have been involved anyway since you have 1-AIP-every-30-min on, which is good because without it it really is just a matter of whittling down the AI without freeing too much at once).

Quote
I wouldn't mind something that could reduce wall clock time but I also want to reduce game time.  I rarely play at +10 time, because things get away from me too easily, and I tend to micromanage a bunch of things or use that time to clean up minor problems.  Also, game time is an issue too.
Well, I'm not sure what you mean, but the relationship between "game time elapsed" and things like aip-over-time, reinforcements, wave generation, wave countdowns, cpa generation, cpa countdowns, etc... that's all intentional.  If you bust your entire I-IV fleet and your economy cannot support rebuilding it in less than 3000 game seconds rebuilding, that's 3000 game seconds the AI has to counterattack against your weakened forces and reinforce its own position against future attacks.  That seems balanced ok, and is the whole reason there's an m+c system at all, really.  The only thing I'm wanting to change is the necessary correlation between game seconds and how much time you physically have to spend on the game.

Quote
On a side note, one of the most frustrating things I ran into was the need for the baiting mechanic.  Oh, I understand not sending 100 ships into a whipping boy to die.  But it's just confusing.  I had 2200 ships sitting on the other side of a wormhole playing poker, MK IV and V, and not entering.  Their firepower was MASSIVE.  They wouldn't budge.  Yet later, with only 500 of those same ships sitting against the whipping boy, they flooded in with a tiny little 400 ship MK III wave.  Against MUCH stronger firepower.  I don't get it.
If you have a save of each (or get saves of similar situations in the future) I can interrogate the little buggers and make them show their math.

On the factory ships, they do serve a purpose but really could do with being easier to use.  Cloaking, maybe, though I dunno if that would make it too easy to just have a "spawn my ships here" invincible-mouse-cursor.  On the other hand I wonder about giving them some kind of "warp gate" capability so you can also have your factories produce to them.  But I'm sure the exploits would be strong with that one ;)

Anyway, interesting stuff :)


Oh look, 9.6 already :)  Time to see what happened...
Have ideas or bug reports for one of our games? Mantis for Suggestions and Bug Reports. Thanks for helping to make our games better!

Offline zoutzakje

  • Hero Member Mark III
  • *****
  • Posts: 1,052
  • Crosshatch Conqueror
Re: The AI 9 through 10 run
« Reply #101 on: February 29, 2012, 04:55:35 pm »
it doesn't matter if you leave no scouts on the dyson planet. It will start spewing out ships once you scouted it once, regardless of scouts being present. At least that's what I believe.

Offline Wanderer

  • Master Member Mark II
  • *****
  • Posts: 1,579
  • If you're not drunk you're doing it wrong.
Re: The AI 9 through 10 run
« Reply #102 on: February 29, 2012, 05:08:51 pm »
Personally I'd merely like to see (generically) if Raid engine on planet no Eye.
Yea, that would probably be a positive step.  I need to look at the core guard post seeding in general just to make sure I understand what kind of variance we're looking at.  The raid/cpa posts are supposed to be way more brutal than the other ones, but it's like the other ones don't even matter compared to those ;)  Which may be as it should be
The only other post that is a significant annoyance is the Heavy Beam Guardpost.  It annihilates Raid Starships trying to get at Ion Cannons (and other assundry annoyances, like Orbital Mass Drivers) that are parked next to them.  Add a FF and they're a significant pain.  Otherwise... no, not really.  Neinzul cockroaches can be a problem if you fire one up too early, but that's mostly because your scouts fired up a homeworld defense you had no intention of dealing with yet.

Quote
Yea; for instance taking planets for resource spots isn't something one sees in successful 9+ play, generally :)

What I was talking about was basically saying that you would get the same amount of resources produced for the same amount of "bad stuff happening elsewhere" that you would without the device/option/whatever, you would just get it without having to wait it out in real-time.  In theory the actual game result is exactly the same, and if so I don't know why you'd not use it simply because there happened to be AIP involved (because it would have been involved anyway since you have 1-AIP-every-30-min on, which is good because without it it really is just a matter of whittling down the AI without freeing too much at once).
Because I can use half a fleet to take care of small objectives, like threat removal, while the rest rebuilds.  If I turn that on, the timers could get away from me.  I understand exactly what you're saying and you're right, which is why I don't have a good answer.  I'd like something that could improve my econ but not, at a 1:1 basis, speed up the game clock's repercussions for it.  Something different, maybe.  Otherwise, not sure I'd use it.  It'd be a case by case basis as to what I felt I could handle vs. what I wanted to deal with.

Quote
If you have a save of each (or get saves of similar situations in the future) I can interrogate the little buggers and make them show their math.
Will do.  I'm not sure I've got a save of that one.  I'm quite sure it'll happen again.

Quote
On the factory ships, they do serve a purpose but really could do with being easier to use.  Cloaking, maybe, though I dunno if that would make it too easy to just have a "spawn my ships here" invincible-mouse-cursor.
They need a serious armor/hp upgrade, at the least.  Right now if you don't control the system or have a near-invincible beach-head, they die.  They also could use with built-in engineer support.  Must Have Engineer IIIs (which die to a left-over sneeze) reduces their value in combat scenarios.  Right now their only use is to reduce time to go home and collect reinforcements during near-deep strikes.

Quote
On the other hand I wonder about giving them some kind of "warp gate" capability so you can also have your factories produce to them.  But I'm sure the exploits would be strong with that one ;)
That... could be useful.  Significantly devalue the worth of the MK IV though, which is as an alternate to your Fact IV if you lost it.  Would need a very significant K reduction.
... and then we'll have cake.

Offline TechSY730

  • Core Member Mark V
  • *****
  • Posts: 4,570
Re: The AI 9 through 10 run
« Reply #103 on: February 29, 2012, 05:12:57 pm »
Comments based on what I have seen tracking this topic.

  • Bombers are still a little too "always useful/needed" compared to the other triangle ship types, though it is better now than it used to be (See http://www.arcengames.com/mantisbt/view.php?id=2579 for related discussion)
  • I think that the ships that come in to a WAY defended planet in some cases were special forces ships. There is a chance that a special forces ship will ignore the "is it safe for me to go in and not get insta-deaded" logic when a human planet is in the path they are patrolling
  • On a similar note, right now, the AI seems to be undervaluing turret firepower, meaning ships will gladly throw themselves onto turrets with just a little bit of build-up, and of course just to get insta-deaded by the turrets. (See http://www.arcengames.com/mantisbt/view.php?id=5001)
  • Higher Mk. Neinzul enclave ship builders need more reward than just allowing higher mk builds. I heard a suggestion about giving them higher intrinsic construction rates at higher marks, and I like that idea. Enough of a boost without radically changing how the ship works.
  • The Mk. I Neinzul enclave ship builder by itself pumps out most MK. I ships at a pathetic rate without assistance. A boost in intrinsic the construction rate across the board (including Mk I) is, IMO, almost certainly in order. Additional health may help too
  • For higher Mk. resource gatherers, they need either a nerf in knowledge to unlock, a buff in how many extra resources that gather per mk, or a merging of the two lines (eg, unlocking metal harvester Mk. II comes with crystal harvester Mk. II), or some combination of these (IMO, some magnitude of all of these is in order)
  • Thanks to the conflicting pressures of a raid engine and the AI eye, having both on the same planet seems like an unfair lose-lose scenario. Not allowing AI eyes and raid engines on the same planet I think would be a great idea
  • The current balance of the upcoming changes to the knowledge hacking may make it that the point where hacking is no longer viable due to AI response magnitude may be way too early for the amount of AIP you need to avoid raising in a >9 game (too early is fine, and fits with the whole "unfair" thing that >9 is supposed to be, but if it is too early, it becomes practically impossible, which is not good for <10)
  • Similar to above, but for Superterminals instead of knowledge hacking

That's all for now.

Offline Wanderer

  • Master Member Mark II
  • *****
  • Posts: 1,579
  • If you're not drunk you're doing it wrong.
Re: The AI 9 through 10 run
« Reply #104 on: February 29, 2012, 05:55:23 pm »
45 minutes in and not a lot has happened.  Here's the deal, Huge has a Counterstrike II on it.  I want AIP as low as possible when I trigger it and breaking that down as well as the fort has been painful, particularly with fleet retreats to cover Sushi for waves.  Sushi's starting to build up a turret defense but digging through Roulette has been ugly.

Huge has an AI Eye on it so once I break down most of Roulette I'll just raid the CounterStrike II post and deal with that.  Once that's done I can get econ up on Janeway.  Popping the Warp Gate on Shawshank with that intent was pre-emptive, and I've been paying for that AIP increase with only homeworld.

I lost a good bit of fleet over on Roulette and it's taking a bit to rebuild.  I may have to take Janeway early like it or not, but I'm going to get hung up on the pair of FF IIs that's on it.  I may just raid it out and collect the backwash in Shawshank.

... and then we'll have cake.