AI War Beta 7.017 "Ominous Rumblings" Released!

[keith.lamothe]

Hunter/Killer Mk Is are considered strength 3,960 strength, Mk IIs are 7,920 strength.  I'd rate a Battleship right around 7,000 strength.  Even with random modules, around 6,000 strength.

Fair enough, I can have it use special strength budgeting there (if I change the actual BB strength then it will probably over-awe the AI in terms of evaulating when to attack, as the modules will count too where they aren't being paid for here).

But what should the FF, DD, and CA costs be?  Is 192 reasonable for a FF?  And maybe 700 for a DD and 2500 for a CA?

So that wave?  It basically had two H/K Vs and three H/K Is added on to it.

(a look of deep worry and concern fails to cross the developer's face)

[RockyBst]

Dual homeworlds too, so two waves at once. Now I don't feel so bad about losing my crunchy outer defensive worlds, even with the artillery golem, regen golem and a 4,000 ship fleet all smacking on them.

EDIT: And when the waves decided to sync with a CPA synced to a Fallen Spire exo lead by golems... my defences weep. Battleships really do melt forcefields, even a mark 3 goes down in seconds.

[Aklyon]

Aside from that, its sorta funny that the Heroic AI throws dice at you that turn into champs.

[Toranth]

Hunter/Killer Mk Is are considered strength 3,960 strength, Mk IIs are 7,920 strength.  I'd rate a Battleship right around 7,000 strength.  Even with random modules, around 6,000 strength.

Fair enough, I can have it use special strength budgeting there (if I change the actual BB strength then it will probably over-awe the AI in terms of evaulating when to attack, as the modules will count too where they aren't being paid for here).

But what should the FF, DD, and CA costs be?  Is 192 reasonable for a FF?  And maybe 700 for a DD and 2500 for a CA?

I know (from experience - I hate H/K factories) that 8 Level 1 Human Frigates can kill an H/K Mk I if they use their Shadow Shields.  Without it, they do about 60-70% damage.  With the correct unlocks (FF Mk II, Polarizer Mk II, Missile Mk II) the Frigates can kill the H/K without using abilities.
So, I'd rate a Frigate a little less than 1/10 of an H/K Mk I, or somewhere in the 300 Strength range.  Maybe as low as 250, if you count up the advantage of numbers a lot.  A Frigate with Mk II modules can beat a Mk I Zenith Starship (Strength 192) and does 60% to a Mk II (Strength 384).  So, somewhere in between.

Three Destroyers with DAs, Polarizers, or IREs can almost kill it (90% damage) without using abilities.  Four DDs with base weapons can always kill one.  Two with unlocks and abilities.
I'd rate a Destroyer at 1/4 an H/K Mk I, or 1000-ish Strength.  A single DD is also better than 2 AI frigates but not quite 3.  Again, can drop it a bit if you count outnumbering heavily.

Cruisers step it up again.  One Cruiser with base weapons can only do 40% damage to an H/K Mk I, but 2 can kill it without loss.  One do 85% damage using Shadow Shields.  One Spire CA with Photon Lances and Polarizers can mutual kill an H/k without using shields.  In several tries, sometimes the H/K survived with 1% health if it moved a little during the fight.  4 v 4 of that configuration was a win for the player almost every time.  On the other hand, 6 AI frigates can kill that decked-out Cruiser every time.  4-5 die, but they never lost.
So I'd rate a Cruiser at near an H/K Mk I.  3,500-3,800 Strength in general, maybe, to make up for the lack of guaranteed to have the best modules?  If you plan for multipliers and best modules, then put it at exactly an H/K Mk I match - 3,960.

And then the Battleship at 7,000 or so.  I managed to get a configuration for 8 Battleships that would consistently kill 3 H/K Mk Vs, which would put it at a bit less than 7,500.

In which case

Hull	Strength
FF	300
DD	1,000
CA	3,500
BB	7,000

[keith.lamothe]

Sounds like being an H/K on your computer is a losing proposition

Thanks for the numbers, they make a lot of sense and I'll probably go with them.

Currently debugging my way through some Special Forces logic tomfoolery.  I never seem to understand a part of this thing until I stuff it to the gills with logging code.

[Eternaly_Lost]

Sounds like being an H/K on your computer is a losing proposition

Thanks for the numbers, they make a lot of sense and I'll probably go with them.

Currently debugging my way through some Special Forces logic tomfoolery.  I never seem to understand a part of this thing until I stuff it to the gills with logging code.

I am sure that the Hunter Killers feel the same on my machine as well. I do kill them by the hundreds, and that is the MK5 Version. I don't bother to count anything less then that.

My Homeworld Command Station would not have enough wall space to mark all the kills if I did.

[keith.lamothe]

Oh, I'm sure your machine occupies a very special place in their cosmology

[Chthon]

I think, when compared to spire fleets or shadow champions, hunter killers are rather lackluster.

[keith.lamothe]

I think, when compared to spire fleets or shadow champions, hunter killers are rather lackluster.

Against much short of those (the most powerful superweapons you can get) they're nigh-unstoppable, though.

They're less impressive in many ways due to the lack of modules.

They could be given modules.  You Do Not Want That To Happen.

[RockyBst]

Although the strength estimates posted above sounds pretty accurate, let's also remember that this is a 400 AIP wave for a very hard type AI on diff 9. I'd expect 2-3 battleships, if we go with 7,000 strength for one then most of the wave's dirty tricks budget is blown on that alone.

[Draco18s]

* Draco18s has still never even seen a H/K

[Eternaly_Lost]

Oh, I'm sure your machine occupies a very special place in their cosmology

Yea, they love making it crash.

On that note. I did get a crash with 7.017, but it did not say that it was an out of memory crash, so I am trying again to see what happens this time.

I guess the Spire fleet will have to put up with taking over the Galaxy a second time in row. All those AI ships are not going to kill themselves.

[Aklyon]

* Draco18s has still never even seen a H/K

Turn on Hunter plot 4/10 and go meet the SF in an hour or so, you'll get to see one for free!*

[RockyBst]

These waves are officially more punishing than the two spire city, 7 golem fallen spire exos. Primarily because they beeline for the command centers and laugh at force fields. 9 battleships will happily melt my current spire fleet, even cruisers can't stand up to them for long.

The biggest problem though is the complete lack of salvage. I 'let' a H/K through onto my homeworld earlier, netting about 2 1/2 million salvage. For a golem, you seem to get about 5 million. On an average fallen spire exo you're looking at 50+ million salvage, albeit at a 8-12% collection rate (mil 2/3 stations). The champion ships, on the other hand, drop practically nothing. It's getting to the point where I'm waiting for exos just to rebuild from the waves 

EDIT: Huh, just had an out of memory crash. Looks like it maybe fell over trying to auto-save. I didn't kill THAT many AI ships, surely? Crash log attached.

[Toranth]

So, the new reinforcements model.  I think there's a problem in the budgetary spending - maybe the AI was programmed by US economists?

Frequently, it will spend several hundred points during the initial Command Station reinforcement pulse to purchase a starship, putting it massively negative.  Then when each guardpost's fleetship reinforcement cycle arrives, there's no budget left to actually buy ships.

In the example below, a Mk IV world bought a Mk III Zenith Starship, then no fleetships were ever bought again.  Not just on this world, but on any other world during the reinforcement cycle either.

Code: [Select]

= Doing central reinforcement pulse (this will go to the command station if there is one, and pick another spot if not)
pulseFleetShipStrengthBudget = 1.5 * AIDiff * handicapMultiplier = 13.5
pulseFleetShipStrengthBudget += 4 + random(0,AIDifficulty) = 19.5
pulseFleetShipStrengthBudget += ((AIP/10) * (AIDiff/10)) = 23.28
pulseFleetShipStrengthBudget *= 0.14 = 3.26
pulseFleetShipStrengthBudget must be at most 35; = 3.26
pulseFleetShipStrengthBudget *= MultiplierFromHumanHomeworlds = 3.26
pulseFleetShipStrengthBudget *= (1 + guardPostReinforcementPulsesShiftedToCentralPulse) = 3.26
pulseFleetShipStrengthBudget *= player.AITypeData.ReinforcementsMultiplier = 2.93
player.AITypeData.StarshipBudgetMultiplier: 0.25
= pulseStarshipBudget = pulseFleetShipStrengthBudget * player.AITypeData.StarshipBudgetMultiplier = 0.73
maxStarshipsAtPlanet = planetReinforcementAITechLevel * 3 = 12
humanRollup.EnemyStarships.Count is 7 which is under the max, so proceeding
runningStrengthBudget += pulseStarshipBudget = 0.73
called Helper_PickUnitsForReinforcement with: FlagshipV, AIRaidStarshipIII, LeechStarshipIV, DreadnoughtIV, BomberStarshipIII, ZinthStarship, SpireStarship
** Picked
1 ZinthStarship @ 576 each = 576
After picking starships, runningStrengthBudget = is now -575.27
Also, notice how this AI is spending 576 points when the budget is 0.73 points... that's a very large overdraft there.

Trim a bit, then get to the end of this reinforce cycle:

Code: [Select]

SendReinforcementsToPlanet result: happened, remainingReinforcements = 0, StrengthBudget = -496.03

totals for planet 41 (Yoquonvo); count: 2, strength: 864
totals for planet 32 (Mirhungmic); count: 1, strength: 192
totals for planet 79 (Wepverslov); count: 1, strength: 192
totals for ZinthStarship; count: 1, strength: 576
totals for AIGuardianRaiderIII; count: 1, strength: 288
totals for AIGuardianRaiderII; count: 2, strength: 384
== grand totals; count: 4, strength: 1248
Instead of buying a bunch of ships, it bought 1, plus the three guardians that come out of a separate budget, and still ended up with a -496 budget (still hadn't paid back 86% of its initial spend).

I suspect this is not intended behavior - the complete failure to purchase any fleetships, that is.