Arcen Games

General Category => AI War Classic => AI War Classic - After Action Reports => : Faulty Logic April 18, 2013, 01:04:30 PM

: Ow.
: Faulty Logic April 18, 2013, 01:04:30 PM
70 planets Spokes 1417703512 Fireflies (I like this ship.)

Marauders 4 (apparently this means a lot of pirates)
Resistance 10 (and they still don't show up very often)
Rebellions 4 (haven't survived long enough to see these yet)
Roaming Enclaves 4
Golems 4
Spirecraft 4 (these exos hurt)
Spire Civilian Leaders (Spire Pain Generators)

10/10 Vanilla/Feeding Parasite (Vanilla is actually harder...)

Normal+Champ (who has been more a liability than asset thus far), Schizo, Complete visibility, No Plots, No AIP/time

As of writing this, my first run ended at 4 hours to the first exo (4 hours? that is crazy fast). My closest save is about half an hour back.

So, what does the black box say?

Mainly that SCLs hurt.

00:00:02 Map is relatively friendly, but far from ideal. The two hubs closest to my HW have little of interest around them (just two ARS out of 19 planets), so I need to deal with either crazy AIP or multiple chokepoints.

Unlock Harvester IIIs because I need to do stuff fast. If I had more time (damn those SCLs), one of the station upgrades III would be a better investment, because my planned empire is relatively resource-poor.

I immediately colonize the nearest hub, and repel the first waves without issue, building the turretball, and supporting it with a stream of fireflies.

At half an hour, I conquer an ARS world (mirrors) with pysite. I use the resulting jumpships to pop all the DCs (6), and 5 of those accursed collaborators. I also have two free by the first hour.
Fortress I and Firefly II take care of defence, for now.

Hour two is spent cleaning up the threat, then freeing the remaining leaders. I also colonized a system with a widow golem, which really helps for cleaning threatballs.

Hour three: Clean up the threat, colonize regen golem and a hub system. The guard post changes make the early game much tougher, as well as giving planetary marks meaning again (mkIV posts hurt. A lot.)

3:10 Economy recovering from regen rebuild/nemesis-repelling fortress.

3:15 CPA of 4300 ships declared.
3:30 CPA dealt with. (Widow under champ shields kills 200 ships per run. Martyr kills 2000 ships)

This also marks where my game will start next AAR portion.

3:40 Commence hacking an ARS/cursed system for tackle drones (the ultimate hacking fleetship). It's really easy. Except that when it spawns a spire blade spawner, it isn't a mindless zombie. It's threatfleet.

3:45 Pop warp counter posts on the hacked system. Despite reading as mkII, these send waves 4 hops away.

3:50 Hack complete, system colonized, TDs unlocked, Cursed repairs commencing, and wow that exo counter advanced a lot faster than I thought it would.

4:00 Exo. I just wasn't ready. (Did you increase exo spawn rates in their recent buff?)

I tried a few times, and it always ended in brownout-then-death.


The game got a lot harder. Guardposts, faster, tougher exos, and the lazy changes... it's almost like you don't want people winning 10/10.

Though the AI unlocking blade spawners and infiltrators (each incredibly painful, but in different ways) does not help.

This particular game is interesting: I have strong motivation to both take my time (my freed SCLs) and move as fast as humanly possible (exos).

Finally: Hunter-Killer mkIII? With SC shield V escorts? On the first exo? <Thread title>
: Re: Ow.
: keith.lamothe April 18, 2013, 02:27:24 PM
4:00 Exo. I just wasn't ready. (Did you increase exo spawn rates in their recent buff?)
Indirectly, yes.  Both you and Kahuna have had your exhaust pipes handed you via platter by the first exo as a result, so I'm guessing it's a bit much.  Basically the goal was to bring with-superweapons up to the challenge level you were encountering on without-superweapons.

Anyway, here's the recent changes in numbers:

In 6.016, in the logic for broken-golems-hard, botnet-golem-hard, and spirecraft-hard where every minute (specifically, on the 22nd second of each minute) the AIP is added to the exo counter (thus this controls the frequency/timing of exos, not the size), but similar to CPAs there's an "Alternate Minimum AIP" to make sure these don't ever really leave you alone even on low-AIP :

It used to be:

:
            int minimumEffectiveAIP = 50 + ( Game.Instance.GameSecond / 360 )
But now it is:

:
        int divisor = 360;
        if ( AIDifficulty >= 8 )
            divisor -= ( 120 * ( AIDifficulty - 8 ) ).IntValue;

        int minimumEffectiveAIP = 50 + ( Game.Instance.GameSecond / divisor );

Or to put it in more natural terms, here's how many points it gets per minute, old and new:

HourOld ResultNew Result On Diff 9New Result On Diff 10
1606580
27080110
38095140
490110170
5100125200
6110140230
7120155260
8130170290
9140185320
10150200350

Note 1: 8 is the same as it used to be, and between 8 and 9 isn't much worse, etc.
Note 2: there are also ceilings on this rule (250 on <= 8, 375 on 9, 750 on 10) but generally you're not going to encounter them in a normal-length game.

If for some reason your AIP is higher than the listed number at the time, that's used instead.

Suggestions on alternate numbers are welcome, just bear in mind that this is for balancing all the cheese you can pull off with golems or spirecraft, etc.  And the fact that you have _both_ on hard on 10/10 is basically guaranteed to hurt.  Multiple exo sources = intentional ow.  10/10 = intentional ow.  Multiple exo sources on 10/10 = ... well, I'm guessing you enjoy this for some reason ;)


In 6.014 there was also the change to the special difficulty multiplier (which affects exo size among other things), it used to be:
If Diff < 7, then Diff*0.15
If Diff = 7, then 1
If Diff = 8, then 1.5
If Diff = 9, then 2.5
If Diff = 10, then 5

Now the same is up to 8, but:
If Diff = 9, then 3
If Diff = 10, then 6

(both before and after, midway difficulties are just linearly interpolated, so 7.3 gives a multiplier of 1.15)


it's almost like you don't want people winning 10/10.
Yea it does seem like that doesn't it.


Though the AI unlocking blade spawners and infiltrators (each incredibly painful, but in different ways) does not help.
Yea, that was the RNG coming out swinging.

This particular game is interesting: I have strong motivation to both take my time (my freed SCLs) and move as fast as humanly possible (exos).
I think the exos win that particular tug of war; the SCLs can deal ;)

Finally: Hunter-Killer mkIII? With SC shield V escorts? On the first exo? <Thread title>
Is that how you say "Doom" in your language?
: Re: Ow.
: Faulty Logic April 18, 2013, 03:27:53 PM
Even with a best-case exo loadout (staggered waves on a single choke) and half an hour of preparations (38 mks of spire ram, widow and regen golem) and almost exclusively defensive unlocks, the first exo wins.

When intensities 4 and 4 are much harder than 10 and 6 were, you have a balance issue.

Reset the exo rate to what it was, while maintaining their strength increase (they also got a pretty significant buff WRT starships).

So it will still grow quadratically, will still be an ironclad reason to keep moving, will still kill you/destroy something vital with increasing frequency, and will still require some kind of chokepoint, but it won't be a flat "I win" at hour 4.
: Re: Ow.
: keith.lamothe April 18, 2013, 03:35:30 PM
but it won't be a flat "I win" at hour 4.
But it likes its I-Win button!
: Re: Ow.
: Faulty Logic April 18, 2013, 03:37:18 PM
But it likes its I-Win button!
Nerf the exos, or I pull out the heavy-duty cheese. And the AI does. not. want that to happen  >D
: Re: Ow.
: keith.lamothe April 18, 2013, 03:38:59 PM
Hmm, it saw visions of MkV Riots and Warhead Fortresses, and I think it's reconsidering.
: Re: Ow.
: Faulty Logic April 18, 2013, 03:50:52 PM
More seriously, does resetting the rate while maintaining the strength seem reasonable?
Maybe increase the strength later if it that ends up being too easy?
: Re: Ow.
: keith.lamothe April 18, 2013, 04:06:05 PM
More seriously, does resetting the rate while maintaining the strength seem reasonable?
Maybe increase the strength later if it that ends up being too easy?
I'm thinking of going with:

:
        int divisor = 360;
        if ( AIDifficulty >= 8 )
            divisor -= ( 30 * ( AIDifficulty - 8 ) ).IntValue;

        int minimumEffectiveAIP = 50 + ( Game.Instance.GameSecond / divisor );

Note the "30" instead of "120" on line 3, so:

HourOld Result6.021 Result On Diff 96.021 Result On Diff 10Proposed Result on Diff 9Proposed Result on Diff 10
160658060.9162
2708011071.8274
3809514082.7386
49011017093.6498
5100125200104.55110
6110140230115.45122
7120155260126.36134
8130170290137.27146
9140185320148.18158
10150200350159.09170

In other words, a more subtle (but significant, Diff 10 getting a roughly 17% boost compared to Diff <=8) use of the same rule.  By itself, probably difficult to notice.  Combined with the special difficulty multiplier, pain.  But probably not "I Win".

For reference, before 6.021 the AMP rule meant that the exo counter for those sources would get about 16,800 points in the first four hours (which is still what you get on Diff <= 8 ).  After 6.021 on Diff 10 it would get about 33,600 points in the first four hours.  Under these new proposed numbers on Diff 10 it would get about 20,640 points in the first four hours.
: Re: Ow.
: Faulty Logic April 18, 2013, 04:13:17 PM
Looks good.
: Re: Ow.
: Oralordos April 18, 2013, 04:31:06 PM
Warhead Fortresses
:o
: Re: Ow.
: Winge April 18, 2013, 05:02:05 PM
Warhead Fortresses

That's no moon...
PS:  two words:  scroll right

(http://www.arcengames.com/forums/index.php?action=dlattach;topic=12514.0;attach=6761)
: Re: Ow.
: keith.lamothe April 18, 2013, 05:42:57 PM
No, it is a moon, it's just a moon that happens to also be a nuclear warhead.

Construction was complicated.
: Re: Ow.
: LaughingThesaurus April 18, 2013, 06:20:15 PM
Keith, you don't often construct moons. They're natural.

So imagine that. A naturally occurring moon-sized nuclear warhead. I mean, the galaxy created massive fusion reactors naturally, created invisible matter naturally... why not a nuke?
: Re: Ow.
: Faulty Logic May 02, 2013, 10:42:48 AM
Given the name of the next expansion, I thought I'd give the AI some more reasons to direct that vengeance at me:

70 planet Spokes 265230677 Fireflies

Marauders 4 (apparently this means a lot of pirates)
Resistance 10 (and they still don't show up very often)
Rebellions 4 (haven't survived long enough to see these yet)
Roaming Enclaves 4
Golems 4
Spirecraft 4 (survival possible?)
Spire Civilian Leaders (want to win with these enabled)

10/10 Feeding Parasite/Vanilla (Vanilla is actually harder...)

Normal+champ, comp-vis, schizo, no plots.

And the AI has tackle drone launchers. OW. I have regularly faced groups of 40 of these things. The mk hubI started with 17. These need a nerf in AI hands, otherwise it basically says "no fleetships allowed." Everywhere. My fleet's been small enough to be covered by diplomatic immunity so far, though.

They also have infiltrators, because the AI should always have something with both cloaking and forcefield immunity  ::)

===Operation Collaborator Termination====

I immediately unlock Harvester IIIs, conquer my only neighbor, and put down the full cap of missile turrets. These, fireflies, fighters, and Negotiation take me past the first couple waves.

Conquered the next world before it reinforces, gaining access to pysite. Jumpships (buffed indirectly by the starship changes) and raid/bombers kill the 8 data centers, terminate 5 collaborators, and free 3 civvies by gametime 1 hour.

Pause was used extensively, because I hate time pressure.

Cleaned up threat fallout, conquered next two hubs (one had a co-p, annoyingly) repaired the cursed golem, and I'm set to free the other two civilians with jumpships, though this will require eating two counterwaves. I'll also pop the other 3 co-ps, to keep AIP close to reasonable.

I've also colonized the three civvies I freed, one coming with a black widow, as yet unrepaired.

Fireflies mkI-III really help when you need to move fast.

===Operation Civilian Freedom====
They'd better be grateful.

Wasn't actually that complicated. Unlocked flagship/bomber starship mkII, and freed the remaining two leaders, one at a time, with the youngling stream, cursed golem, and Negotiator handling threat fallout. Pulled back the cursed to deal with counterwaves, and the fleet to deal with a double normal wave.

Golems seem to be made of tissue paper, these days. Maybe it's just that the AI always targets them above anything else. I think golems (and spirecraft, to a lesser extent) need some rebuffing. They don't feel like like superweapons anymore, when they lose to a ~500 mark II ships.

Anyway, I have the last two hubs (one ARS; scapegoats) and last two leaders (one with a hive, one with a core paralyzer fab) colonized by the time the first CPA declares.

It threatballs on the choke, except for ~500 hiding on a core world. Widow runs under champ shield reduce it to mostly TDLs, which are a certifiable nightmare in AI hands.

It's like when the low-cap ships were designed, someone went through what they thought those ships would be vulnerable to, and then gave them all of those immunities.

Hmm, really getting sidetracked today. Stopping here. Gametime 3:19, AIP 98

K unlocks: fortress, harvester IIIs, riotIIs, fireflies II/III, missile turret II, HBC, grav Is, miniforts, hardened ffs I, flagshipIIs, heavy bomber starship IIs, warp jammers. 9750 k unspent.

Captures: Cursed and Widow (active), Hive (unrepaired), 1 ARS (scapegoats), and three fabricators (core etherjets, core paralyzers, core electric bombers).

State of Empire:
: Re: Ow.
: TechSY730 May 02, 2013, 12:30:53 PM
And the AI has tackle drone launchers. OW. I have regularly faced groups of 40 of these things. The mk hubI started with 17. These need a nerf in AI hands, otherwise it basically says "no fleetships allowed."

How the crud did they get 17 of them on game spawn? Did that planet have only 17 of them and like nothing else mobile on that planet? Even with difficulty 10, and the planet "specialization" logic, this seems like something fishy/buggy is going on.

Anyways, I have been wondering about these "spawner" units, and I think I realized something. Due to the nature of spawner units (Spire blade spawners and Tackle Drone Launchers), their net "utility" goes up faster than linear over the number of spawners. This means that the AI paying the properly scaled costs due to their low ship cap, especially past normal cap, still isn't enough; it would still be giving the AI too much and making it pay to little as they get over cap, as each additional "spawner" over cap gives even more return for their cost.
Maybe do something like what was done for the Zombard, increase the "cost" of spawner units for the AI? Yea, that won't fix it truly, as that point where they get increasingly "more bang for their buck" would still happen (it will always happen eventually so long as their count isn't capped AND their cost is constant), but it may delay it to the point where you have a chance to reasonably fight it in the early to mid game if the AI decides to "specialize" lots of planets to use them.


They also have infiltrators, because the AI should always have something with both cloaking and forcefield immunity  ::)

Hooray, the AI getting a good raiding ship. Fun times. :D

Now you might be happy that they get anti-matter ammo. ;)
: Re: Ow.
: TechSY730 May 02, 2013, 02:30:28 PM
Made new thread about the spawner ships in AI hands thing.
http://www.arcengames.com/forums/index.php/topic,12960.0.html
: Re: Ow.
: Faulty Logic May 02, 2013, 07:10:02 PM
How the crud did they get 17 of them on game spawn? Did that planet have only 17 of them and like nothing else mobile on that planet?
Precisely. I don't think 2 caps of mobile ships is a bug, per se. It was certainly a roadblock, though.

Now you might be happy that they get anti-dark matter ammo.
Yeah, spending 1000 k not to worry about them seems reasonable.
: Re: Ow.
: Faulty Logic May 02, 2013, 11:46:12 PM
===Operation Entrenchment====
An exo is coming. I must prepare.

I take two more planets for an artillery golem and nanoswarms.

I repair the Hive.

I unlock military stations III for income and attack boost.

I unlock HBCs II-III.

I do 4 nebulae (not quite enough time to become a cruiser), deploying two mod forts at the choke.

And I build lots of spirecraft.

Gametime 5:10. Exo time.

One big hammer on the choke. Led by an H/K mkIII. My huge group of penetrators insta-kills him. 28 Rams deal with the three enemy golems, my golems, fleet, and fixed defences mop up.

That was easier than I expected. Of course, I did spend about 2 hours just preparing for the exo. I think a mild strength buff is still in order (their rate is good, though).

The last explosions from that battle have barely died, when CPA number 2 declares with 7500 ships. Joy.

Stopping here. State of Empire (not quite current, CPA will declare a couple minutes later):
: Re: Ow.
: TechSY730 May 03, 2013, 12:14:06 AM
Impressive work shutting the AIs stuff down so far.

However, you did just dump a bunch of spirecraft onto it, including some Rams, which you don't get to reuse, and you get a finite number of.

And also, only 5 hours in and you are already getting Mk. III H/Ks escorted by golems in exos, followed by a 7500 ship count CPA!? Ouch!! 10/10 hurts.  :o
: Re: Ow.
: keith.lamothe May 03, 2013, 11:39:31 AM
One big hammer on the choke. Led by an H/K mkIII.
Oh, yeah! (http://blazbaros.deviantart.com/art/Brother-Koolaidius-180110431)

My huge group of penetrators insta-kills him.
Oh, no! (http://www.penny-arcade.com/comic/2000/7/21/)

28 Rams deal with the three enemy golems
You really know how to make the AI feel welcome.



: TechSY730
only 5 hours in and you are already getting Mk. III H/Ks escorted by golems in exos, followed by a 7500 ship count CPA!? Ouch!! 10/10 hurts.  :o
Yea, but

: Faulty Logic
I think a mild strength buff is still in order (their rate is good, though).
He still asks for more.  On the exos, at least.

We'll see about the CPA, but I think we all know where this is going.
: Re: Ow.
: TechSY730 May 03, 2013, 11:57:20 AM
Warhead?

Yea, that would work, but
1. The CPA will be threatfleet, so there is no telling where they would be. (Oh wait, is he playing with complete visibility? if so, this becomes a moot point.)
2. The AIP will go up, making the next CPAs even thougher. There does eventually come a point where attack forces get so large, not even warheads will be able to kill them all in time.
Not to mention all the other ways AIP makes things harder to deal with. ;)


As for the exos, well, spirecraft are finite in number, and the exos will just keep getting nastier (don't they also come more often if AIP goes up?). He can't keep up the implosion/ram combo up forever. ;)


I guess the key question is if he will be able to reach the CSGs and homeworlds in time before he is forced to use this sort of stuff so many times it no longer becomes sustainable (running out of spirecraft, or attacks and CPAs get too big for even martyrs and warheads to handle in time, Mk. V carrier spam, whatever)
: Re: Ow.
: Faulty Logic May 06, 2013, 07:05:27 PM
Been pretty busy, with the return of health problems, the skyward alpha, and trying all of arcens' other games. But AI War remains my favorite, and I find time for it.

You really know how to make the AI feel welcome.
That was merely a welcome mat. This (http://www.arcengames.com/forums/index.php?action=dlattach;topic=11286.0;attach=5717;image) is a full greeting. And that isn't the craziest choke I've set up.

We'll see about the CPA, but I think we all know where this is going.
I read that as "my victory." Not there yet, though.

I guess the key question is if he will be able to reach the CSGs and homeworlds in time before he is forced to use this sort of stuff so many times it no longer becomes sustainable (running out of spirecraft, or attacks and CPAs get too big for even martyrs and warheads to handle in time, Mk. V carrier spam, whatever)
Indeed. I don't think I'll run out of spirecraft for quite a while, though.

===Operation Swat the CPA====

Yes, "swat" is the appropriate verb. A shadow cruiser (did the Mourner scenario in the 10 mins between announcement and arrival) and black widow kill threatballs fast. I used couple martyrs on the main threatball (3500), and cleaned the others. It wasn't hard, but those TDLs are really, really annoying to clean up. I think a health nerf is in order. Anyway, gametime is now 6:15. Like I said: tedious and time-consuming.

The new toggles are awesome. I'd like to thank whoever suggested them (yes, I'm shameless).

===Operation Advanced Negotiation====

Operational goals: Conquering the three-planet, FactIV/Widow/Arti/Z Reproc V branch, and achieving battleship size. AIP 129.

While Negotiating, repaired new golems, and built a new batch of spirecraft. Exo at 55% by the time I have a battleship. I'll take a couple ARS systems, then see about the latest batch of hacking buffs.

I capture two of the three remaining ARSs (grav drain, sentinel frigate), and gate raid to preserve single-entry (AIP 154), when an 11,000 ship CPA declares. Sigh.

Apparently the AI thinks its reserve is made entirely of bombers. Every one of the ~5000 ships contributed by the SR is a bomber.

The various threatballs are too cowardly to attack, so I kill them one by one. Gametime 8:20.

Exo at 78%.

Should be enough time for some hacking.

Observations
You might want to update guardian-creation-from-carrier costs to reflect their new nastiness.

Firefly bombs are fun.

The SF-Alarm posts may need a revisit. ~10,000 extra threat is no picnic, especially led by enemy frigates.

AI homeworlds should have planetary tachyon emission. As part of the home command station. There's just too much cloaking-related cheese that can occur (penetrator/ram assassinations are the worst of it).

State of Empire:
: Re: Ow.
: keith.lamothe May 06, 2013, 07:28:20 PM
Been pretty busy, with the return of health problems, the skyward alpha, and trying all of arcens' other games. But AI War remains my favorite, and I find time for it.
I hope you feel better soon (and that those problems are not serious).  Glad the games are there to entertain you.  For some definition of "entertain".

Yes, "swat" is the appropriate verb. A shadow cruiser (did the Mourner scenario in the 10 mins between announcement and arrival) and black widow kill threatballs fast.
Clearly the AI needs more fear of wormholes.

Apparently the AI thinks its reserve is made entirely of bombers. Every one of the ~5000 ships contributed by the SR is a bomber.
It has this curious notion that 5000+ bombers can actually penetrate your defenses.

You might want to update guardian-creation-from-carrier costs to reflect their new nastiness.
Hmm, yea, I guess that aspect of their pricing didn't get updated, though the other ones did in various ways.

The SF-Alarm posts may need a revisit. ~10,000 extra threat is no picnic, especially led by enemy frigates.
They may need a revisit in the sense of emphasizing to the player that "there is something on this planet that can kill you".  The mechanic itself I like quite a bit.
: Re: Ow.
: Faulty Logic May 07, 2013, 12:48:10 PM
I hope you feel better soon (and that those problems are not serious).  Glad the games are there to entertain you.  For some definition of "entertain".
Nothing close to life-threatening, but it seriously interferes with going to school. AI War (among others, but mainly this) has been the difference between bad and terrible days.

It has this curious notion that 5000+ bombers can actually penetrate your defenses.
Silly AI. Can't it read the "no fleetships" sign, made of insanity inducers, translocators, and turrets?

They may need a revisit in the sense of emphasizing to the player that "there is something on this planet that can kill you".  The mechanic itself I like quite a bit.
I'd up the AIP to 10, but if you destroy the post, you definitely won't get a special forces attack.

===Operation Mad Hacker====

First hack: complete. Cute new mechanic, there, that warp-counterwave at 30 seconds almost cost me a fabricator.

Unlocks so far:
Harvester IIIs
Military command IIIs
Engi IIs
Y firefly, nanoswarm IIIs
Grav drain IIIs
Sentinel IIs
Enclave IIIs
FlagshipIIs
Heavy Bomber IIs
Zenith/Spire I
Riot IIs
FortressIs
Miniforts.

There is never enough k.

Gametime 8:33, exo progress 82%.

Hack 2 was nastier, due to guardian changes more than anything else. My spider golems were severely damaged by enemy artillery guardians.

Conquer a couple more planets (continuity, bulletproof V/exp engi), and unlock riotIIIs and sentinel IIIs.

By the time my golems are repaired and riots are built, it's about 9 hours. And with only 5% to go, I'll simply wait for the exo.

===Operation Exoswat====

Negotiator and minions deal with the group going for the ACS, and the other exos target the golems (all at the choke). A wave (2300), and threatfleet (7 frigates, 300 TDLS, ~20 guardians, 500 misc) join it. Leading is an H/K III. It gets the same treatment as last time, and the other stuff is pretty much slaughtered.

In other news, a colony rebellion declared, on an artillery golem/raid engine/ grav drill planet, and my AIP is back down to 162. It's not looking good for the AI.

State of Empire (and chokepoint picture (add 32 rams for what it looked like pre-exo)):
: Re: Ow.
: keith.lamothe May 07, 2013, 02:10:37 PM
Nothing close to life-threatening, but it seriously interferes with going to school. AI War (among others, but mainly this) has been the difference between bad and terrible days.
Sorry it's so rough.  Glad that cheesing the living daylights out of an unsuspecting piece of circuitry can be of some assistance, though.


Silly AI. Can't it read the "no fleetships" sign, made of insanity inducers, translocators, and turrets?
None of the ships seem to file proper reports before dropping off the comm net, for some reason.

"Dispatching Taskforce #111001101, objective: discern status of Taskforce #111001100"


Operation Exoswat
More mechanical futility, I see.

I wonder how much of this is the spokes.  At least for the CPAs/threatfleet.  The absolutely unreadable nature of most spokes maps has probably kept people from exploiting this much, but with manual graph untangling the AI's lack of tolerance for disjoint graphs seems to be showing.
: Re: Ow.
: TechSY730 May 07, 2013, 02:39:14 PM
Gee, I wonder how much of this probable coming up victory can be written up to "Spirecraft cheese" (martyr, ram, and penetrator, I'm looking at you) and "AI lack of response to mobile cloak if it gets past the static tachyon defenses" cheese.

Also, the sometimes...odd behavior of the AI when there are more than 1 large-ish disconnected clusters that make up its AI space is indeed quite abusable. Many of the "this group of ships should go to a planet" type systems will cause ships to just blindly stumble into human worlds if there are no other ways they can go. (This is opposed to "I should go to a planet" decisions that happen on a per-ship basis. Those generally will wait until its somewhat safe before charging in)
This can lead to some glorious cheese, especially for "weakly connected" map types like tree, spokes, and the various maze types.
: Re: Ow.
: Faulty Logic May 07, 2013, 03:03:21 PM
Gee, I wonder how much of this probable coming up victory can be written up to "Spirecraft cheese" (martyr, ram, and penetrator, I'm looking at you) and "AI lack of response to mobile cloak if it gets past the static tachyon defenses" cheese.
I don't consider the spirecraft cheese (unless I did something like kill the HW with just penetrators/cloaked rams), because they are, after all, superweapons. Martyrs have saved me time, but have never been vital. Rams did exactly what they were supposed to, same with penetrators.

I really don't mind the cloaking situation as it stands, except on the AI HWs, which, IMO, should always have planetary tachyons.

Many of the "this group of ships should go to a planet" type systems will cause ships to just blindly stumble into human worlds if there are no other ways they can go.
This hasn't happened. If anything, it's the opposite problem, of them being too cowardly to attack, even if they would win (or at least do some damage).

Do threatfleets consider the firepower of neighboring threatfleets when they make their "should I attack" decision? Because I've had systems defended by only a fort+miniforts, and 1500 ships around it, in groups of about 300, and no attack.

I wonder how much of this is the spokes.  At least for the CPAs/threatfleet.
Well, I did build my strategy around the map, making it difficult to really be sure. However, spokes definitely helps, for chokepointing waves (it's not nearly so good for exos, though), and isolating AI fleets from each other.

 I'll typically burn a few warheads at CPA time, and I haven't had to this game, so it has saved me about 10 AIP (at most), though the Golems/Spirecraft alleviate base game threats quite a bit.

cheesing the living daylights out of an unsuspecting piece of circuitry
Anything standing out as disqualifyingly cheesy?
: Re: Ow.
: keith.lamothe May 07, 2013, 03:13:26 PM
cheesing the living daylights out of an unsuspecting piece of circuitry
Anything standing out as disqualifyingly cheesy?
Not really.  My main concern is that the AI is just not handling the disjoint graph problem well; I honestly don't know why those threatfleet ships are just waiting to cross rather than just crossing if they've got a target planet (which they always do).  That might not be better but at least it would have a chance of hurting you instead of just feeding the widow golem.

To your question: the wait logic does consider other threatballs targeting the same planet, yes.

Other than that I wonder about the Spirecraft and how their benefit can be balanced.  Golems too, to a lesser extent.  I mean, I can keep making the exos bigger on Diff 10, but I don't know that this is the right way.  Perhaps having them try to coordinate with CPAs a bit or something like that.
: Re: Ow.
: Faulty Logic May 07, 2013, 03:25:42 PM
Here's the CPA:

It'll hit in about 10 seconds, freeze for about a minute, then pile up on the wormholes. Debug shows that they typically target planets a few hops away, don't know if that explains it.

Edit: sure I have logs:
: Re: Ow.
: TechSY730 May 07, 2013, 03:28:47 PM
Do threatfleets consider the firepower of neighboring threatfleets when they make their "should I attack" decision? Because I've had systems defended by only a fort+miniforts, and 1500 ships around it, in groups of about 300, and no attack.

I think this is the key thing to investigate. Why didn't they attack in this case? Does regular threat also do this "also adjacent planets" check? Does regular threat also factor in threat fleet that is planning on attacking the planet they are waiting on, and vica versa?

If so, why didn't they attack in this case?

I have seen this several times myself as well. Is it the "consider human planets to have X times they really have" thing going that happens at higher difficulties kicking in a bit too aggressively? (Aka, is X set too high for high difficulties) Is there some sort of bug making them not go in when the intended logic says they should?
Hmm, got logs?


Also, mentioning the "afraid of wormholes" thing. I think it would be a smart move if some of the ships waiting around a wormhole pulled back a bit (but still on the same planet) so they don't get insta-alpha'd by a strong AOE coming through. Again, only some, not all, as having some right one top of the wormhole does often allow their alpha to shutdown more moderate forms of attack coming through.
: Re: Ow.
: TechSY730 May 07, 2013, 03:33:45 PM
Here's the CPA:

It'll hit in about 10 seconds, freeze for about a minute, then pile up on the wormholes. Debug shows that they typically target planets a few hops away, don't know if that explains it.

Hmm, could that be it? When waiting to enter planet X (where X might or might not be their final destination), do they consider their "ships pending to attack X" to "enemy stuff on X" firepower ratio, or do they do that ratio comparison relative to the final destination, even if X isn't their final destination?


Optimally, they should be doing some of both. They should be doing it for the final destination to choose which planet they should aim for, and whether to "break it off" and find a new potential planet. If needed, they should also do it for the next enemy planet they need to "lurk" if they must use that planet to get to their destination for determining when to take that next step on their path to the final destination.
Aka, use final destination comparison for "final goal" selection, use the "next" planet comparison for "should I go into the next planet in my path" decision.
: Re: Ow.
: keith.lamothe May 08, 2013, 03:13:11 PM
I just finished ripping all the references to "Firepower" out of the game and replacing it with the relative-strength logic (had to put in numbers for some things that didn't have that, but it's done now).

Testing your scenario the AI fairly quickly overruns Recursion, No Context, and Box of Mazes.

But it looks at "Midnight" and says "Good Night!".  It's about 70,000 strength of Threatfleet (another 12,000 pinned down elsewhere) staring across the wormholes at over 52,000 strength of your mobile and non-mobile defenses.

There's enough AI TF strength (about 4,500) in your "local cluster" to overrun Milkman's Lair and go after High Lab (which the TF is fixated on despite not being able to reach it, due to a lack of "reevaluate whether we can even get there" logic, a lack I'm currently testing a remedy for), but the ally enclaves are camping Milkman's Lair with some 3,000 of their own strength and the AI wants none of that.

Testing continues...
: Re: Ow.
: TechSY730 May 08, 2013, 03:16:24 PM
I just finished ripping all the references to "Firepower" out of the game and replacing it with the relative-strength logic (had to put in numbers for some things that didn't have that, but it's done now).

Testing your scenario the AI fairly quickly overruns Recursion, No Context, and Box of Mazes.

But it looks at "Midnight" and says "Good Night!".  It's about 70,000 strength of Threatfleet (another 12,000 pinned down elsewhere) staring across the wormholes at over 52,000 strength of your mobile and non-mobile defenses.

There's enough AI TF strength (about 4,500) in your "local cluster" to overrun Milkman's Lair and go after High Lab (which the TF is fixated on despite not being able to reach it, due to a lack of "reevaluate whether we can even get there" logic, a lack I'm currently testing a remedy for), but the ally enclaves are camping Milkman's Lair with some 3,000 of their own strength and the AI wants none of that.

Testing continues...

Woa. I knew you mentioned you would like to have only one type of per ship relative strength/cost property, but I didn't think you would actually go through with it this early.
Must of been a lot of work.


Also, how are the firepower displays in the UI handled? Do they now just show the relative strength sum now?
: Re: Ow.
: keith.lamothe May 08, 2013, 03:19:35 PM
Woa. I knew you mentioned you would like to have only one type of per ship relative strength/cost property, but I didn't think you would actually go through with it this early.
Firepower just didn't work as a consistent measure of power.  It was basically just BaseDPS.  Not even counting bonuses or whatever.

Exo budgets are still done their way (not firepower, not strength) since they at least function in a desirable way even if they don't quite make sense at higher tiers.

Also, how are the firepower displays in the UI handled? Do they now just show the relative strength sum now?
They're all updated to use strength now.
: Re: Ow.
: keith.lamothe May 08, 2013, 04:48:43 PM
Found that I was mistaken about thinking that wait points factored in AI ships on other neighboring planets waiting against the same planet.  Well, they do now.

Still not enough to get them to go after Midnight.  I think that's because the wait point logic itself factors in munitions boosts on the target planet and does not factor in munitions boosts on the threat ships that are considering attacking it.  And I agree with that logic; it's somewhat conservative (it could check for mobile munitions boosting ships in the threat itself, which there is in this case) but I think that's appropriate given that a smart player will just assassinate their mobile munitions boosters if it's really making that much difference, whereas the AI will probably not be able to neutralize the human's boosting (largely from a military III in this case, I believe).

Anyway, what I need to track down now is why the threatfleet logic (as opposed to the wait point logic) is being less conservative than the wait logic when it's supposed to be more conservative than the wait logic (to avoid situations like this where chunks of the TF get "stuck" by the wait logic).  I'm guessing it's handling munitions boosts in a different way.
: Re: Ow.
: keith.lamothe May 08, 2013, 05:33:10 PM
In the process of adding some detailed logging to check the strength computations, etc, I apparently introduced a bug which short-circuited the wait point logic in that particular case.

So the entire 82k+ strength TF attacked simultaneously.

Midnight's defenses destroyed them all.

Unattended by player control.

Your Cursed Golem slept through the whole thing.  Nor did any of Tripwire's golems or spirecraft get involved.
: Re: Ow.
: Faulty Logic May 08, 2013, 06:27:30 PM
I just finished ripping all the references to "Firepower" out of the game and replacing it with the relative-strength logic (had to put in numbers for some things that didn't have that, but it's done now).
Very nice.

Testing your scenario the AI fairly quickly overruns Recursion, No Context, and Box of Mazes.
Yeah, they aren't intended to repel attacks, merely discourage them/support the fleet.

But it looks at "Midnight" and says "Good Night!".
[...]Still not enough to get them to go after Midnight.
Midnight, on the other hand...

So the entire 82k+ strength TF attacked simultaneously.

Midnight's defenses destroyed them all.

Unattended by player control.

Your Cursed Golem slept through the whole thing.  Nor did any of Tripwire's golems or spirecraft get involved.
...is an all-out, exo-grade chokepoint. I am not surprised at the result, though it is good to know.

You might want to have stuff like gravity turrets and forcefields factor in to planetary strength evaluations (if they don't already).


Basically, the CPA should be attacking everywhere but Midnight (and possibly High Lab/Milkman's lair, with mercs+allied enclaves) but be hitting hard everywhere else. And the ones that can't shouldn't wait at the wormholes.
: Re: Ow.
: keith.lamothe May 08, 2013, 06:31:32 PM
You might want to have stuff like gravity turrets and forcefields factor in to planetary strength evaluations (if they don't already).
They aren't counted (and haven't been under firepower rules).  I'm not sure how to count them as they shouldn't really scare anyone off on their own, etc.  But it probably should do some degree of "if the player's using those smartly, this is going to hurt a lot more than otherwise indicated, so..."


Basically, the CPA should be attacking everywhere but Midnight (and possibly High Lab/Milkman's lair, with mercs+allied enclaves) but be hitting hard everywhere else.
That's actually exactly what it did with the strength rules, in an earlier test.


And the ones that can't shouldn't wait at the wormholes.
Still working on that ;)
: Re: Ow.
: Faulty Logic May 08, 2013, 06:40:39 PM
BI'm not sure how to count them as they shouldn't really scare anyone off on their own, etc.  But it probably should do some degree of "if the player's using those smartly, this is going to hurt a lot more than otherwise indicated, so..."
Idea: have the "civilian" things multiply the military strength, by (1+ some evalution).

So it won't discourage an attack on an essentially undefended planet, but will factor in to big chokepoints. The actual result of "some evaluation" will need some testing, but I think for each mk*cap of ffs/gravity turrets, it should be about .2 to start.
: Re: Ow.
: keith.lamothe May 08, 2013, 06:42:29 PM
Added a command for logging the AI's wait logic against the planet being viewed (at the time of the command being given).

When the AI looks at Midnight, this is what it sees:

:
5/8/2013 6:34:32 PM (6.027)
-----------------------------------

Logging CalculateWorkingAlliedAndEnemyStrength on Midnight

Allied Total Strength: 19.77; from 1 total ships:
1 of SpecialForcesCommandPost sub-totalling 19.77

Enemy Total Strength: 87359.65; from 3333 total ships:
8 of HeavyBeamTurretII sub-totalling 6336
12 of HeavyBeamTurret sub-totalling 4752
4 of HeavyBeamTurretIII sub-totalling 4752
1 of CursedGolem sub-totalling 2970
8 of FortressMissileModuleII sub-totalling 2304.84
8 of FortressTranslocatorModuleII sub-totalling 2304.84
96 of MissileTurretII sub-totalling 2304
89 of LazerTurretII sub-totalling 2136
96 of CoreZenithParalyzer sub-totalling 1920
19 of CoreZenithElectricBomber sub-totalling 1919.99
3 of Fortress sub-totalling 1782
8 of FortressInsanityInducerModuleIII sub-totalling 1728.63
12 of PlayerControlledZenithRefugeeAdvancedCombatShip sub-totalling 1573.17
48 of SpireGravityDrainIV sub-totalling 1536
96 of YounglingNanoswarmIV sub-totalling 1536
96 of YounglingFireflyIV sub-totalling 1536
4 of PlayerControlledZenithRefugeeCommandShip sub-totalling 1237.5
1 of MilitaryOrbitalCommandStationIII sub-totalling 1188
9 of PlayerControlledGraySpireAdvancedCombatShip sub-totalling 1179.88
9 of PlayerControlledHumanMinerAdvancedCombatShip sub-totalling 1179.88
9 of PlayerControlledNeinzulMournerAdvancedCombatShip sub-totalling 1179.88
9 of PlayerControlledHumanResistanceAdvancedCombatShip sub-totalling 1179.88
96 of Turret sub-totalling 1152
48 of FlakTurret sub-totalling 1152
48 of SpireGravityDrainIII sub-totalling 1152
48 of LightningTurret sub-totalling 1152
96 of YounglingNanoswarmIII sub-totalling 1152
96 of MLRSTurret sub-totalling 1152
96 of YounglingFireflyIII sub-totalling 1152
96 of SniperTurret sub-totalling 1152
95 of MissileTurret sub-totalling 1140
95 of LazerTurret sub-totalling 1140
92 of FighterIII sub-totalling 1104
4 of HeroShieldModuleV sub-totalling 990
8 of CoreZenithReprocessor sub-totalling 960
144 of CoreEtherJet sub-totalling 959.94
12 of PlayerControlledZenithRefugeeNormalCombatShip sub-totalling 786.59
96 of YounglingFireflyII sub-totalling 768
4 of RiotControlStarshipII sub-totalling 768
2 of Flagship sub-totalling 768
96 of MissileShipII sub-totalling 768
96 of FighterII sub-totalling 768
48 of SpireGravityDrainII sub-totalling 768
2 of BomberStarshipII sub-totalling 768
1 of HeroHumanBattleship sub-totalling 768
96 of YounglingNanoswarmII sub-totalling 768
3 of PlayerControlledHumanResistanceCommandShip sub-totalling 742.5
3 of PlayerControlledHumanMinerCommandShip sub-totalling 742.5
3 of PlayerControlledNeinzulMournerCommandShip sub-totalling 742.5
3 of PlayerControlledGraySpireCommandShip sub-totalling 742.5
8 of HeroMLRSModuleV sub-totalling 719.3
9 of PlayerControlledHumanMinerNormalCombatShip sub-totalling 589.94
9 of PlayerControlledGraySpireNormalCombatShip sub-totalling 589.94
9 of PlayerControlledNeinzulMournerNormalCombatShip sub-totalling 589.94
9 of PlayerControlledHumanResistanceNormalCombatShip sub-totalling 589.94
6 of HeroLaserModuleV sub-totalling 539.47
96 of Fighter sub-totalling 384
3 of RaidStarship sub-totalling 384
2 of LeechStarship sub-totalling 384
2 of LightStarship sub-totalling 384
4 of RiotControlStarshipI sub-totalling 384
48 of NeinzulScapegoat sub-totalling 384
2 of Dreadnought sub-totalling 384
48 of SpireGravityDrain sub-totalling 384
96 of MissileShip sub-totalling 384
96 of YounglingNanoswarm sub-totalling 384
1 of SpireStarshipI sub-totalling 384
1 of ZenithStarshipI sub-totalling 384
96 of YounglingFirefly sub-totalling 384
2 of BomberStarship sub-totalling 384
96 of Bomber sub-totalling 384
19 of SentinelFrigate sub-totalling 384
6 of HeroNeedlerModuleIII sub-totalling 323.68
36 of BomberII sub-totalling 288
19 of Cruiser sub-totalling 228
1 of HeroMissileModuleIV sub-totalling 198
1 of ModularSpireFortress sub-totalling 192
1 of ModularNeinzulFortress sub-totalling 192
1 of ModularZenithFortress sub-totalling 192
2 of MiniFortress sub-totalling 178.08
15 of RiotControlLaserI sub-totalling 176.92
12 of BomberIII sub-totalling 144
12 of RiotShieldI sub-totalling 142.12
9 of RiotControlMachineGunI sub-totalling 106.15
2 of RiotControlShotgunII sub-totalling 94.75
6 of ImpulseReactionEmitterIII sub-totalling 72
2 of RiotControlTazerI sub-totalling 47.37
2 of RiotControlLaserII sub-totalling 47.18
2 of RiotControlMachineGunII sub-totalling 47.18
3 of RiotControlTractorI sub-totalling 35.53
1 of RiotControlShotgunI sub-totalling 23.69
2 of VampireClawII sub-totalling 13.71
8 of NeinzulMLRSDroneIV sub-totalling 11.91
8 of NeinzulNeedlerDroneIV sub-totalling 11.91
8 of NeinzulLaserDroneIV sub-totalling 11.91
8 of NeinzulMissileDroneIV sub-totalling 11.91
8 of NeinzulMissileDroneIII sub-totalling 8.93
8 of NeinzulMLRSDroneIII sub-totalling 8.93
8 of NeinzulLaserDroneIII sub-totalling 8.93
8 of NeinzulNeedlerDroneIII sub-totalling 8.93
20 of NeinzulMLRSDrone sub-totalling 7.23
20 of NeinzulMissileDrone sub-totalling 7.23
20 of NeinzulNeedlerDrone sub-totalling 7.23
20 of NeinzulLaserDrone sub-totalling 7.23
8 of NeinzulLaserDroneII sub-totalling 5.96
8 of NeinzulNeedlerDroneII sub-totalling 5.96
8 of NeinzulMLRSDroneII sub-totalling 5.96
8 of NeinzulMissileDroneII sub-totalling 5.96
1 of InfiltratorIII sub-totalling 3.6
1 of MobileRallyPost sub-totalling 0
1 of EnergyCollector sub-totalling 0
1 of MercenaryNeinzulEnclaveStarship sub-totalling 0
6 of ForceFieldGenerator sub-totalling 0
4 of MetalHarvesterIII sub-totalling 0
34 of GravitationalTurret sub-totalling 0
6 of MineLayer sub-totalling 0
6 of SpaceDock sub-totalling 0
11 of EngineerDrone sub-totalling 0
5 of LightForceFieldGenerator sub-totalling 0
78 of Mine sub-totalling 0
1 of CloakerStarship sub-totalling 0
6 of EngineerDroneII sub-totalling 0
1 of TransportShip sub-totalling 0
2 of CrystalHarvesterIII sub-totalling 0
4 of ScoutStarship sub-totalling 0
12 of FortressShieldModuleII sub-totalling 0
1 of SpirecraftJumpship sub-totalling 0
2 of NeinzulEnclaveStarship sub-totalling 0
5 of HardenedForceFieldGenerator sub-totalling 0

Inventory of threat power on wait points against Midnight
Total Strength: 80003.41; from 2867 total ships:
28 of AICarrier sub-totalling 56606.79
150 of TackleDroneLauncherII sub-totalling 7200
1759 of BomberII sub-totalling 7036
574 of FighterII sub-totalling 2296
104 of MunitionsBoosterII sub-totalling 2101.89
62 of ParasiteII sub-totalling 496
5 of AIGuardianBeamII sub-totalling 480
2 of AIHeroSpireFrigate sub-totalling 384
2 of AIGuardianLaserIV sub-totalling 384
2 of AIGuardianSpireImplosionIV sub-totalling 384
2 of AIHeroNeinzulFrigate sub-totalling 384
1 of WarbirdStarship sub-totalling 320
75 of MissileShipII sub-totalling 300
3 of AIGuardianLaserII sub-totalling 288
41 of VampireClawII sub-totalling 281.13
1 of AIGuardianFlakIV sub-totalling 192
1 of AIGuardianArtilleryIV sub-totalling 192
20 of SpireTeleportingLeechII sub-totalling 160
1 of AIGuardianZombieII sub-totalling 96
1 of AIGuardianArtilleryII sub-totalling 96
1 of AIGuardianFlakII sub-totalling 96
1 of AIGuardianSpireImplosionII sub-totalling 96
3 of ParasiteIV sub-totalling 48
19 of InfiltratorII sub-totalling 45.6
8 of ImpulseReactionEmitterII sub-totalling 32
1 of CruiserIV sub-totalling 8

Checking for "we have enough strength" clearing wait points against Midnight
ShipsAlreadyWaitingReady.Count:2890
strengthOfThreatWaitingAgainstTargetPlanet:80003.41
targetPlanetRollup.WorkingEnemyStrength:87359.65
targetPlanetRollup.WorkingAlliedStrength:19.77
targetPlanetNetEnemyStrength:7336.47
confirmingClearWaitPointOrder:false

http://dilbert.com/strips/2013-04-12


It's actually closer to going through than I thought.  And it would still lose.

Btw, with the "threat waiting considers all ships waiting on that planet" logic the AI actually did make the dash through Milkman's Lair to try to get to Midnight via Tripwire.  A goodly chunk of the AI actually made it through the atmospheric density of ally-faction enclave younglings to the Tripwire wormhole... and promptly turned around, figuring the few extra younglings it could dispatch on ML was more than it could do against what awaited on Tripwire.

What it should have done is go after High Lab; just a result of its "one threatfleet to rule them all" characteristics and Midnight being the target accessible from the "origin" (the AI HWs).  Not entirely sure if I want to try to make it multi-fleet as that's gonna be quite complicated, but this is one argument for it.
: Re: Ow.
: keith.lamothe May 08, 2013, 06:44:52 PM
Idea: have the "civilian" things multiply the military strength, by (1+ some evalution).

So it won't discourage an attack on an essentially undefended planet, but will factor in to big chokepoints. The actual result of "some evaluation" will need some testing, but I think for each mk*cap of ffs/gravity turrets, it should be about .2 to start.
Yea, that occurred to me too, but does it actually work?  I can see a mkI cap of ffs making a 10,000 strength defense "look like" a 12,000 strength defense.  But should it really make an 80,000 strength defense "look like" a 96,000 strength defense?

In this case, I suppose so ;)
: Re: Ow.
: Faulty Logic May 08, 2013, 06:56:16 PM
But should it really make an 80,000 strength defense "look like" a 96,000 strength defense?
Yes. An 80,000 strength defence performs a lot more than 8 times better than a 10,000 defence, and this would help the AI "realize" that.

That's actually exactly what it did with the strength rules, in an earlier test.
Wait, then why did you change it?

he AI actually did make the dash through Milkman's Lair to try to get to Midnight via Tripwire.
"Alright, we just need to wade through a horde of younglings, and get past all the superweapons the human isn't using right now to get through to his massive chokepoint! Wait..."

What it should have done is go after High Lab
"Alright, we just need to wade through a horde of younglings, and then the mercenaries, and then... what is that huge ship that just appeared? It's tearing us apart!"

Not entirely sure if I want to try to make it multi-fleet as that's gonna be quite complicated, but this is one argument for it.
Wait, you mean every threatfleet in the galaxy acts like one? It definitely shouldn't.
: Re: Ow.
: keith.lamothe May 08, 2013, 07:04:38 PM
That's actually exactly what it did with the strength rules, in an earlier test.
Wait, then why did you change it?
That's what it's still doing; I meant "strength" (the new numbers) as opposed to "firepower" (the old numbers).  The test was "earlier" in the sense of hours ago, whereas the tests I'm currently running are at a later stage in the game (the woebegone siege of Midnight).


he AI actually did make the dash through Milkman's Lair to try to get to Midnight via Tripwire.
"Alright, we just need to wade through a horde of younglings, and get past all the superweapons the human isn't using right now to get through to his massive chokepoint! Wait..."

What it should have done is go after High Lab
"Alright, we just need to wade through a horde of younglings, and then the mercenaries, and then... what is that huge ship that just appeared? It's tearing us apart!"
The AI just can't get no respect.


Not entirely sure if I want to try to make it multi-fleet as that's gonna be quite complicated, but this is one argument for it.
Wait, you mean every threatfleet in the galaxy acts like one? It definitely shouldn't.
There is only one threatfleet, correct.  Before joining the threatfleet threat behaves normally for 30 minutes, and that is more "distributed" as you'd expect.

But when CPA ships were changed to get dunked directly into threatfleet...  Well, I'm making a change now to not try to TF any ship that isn't on the AI's "side" of the galaxy.  Better that they stick to the "local" logic if they're cut off in some corner of the galaxy on the wrong side of one mean chokepoint.
: Re: Ow.
: TechSY730 May 08, 2013, 07:06:17 PM
Not entirely sure if I want to try to make it multi-fleet as that's gonna be quite complicated, but this is one argument for it.
Wait, you mean every threatfleet in the galaxy acts like one? It definitely shouldn't.

Yes. Same with special forces. It causes strange behavior when the AI space is not connected. I've mentioned this about special forces before, and the same arguments apply to threat fleet as well.
: Re: Ow.
: Faulty Logic May 08, 2013, 07:11:13 PM
The AI just can't get no respect.
I have plenty of respect for it. In fact, being willing to do crazy things like that when I least expect is a big part of it. But the price of doing the right thing when it looks wrong is doing the wrong thing when it looks (and is) totally crazy.

Well, I'm making a change now to not try to TF any ship that isn't on the AI's "side" of the galaxy.
Sounds great.

: Re: Ow.
: TechSY730 May 08, 2013, 07:16:00 PM
But when CPA ships were changed to get dunked directly into threatfleet...  Well, I'm making a change now to not try to TF any ship that isn't on the AI's "side" of the galaxy.  Better that they stick to the "local" logic if they're cut off in some corner of the galaxy on the wrong side of one mean chokepoint.

That will sort of break down if there are two or three "parts" of roughly or exactly the same size. Especially if the to AI homeworlds do not have a fully AI space path to each other, which is pretty trivially doable on some map types like tree, snake, and X.


Still, I am glad that "should I go in" logic is getting a lookover, especially in terms of actual how to measure "threat levels", target selection, how do deal with enemy planets on the path to the target, other "groups" of stuff also going to that target, etc...

The...oddities with this logic have been a major thing holding the AI back. Good to see you addressing it. :)



And I agree with treating "support" units like forcefields, tractors, and grav turrets as force multipliers. That would seem to be a decent approximation. Finding good multipliers of course will be tricky. ;)
: Re: Ow.
: keith.lamothe May 08, 2013, 10:31:48 PM
I've got it doing the waiting well away from the wormholes, but I'm wondering if this isn't going to replace one vulnerability (getting tractored) with another (getting defeated in detail on the other side of the wormhole when going through due to spreading out from speed differences).

And having some camp the wormhole and others hang back would actually be the worst of both worlds: the campers would be even easier to tractor, and the hangers-back would have an even higher chance of coming through after a huge chunk has already died.
: Re: Ow.
: keith.lamothe May 08, 2013, 10:51:06 PM
Ok, the labors of the day are in for the next version now:

* Replaced the "Firepower" metric (which was basically just BaseDPS, not considering other stats) with the "Strength" metric (which, while not perfect, has been used with success elsewhere and seems to be a much better measure for comparing groups of units to see which one is stronger; something that the game does rather a lot, particularly in AI logic).
** You're likely to see situations where the AI is undervaluing/overvaluing some things and making... "questionable" decisions about when to commit to an attack, etc.  Your reports (and save games) will be helpful in fixing these cases.

* Substantially improved the AI's "intelligence" in handling its threat ships:
** Previously when asking "is it safe to move this threat to its next destination planet?" it would only consider the strength of the threat on that group's current planet (and anything actually on the target planet), but now it also considers any other threat (on other planets adjacent to the target) waiting to go through.
** It now asks that question (is it safe, blah blah) even when moving on to other AI-held planets.  Unless it's a special forces ship, which is just supposed to charge in if it's an AI planet.
** If it decides a ship needs to wait, the ship now waits somewhere not directly on top of the wormhole leading to the target planet, because that position is far too vulnerable to tractor shenanigans and such.
** If a ship is in a part of the galaxy that the AI cannot reasonably easily gather forces from (meaning there's some nasty human chokepoint cutting it off from the AI homeworlds), the AI will no longer try to enlist that ship into the Threatfleet.  The "local threat" logic is better in those cases.
*** Note that ships already in threatfleet behavior (from previous saves, or before you cut it off from the AI HWs, etc) will stay that way.

* Added a debug chat command: "cmd:log ai wait"
** If used in MP this only affects the host, since only the host runs the AI thread.
** If you're in planet view when you give the command, this tells the AI thread to generate a "LogicLog_AICheckForClearingWaitPointsAgainstSpecifiedPlanet_AIThread.txt" log file in your RuntimeData directory whenever evaluating whether to clear wait points (i.e. charge!) against that planet.
** If you're in galaxy view when you give the command, this tells the AI thread to stop said logging.
** The contents of this log are fairly readable if you understand the terms, but the main benefit is so you can send it to us if you have questions about why the AI was (or was not) waiting to attack in a particular instance.

Not sure when the next release will be, but I'll definitely want to make this available for testing sooner than later.  Gotta break things to improve them when it comes to this stuff, but I'm hoping they won't be broken or at least won't stay there long.  Things seemed fine in my tests but that often means little.
: Re: Ow.
: keith.lamothe May 09, 2013, 11:23:35 AM
Just got the "clear wait points" logic to group-move each planet's chunk of threat towards its target wormhole, so that deals with one of the downsides of "don't camp right on the wormhole, you moron".

The other is that right now I've got them waiting either at the wait-planet's command station or at a point selected in a way similar to adv-fact placement: near the inner grav ring, as far from wormholes as possible.

This does reduce the threatball's defensive utility (they'll FRD into attackers from there but it can take a while to get in-system) but I don't mind that so much: the AI's priority for these should probably be "conserve until I can use them offensively", and leave defenses to the guards and the SF and such.

The actual problem is that it can take quite a while for the threatball to actually reach the wormhole once the "stop waiting" order comes down.  It can also take a pretty variable length of time for one planet's threatball to reach the wormhole compared to another planet (that was waiting against the same target, for instance).

So I'm thinking that they need to pick a closer spot to the wormhole.

My question to y'all is: what range? 

In theory it could be variable (based on "does the human have a black widow golem?", probably ;) ) but I think a single global range (or range of ranges) would be better for a first try.
: Re: Ow.
: keith.lamothe May 09, 2013, 11:29:00 AM
Btw, in the test I just ran the AI did a good job of wiping out most of the satellite planets.  Basically only High Lab, Tripwire, Midnight, and some of the "satellites" around Box of Mazes stood; the friendly enclaves stopped the enemy from getting further in than Milkman's Lair and killed the stuff on Box that would have cleaned out its satellites.

Once the human went into brownout the AI attacked Midnight (not entirely sure why, I don't think it knows to specifically recognize that situation) and actually almost took it down.  But the energy balance came up enough from losses that the FF net came back on line.

Naturally any player-attention would have fixed the brownout much faster, but I think things are improving :)
: Re: Ow.
: Faulty Logic May 09, 2013, 11:37:23 AM
So I'm thinking that they need to pick a closer spot to the wormhole.

My question to y'all is: what range? 
An AAR is probably not a good spot to get a lot of responses.

I'd say a flat 20,000 from the target wormhole.

In theory it could be variable (based on "does the human have a black widow golem?", probably   ;)
No. Riots don't work as well as widows for this specific thing, but they get the job done.

Btw, in the test I just ran the AI did a good job of wiping out most of the satellite planets.  Basically only High Lab, Tripwire, Midnight, and some of the "satellites" around Box of Mazes stood; the friendly enclaves stopped the enemy from getting further in than Milkman's Lair and killed the stuff on Box that would have cleaned out its satellites.
I've got a bad feeling about this change. Of course, this result was without player intervention.
: Re: Ow.
: keith.lamothe May 09, 2013, 11:40:34 AM
An AAR is probably not a good spot to get a lot of responses.
Yea, I was mainly asking the two of you actually tracking with me here.  20k sounds like a good first shot.
: Re: Ow.
: TechSY730 May 09, 2013, 12:06:03 PM
Sorry, I haven't quite internalized the unit used for distances and "practical" data points for them.

What the range for a plasma siege starship again? Zombard? Missile frigate? Ether jets? Tractor platforms? Martyr? Widow golem? *contemplates trip to the wiki*
: Re: Ow.
: Faulty Logic May 09, 2013, 12:40:07 PM
20000 is in zombard range, but none of the others. Widow tractors operate at 12000 range.

What the range for a plasma siege starship again? Zombard? Missile frigate? Ether jets? Tractor platforms? Martyr? Widow golem? *contemplates trip to the wiki*
10000, 30000, 10000, 5000, 5000 (~1/2 for tractors), 3500, 19,000 (tractors at 12000). Note that some ranges increase with mark.
: Re: Ow.
: keith.lamothe May 09, 2013, 12:53:19 PM
Ok, retested with the wait-20k-out rule and with the exos syncing (via pre-100% launch) with the CPA.  The CPA still isn't jumping in after the exo as thoroughly as it should but it did do partially.  The combined impact took down the Midnight command station but there was still just too much left on Midnight for the AI to take out.  It overran everything outside the Midnight chokepoint, though, except for a couple isolated planets it just wouldn't attack (debugging that now).
: Re: Ow.
: Faulty Logic May 09, 2013, 01:04:01 PM
Ok, retested with the wait-20k-out rule and with the exos syncing (via pre-100% launch) with the CPA.
I'd be very careful with that rule. CPAs happen at close to twice the rate of exos, so having them try to sync could make it so I never see a full exo.

The combined impact took down the Midnight command station
What?! How? Infiltrators, or brute force?
: Re: Ow.
: keith.lamothe May 09, 2013, 01:07:41 PM
I'd be very careful with that rule. CPAs happen at close to twice the rate of exos, so having them try to sync could make it so I never see a full exo.
True.  The alternative is to try to sync CPAs up to the exos (so you'd have something like CPA -> CPA/exo -> CPA -> CPA/exo, etc).

Anyway, it does seem that the AI's responses to your normal stuff and its responses to your superweapons need to cooperate, or there's no way through that door ;)

The combined impact took down the Midnight command station
What?! How? Infiltrators, or brute force?
I wasn't watching closely, but I suspect the H/Ks.  That part of Midnight was hammered flat; nothing left.
: Re: Ow.
: Faulty Logic May 09, 2013, 01:16:48 PM
Anyway, it does seem that the AI's responses to your normal stuff and its responses to your superweapons need to cooperate, or there's no way through that door
This principle applies to all the significant asynchronous responses (FS exos, Golem/Botnet/Spirecraft exos, Nemeses, CPAs). They should all sync up to some degree, or the player gets an advantage for each new, reasonably fair "deal."

True.  The alternative is to try to sync CPAs up to the exos (so you'd have something like CPA -> CPA/exo -> CPA -> CPA/exo, etc).
I think that would work out pretty well (for the AI).
: Re: Ow.
: Hearteater May 09, 2013, 01:50:54 PM
I think the AI does need to combine its efforts to a degree, otherwise it is too easy to defend, especially with Superweapons.  You could make the degree to which it bends the rules for sync'ing be related to game time.  So early game syncs are just random.  But the longer the game drags on, the more "effort" is put into syncing.  For example, for every one hour of game time, a CPA or Exo can be bumped two minutes later to coincide with the other.
: Re: Ow.
: Faulty Logic May 09, 2013, 01:59:12 PM
You could make the degree to which it bends the rules for sync'ing be related to game time.  So early game syncs are just random.  But the longer the game drags on, the more "effort" is put into syncing.  For example, for every one hour of game time, a CPA or Exo can be bumped two minutes later to coincide with the other.
That is an excellent idea.
: Re: Ow.
: Toranth May 09, 2013, 03:55:46 PM
An AAR is probably not a good spot to get a lot of responses.
Yea, I was mainly asking the two of you actually tracking with me here.  20k sounds like a good first shot.
Not sure what you want with the 'waiting-off-the-wormhole' bit.  If you want to protect the threatball from warheads and tractor raids, then yeah, 20,000 or so is fine.  But that means most units will be out of range to shoot anything.
I'd think you either want to have the units pull back to say 90% of their range (so all units are still barely covering the wormhole) or 5 seconds at that units move speed from the wormhole.
Either of these, especially if you spread it out in an arc, will scatter the units but still leave them effective.  The range pull back is the combination of safest/most aggressive, while the time pull back is so that when the threatball DOES decide to attack, all units would arrive at about the same time.

You could make the degree to which it bends the rules for sync'ing be related to game time.  So early game syncs are just random.  But the longer the game drags on, the more "effort" is put into syncing.  For example, for every one hour of game time, a CPA or Exo can be bumped two minutes later to coincide with the other.
That is an excellent idea.
I agree.  I've always wanted a little more variance in Exos in general (like normal waves), and timing them to sync a little better both does that and improves the 'impact' as the importance of syncing goes up.
: Re: Ow.
: Faulty Logic May 09, 2013, 08:18:29 PM
Note to self: don't accidentally click lock topic.

Anyway, this is an AAR, so:

===Operation Excess Spending====
The rebels come with a third artillery golem, and another in the branch has a second regen (still my least favorite of the golems, but not nearly as bad as it was). I now have all the golems, and unlock riots III, zenith II, and flagships III.

Then I hack a couple more systems. Still going with two hackers.

Then I clean up another CPA (14,000 ships), leaving about 2000 around.

Raiding for the only titanite and then the four adamantite (penetrators IV/V, implosion V, attritioners IV) take the gametime to 11 hours. Mercenary fleet at full fighter, beam frigate, and enclave caps, with 73 bombers as well. I like having incomes of ~5,000 each per second.

Exo at 58%.

Now, about those hacking changes...

===Operation ICE Breaker====

The fifth hack gets interesting. Those implosion guardians really do a number on the golems.

Exo-galactic strikeforce responding to hacking
What the?! Ok that got my attention; pausing the hack (1044 k extracted).

Well, that was unexpected. A few groups come at Midnight and evaporate, but one went for High Lab, and the mercenary fleet took some casualties.

Repaired, econ back at ~500,000 each, and I redeploy, with more of the fleet available for response, and less on site.

Wow, this hurts. Subhacked until I got to 3000 k. Unlocked Assault Transport.

Gate raided my local cluster, so the AI will have to break through Midnight to kill me.

Negotiated a cloak path to the blind homeworld. This does trigger a CPA post.

Same to the one with the eye.

Gametime is now 11:40

===Operation Five Minutes to Midnight====

I take the (whole) fleet and smash my way up to the first HW. Warheads under cloak kill the SR, and then I slaughter the guardposts, while the artillery golems knock out the core forcefields. I pop the command station, and go back to Midnight (which had no problems with the mini-exos).

Gametime 11:50.

Pop a few D nodes to refleet essentially instantly, and then send the entire fleet to the other homeworld. Eye killed instantly with penetrators, strategic reserve warheaded, forcefields artilleried, and victory achieved.

Observations

CPA posts need to activate immediately to have much impact.
The HWs should have planetary tachyons (I could have won simply with cloaked rams and penetrators).
Special forces should base themselves at the HWs, to prevent situations like me cutting them off.
If they are cut off, then they should redeploy via carriers.
SC shieldbearers are awesome come HW attack time.
Hacking responses are about right (at hack 5, it is usually easier to just conquer a planet).
SCLs might not be that bad. Possibly.
: Re: Ow.
: keith.lamothe May 09, 2013, 08:33:07 PM
Anyway, this is an AAR
I have Aura of Thread Derailing +5, sorry ;)

Exo-galactic strikeforce responding to hacking
What the?! Ok that got my attention; pausing the hack (1044 k extracted).
I thought you'd appreciate that.

Pop a few D nodes to refleet essentially instantly, and then send the entire fleet to the other homeworld. Eye killed instantly with penetrators, strategic reserve warheaded, forcefields artilleried
"Our powers combined..."

and victory achieved.
Congratulations!  Again.  (sound of AI tinkering in the background... was that a scream?)

Special forces should base themselves at the HWs, to prevent situations like me cutting them off.
If they are cut off, then they should redeploy via carriers.
As in cheat to hop to the HW?

SC shieldbearers are awesome come HW attack time.
Yea, they're kind of the defensive equivalent of the warhead.  The match seems natural.

SCLs might not be that bad. Possibly.
I noticed you riding the floor with those.  The AI should be a bit better at killing them when I'm through with them.
: Re: Ow.
: Faulty Logic May 09, 2013, 08:41:37 PM
I have Aura of Thread Derailing +5, sorry  ;)
No problem.

(sound of AI tinkering in the background... was that a scream?)
The sound of the AI attack synchronization?

As in cheat to hop to the HW?
Yup. On second thought, it wouldn't really affect me, beyond deploying a nuke.

I noticed you riding the floor with those.  The AI should be a bit better at killing them when I'm through with them.
Yeah, I saw some AI ships literally fly past one of them without killing it.


Oh, I forgot one of the goals of this game: chokepoint comparison. Yeah, single choke is a luxury, but it isn't necessary to victory. No-choke is possible without any superweapons, but you need to be constantly vigilant.
: Re: Ow.
: keith.lamothe May 09, 2013, 08:55:31 PM
(sound of AI tinkering in the background... was that a scream?)
The sound of the AI attack synchronization?
Among other things.

I just tested with a rule change that makes carriers do partial deployments to keep the number of AI ships on each planet (with any kind of battle) up to 2000.

With that rule change, and the coordination with the exos (which don't actually do a whole lot by themselves, unless the HK spawns on a line that takes it to the Midnight CS without going through the black hole), the cpa wiped Midnight clean.

I'm also shoulder-deep in the code that does planet-target selection for "local threat" (free ships that aren't in Threatfleet), and I'm pretty happy with the improvement thus far.  With that and the "don't wait right on top of the wormhole you moron" rule I think CPAs may actually be more effective if I have them not start in threatfleet anymore.  Gives them an opportunity to pick off outlying stuff even if the map is still "connected" for the AI.


I noticed you riding the floor with those.  The AI should be a bit better at killing them when I'm through with them.
Yeah, I saw some AI ships literally fly past one of them without killing it.
Yea, the AI didn't even consider it an "irreplaceable".  No longer.


Oh, I forgot one of the goals of this game: chokepoint comparison. Yeah, single choke is a luxury, but it isn't necessary to victory. No-choke is possible without any superweapons, but you need to be constantly vigilant.
Midnight certainly is pulling choke-duty.  It's also pulling map-fragmentation duty in a big way.  But yea, the Milkman's Lair angle is still open (until you gate raided back there anyway) and I've seen the AI pull off exo kills that way sometimes.  But player-attended it would still be very survivable.
: Re: Ow.
: TechSY730 May 09, 2013, 11:48:20 PM
For the AI HW tachyon thing.
You know how we get a "feature packed" forcefield on our planet (gravity and tachyon in a decently large radius)?

Why not make a AI HW core forcefield with the same sort of "bonus attributes" but is modeled after the AI core forcefield rather than the MK. I forcefield?

Like us, each AI would only get one of them, over their homeworld command station, and not replaceable. Also, maybe the AI should have a tachyon guardian over their HW command station always on game seed, and a higher chance to spawn them afterwards, with a large boost in chance if both that special forcefield and their is no tachyon guardian currently guarding the home.

This should be able to shut down most of the cheapest forms of cheese from being, well, cheap (as in, without having to invest any more than getting there and the units to do it), just like our home forcefield does.



For the special forces thing.
How about making it such that as the human gets closer and closer to an AI homeworld OR as the AI loses core guard posts, the "max staging distance from HW" for the special forces would go down. The "max staging distance from HW" would be based on whichever factor was more "severe" at the time. (The core guard post condition also being there is to prevent special forces from being "lazy" after repeated ultra deep-strike style cheese).

Note, I am referring to the "staging area" to be where special forces ships gather when there are no "jobs" for them to do. They would still go wherever they "feel" is worth defending if the AI under attack.
: Re: Ow.
: TechSY730 May 09, 2013, 11:49:55 PM
Derp. Almost forgot.

Congrats on the win, though I think you may have been able to finally uncover the situations and the "need" to deal with one of these few remaining "gaps" in the AI intelligence (where and when to attack). That may not bode well for your next AAR.  ;)
: Re: Ow.
: Faulty Logic May 09, 2013, 11:59:08 PM
Congrats on the win, though I think you may have been able to finally uncover the situations and the "need" to deal with one of these few remaining "gaps" in the AI intelligence (where and when to attack). That may not bode well for your next AAR.   ;)
Thanks. I'm not too worried; the AI will be able to attack as intelligently as it likes. I will have eight champions and an FS fleet waiting for it. And warheads, of course.

In fact, this (http://www.arcengames.com/mantisbt/view.php?id=11091) is much more of an issue.
: Re: Ow.
: keith.lamothe May 10, 2013, 12:06:02 AM
Thanks. I'm not too worried; the AI will be able to attack as intelligently as it likes. I will have eight champions and an FS fleet waiting for it. And warheads, of course.
Yea, my guess is that the scaling nebula rewards will make you unassailable just by themselves.  32 Mod Forts?  Against single-HW stuff?  Sure you'll face like a billion nemesis units, but they won't scratch that defense.  If you can power it, at least.

In fact, this (http://www.arcengames.com/mantisbt/view.php?id=11091) is much more of an issue.
The counter posts, or the CSGs themselves?
: Re: Ow.
: Faulty Logic May 10, 2013, 12:14:27 AM
Yea, my guess is that the scaling nebula rewards will make you unassailable just by themselves.  32 Mod Forts?  Against single-HW stuff?  Sure you'll face like a billion nemesis units, but they won't scratch that defense.  If you can power it, at least.
Just 24 (human mod fort is a k unlock), and energy will be an issue. Further, the AI doesn't have to kill my defenses, just get past them. And it will be single-HW Exogalactic Transceiver stuff, so I doubt it will be all that easy. And I wouldn't put it past you to implement this (http://www.arcengames.com/mantisbt/view.php?id=9850) in the middle of my AAR.

The counter posts, or the CSGs themselves?
The posts. The CSGs go away as soon as I conquer 1 of them, after all.
: Re: Ow.
: keith.lamothe May 10, 2013, 06:57:00 PM
Ok, spent all day in the individual-ships-tactics AI (writing a lot of toggleable logging that can be activated for one unit at a time, for my own understanding at least but should help people report some of the behavior oddities that don't reproduce after a save/load), made a ton of corrections.

Now that exo+cpa flattens an unattended Midnight while only taking about 20% casualties (didn't measure real close, but I think the high-water-mark was about 120k strength of threat, and it didn't get below 100k during the battle).

Still a bunch of things to fix, and I still expect you'll embarass the AI with 8 champs, but it should do a bit better job once I'm done with the current round of changes, do some make-sure-not-totally-broken testing, etc.