Author Topic: Per-planet Turrets vs Guard Posts  (Read 11680 times)

Offline Radiant Phoenix

  • Sr. Member Mark II
  • ****
  • Posts: 362
Re: Per-planet Turrets vs Guard Posts
« Reply #45 on: June 02, 2015, 07:38:07 pm »
Whoa there, hold up.  Wave size gets a bonus based on how few attack vectors there are?  Where can I find details on that?  I don't remember hearing about that anywhere.

(And is that player planets that can be attacked, or wormholes that can be attacked through?)
It's total number of human systems that can receive waves.  The number of wormholes doesn't count, unfortunately.

Here's the relevant portion of the 5.036 release notes:
Quote from: 5.036
On Diff 7+ the randomization of wave intervals (which feeds directly into step 5 of the wave size formula) has been changed to consider the number of planets that the AI can send waves against (and thus the probable defensive power of those worlds, i.e. chokepoints, and the viability or lack thereof of smaller waves): ## If there are 6+ wave-targetable planets it uses the normal (pre-5.036) formula of adding a random number from (AIDifficulty*-60) to (AIDifficulty*120).
## If there are 5 wave-targetable planets the random is from (AIDifficulty*-30) to (AIDifficulty*120).
## If there are 4 the random is from 0 to (AIDifficulty*120).
## If 3, from (AIDifficulty*30) to (AIDifficulty*120).
## If 2, from (AIDifficulty*60) to (AIDifficulty*120).
## If 1, from (AIDifficulty*90) to (AIDifficulty*150) (making it actually able to hit single-chokepoints harder than it used to).
## The upshot is that it makes the AI a little "smarter" about figuring out that it may as well not bother sending small waves against concentrated defenses and should just save up for stronger attack forces. Accordingly, since the AI on <7 is intentionally a bit dumb it doesn't get this new rule at all.

## Thanks to many players (including zoutzakje, Diazo, Wanderer, rabican, Orelius, chemical_art, and probably others I can't remember right now) for the past few months of 9+ playtesting and related suggestions. No good deed goes unpunished.

Sorry, it only goes up to 25% bigger waves - but the average wave size jumps a massive 100% with just one entrance planet.
That's not the whole formula, and the wiki is confusing

Offline Toranth

  • Hero Member Mark III
  • *****
  • Posts: 1,244
Re: Per-planet Turrets vs Guard Posts
« Reply #46 on: June 02, 2015, 08:12:28 pm »
Whoa there, hold up.  Wave size gets a bonus based on how few attack vectors there are?  Where can I find details on that?  I don't remember hearing about that anywhere.

(And is that player planets that can be attacked, or wormholes that can be attacked through?)
It's total number of human systems that can receive waves.  The number of wormholes doesn't count, unfortunately.

Here's the relevant portion of the 5.036 release notes:
Quote from: 5.036
On Diff 7+ the randomization of wave intervals (which feeds directly into step 5 of the wave size formula) has been changed to consider the number of planets that the AI can send waves against (and thus the probable defensive power of those worlds, i.e. chokepoints, and the viability or lack thereof of smaller waves): ## If there are 6+ wave-targetable planets it uses the normal (pre-5.036) formula of adding a random number from (AIDifficulty*-60) to (AIDifficulty*120).
## If there are 5 wave-targetable planets the random is from (AIDifficulty*-30) to (AIDifficulty*120).
## If there are 4 the random is from 0 to (AIDifficulty*120).
## If 3, from (AIDifficulty*30) to (AIDifficulty*120).
## If 2, from (AIDifficulty*60) to (AIDifficulty*120).
## If 1, from (AIDifficulty*90) to (AIDifficulty*150) (making it actually able to hit single-chokepoints harder than it used to).
## The upshot is that it makes the AI a little "smarter" about figuring out that it may as well not bother sending small waves against concentrated defenses and should just save up for stronger attack forces. Accordingly, since the AI on <7 is intentionally a bit dumb it doesn't get this new rule at all.

## Thanks to many players (including zoutzakje, Diazo, Wanderer, rabican, Orelius, chemical_art, and probably others I can't remember right now) for the past few months of 9+ playtesting and related suggestions. No good deed goes unpunished.

Sorry, it only goes up to 25% bigger waves - but the average wave size jumps a massive 100% with just one entrance planet.
That's not the whole formula, and the wiki is confusing
Oh, no, the whole formula it is not  :D  That's just step 5 in the wiki's description... It's also a little out of date, since waves now operate on a budget, not count scaling.  And I think some of the multipliers have changed...


Here's a sample from a game of mine:
Code: [Select]
= computation of normalWaveStrength:
FleetShipBudget = (base value, equivalent to a single mkI fighter on high caps) = 1
FleetShipBudget *= this.AITechLevel = 1
FleetShipBudget *= (FInt)Game.Instance.Options.TotalHumanPlayerPlusExtraHomeworldsCount = 1
FleetShipBudget *= this.AIProgressionLevelEffective = 193
FleetShipBudget *= player.GetHandicapMultiplier() = 193
preMinStep_DifficultyMultiplier = player.AIDifficulty / ( 13 - player.AIDifficulty ) = 2.25
FleetShipBudget *= preMinStep_DifficultyMultiplier = 434.25
FleetShipBudget *= player.AITypeData.WaveSizeMultiplier = 434.25
postMinStep_DifficultyMultiplier = player.GetDifficultySpecificWaveSizeMultiplier() = 6
FleetShipBudget *= postMinStep_DifficultyMultiplier = 2605.5
Because Diff >= 7 and entry_points = 13, entryPointMultiplier = 0.8
Because game time > 2 hours, using full entryPointMultiplier
normalWaveStrength *= entryPointMultiplier = 2083.89
normalWaveStrength *= this.WaitingBasedMultiplierForNextNormalWave = 2230.41
minimumStrength = 200

And here's the wave that budget actually produced:
Code: [Select]
Starting CreateWaveToPlanet at Game Time: 16:43:51 ; Player.AIType: Brutal ; Player.AIDifficulty: 9 ; EffectiveAIP: 193 ; AITechLevel: 1 ; IsSchizo: False
WaveStrength = 2233.24

= FleetShipBudget for this wave's fleet ships = 2233.24

percentChanceParasiteWave (from AI type) = 0.03
percentChanceBomberWave (from AI type) = 0.06
percentChanceStealthWave (from AI type) = 0.06
= actually chosen WaveType: Normal

AIP is between the current tech level's AIP threshold and the next tech level's, so promotionRatio (percent of fleet ships to be bought from next tech level up) = 0.92
Some strength budget remaining from previous pick phases (if any), so spend what remains on types from the Normal group
* called PickUnitsForWave with: Fighter, Bomber, MissileShip, ZenithBombardment, ZenithParalyzer, ZenithViralShredder, SpireGravityDrain
** Picked 109 Fighter @ 1.67 each = 181.65
** Picked 616 FighterII @ 3.33 each = 2053.28
= After picking fleet ships, FleetShipBudget is now -1.7

player.AITypeData.StarshipBudgetMultiplier: 0.08
= StarshipBudget = originalFleetShipBudget * player.AITypeData.StarshipBudgetMultiplier = 187.56
* called PickUnitsForWave with: LightStarship, Dreadnought, LeechStarship
** Picked 2 LightStarship @ 96 each = 192
= After picking starships, StarshipBudget is now -4.44


== Wave total ships: 727
TypesForDirectAdd count by type:
Fighter : 109
FighterII : 616
LightStarship : 2
TypesForCarrierAdd count by type:
== For a grand-total Strength value of 2874


Since it is complicated, I tend to just remember the few factors that the player controls during the game:
*  Linearly multiplied by AIP
*  Related, multiplied by Tech Level
*  Number of entry points

Everything else major is set in the lobby - Difficulty, AI type, HW count.

The random factors, such as the wave size multiplier, and the wave charge time (related variables) are basically unpredictable, and not really worth worrying about.  Just understand that when you double the AIP, you (more than) double the AI's wave strength.  When you hit Tech Level II, the AI gets another bump.  And when you have very few entry points, the AI gets another bump.

Offline Radiant Phoenix

  • Sr. Member Mark II
  • ****
  • Posts: 362
Re: Per-planet Turrets vs Guard Posts
« Reply #47 on: June 02, 2015, 08:26:46 pm »
Wait...
Code: [Select]
(FInt)Game.Instance.Options.TotalHumanPlayerPlusExtraHomeworldsCount
Does this mean that adding a champion-only player is almost strictly worse than having them be Normal+Champion? What's the formula for this number?

Offline Traveller

  • Jr. Member Mark III
  • **
  • Posts: 96
Re: Per-planet Turrets vs Guard Posts
« Reply #48 on: June 02, 2015, 08:49:39 pm »
I would be very surprised if that variable included champion players.  I think it means "if I am a normal player, and Bob is a normal player, and Bob has two extra homeworlds, then the value is four".

Offline Radiant Phoenix

  • Sr. Member Mark II
  • ****
  • Posts: 362
Re: Per-planet Turrets vs Guard Posts
« Reply #49 on: June 02, 2015, 09:37:49 pm »
So, I decided to follow Red.Queen's example, and use ILSpy (in a virtual machine, because I'm a Linux user)

The relevant code appears to be roughly this:
Code: [Select]
A = 0
for player in Game.Instance.players:
if player.GetShouldCountAsNormalHumanPlayerForSpecialDifficulty():
A++
B = 0
for planet in Game.Instance.planets:
if (!planet.IsObstacle && !planet.IsOffroad)
if (planet.WasOriginallyHomePlanet && planet.ControllingPlayer >= 0)
B++;

TotalHumanPlayerPlusExtraHomeworldsCount = max( A, (B - 2) )
Naturally, the "- 2" is to account for the AI players. (Interestingly, this suggests that if you somehow made an AI start with more than one Homeworld, that would up the difficulty...)

"player.GetShouldCountAsNormalHumanPlayerForSpecialDifficulty()" is, as surmised, only "true" for Commander players.

I'm not sure how adding new Commander players after gamestart works, or if it's even possible, but if you can, there might be some advantage to starting with one Commander with all the HWs, and adding all the other Commanders later.

Offline Radiant Phoenix

  • Sr. Member Mark II
  • ****
  • Posts: 362
Re: Per-planet Turrets vs Guard Posts
« Reply #50 on: June 02, 2015, 09:54:20 pm »
Okay, you can do it: [ESCAPE_KEY] -> Manage Players

The reasons I can think of:
  • This gives you more metal income per planet, while having the same difficulty, as long as Player 1 colonizes all the planets, because Player 1 multiplies income by number of starting homeworlds, while AFAIK, multi-players don't.
  • You get a higher cap on stuff, because Player 1 gets the full multi-HW cap, and everyone else presumably gets some amount of something.
  • The extra players might get "Orbital Command Station Home Core" and get even more income.
These are just based on my vague recollection of how multi-player and multi-HW work, though.

Offline kasnavada

  • Hero Member Mark II
  • *****
  • Posts: 986
Re: Per-planet Turrets vs Guard Posts
« Reply #51 on: June 03, 2015, 03:53:41 am »
You were never allowed to have lightly defended worlds before - CPAs, Exowaves, or even Counterattack posts or Warp Guardians sending an unexpected wave to one of the 10 turret systems would have been doom.  But now, you can defend that system.  Or not, if you want to save the Metal and Energy you'd otherwise use.
The point that paragraph was trying to make was that when you were forced to scatter your defenses, you became weaker everywhere, but the AI would still attack at full strength.

It's not a "point" you've been making. Before you had to scatter some defenses in order to not have worlds defended too lightly, let's say 50 turrets / planets and have a few chokepoints.



Now, you've got 300 turrets / planets and still a few chokepoints. But the AI got stronger waves and more ships.

Basically "light" defending became "300 / planets" instead of "50 / planets". The only people seeing a difference are people not using any per galaxy cap defenses like mine or grav turrets and so on. But nothing changed as basically waves and so on were increased accordingly.

In all of your answers you're forgetting that part. The player was boosted when that change came out but now the AI is boosted accordingly and we're back to square one.

The game just got MOAR stuff in it. Again, you're not argumenting in your favor here, but in mine. The AI still trashes low defense worlds. The AI still requires chokepoint as the main blocking strategy. The lattice still is harder than most map types because of the higher connections.



Normal waves don't normally have any special targeting... Are you thinking of Exowaves?  Those still do have the special targets, and will beeline to your uniques to kill them. 
Also, I'm surprised if you are regularly seeing Threat choose to attack your systems and lose.  I can't remember the last time a Threadfleet attack didn't win easily in any system where I couldn't directly interfere.  The AI does a pretty good job of calculating the strength you have available in or near the target system, and gives itself a hefty overage just in case on top of that.

An individual planet's maximum defensive strength did not go up with the per-planet cap change.  Your chokepoint was always going to be as strong, or stronger, than any current world.  By making it possible to have multiple "chokepoint-equivalent" systems, players are allowed to spread out, and allow more connections to AI space.  Because now, there is no large inherent advantage in hiding everything behind a single chokepoint.

Maybe I didn't explain myself properly. When I first played AI war, I spread turrets and when the AI came in my planet, whether threat, EXO or anything, it targetted different things in my system in order to maximize damage.

Now, I'm playing better and place turrets better. The AI calculates it's chances and thinks it can beeline to the most important target (command station, usually). However, whatever calculation it has been using failed at finding out that my placement of turrets, mines, grav turrets, tractor beams and so on multiplied exponentially the efficiency of those defenses. As far as I've been able to find, the AI actually can't take that part into consideration. So basically it's always choosing the same target.

Therefore it can and will be baited to suicide. As I stated in my post, the problem is alleviated at diff 9+ because there are multipliers to what the AI thinks it requires to win. But at lower diff level an efficient defense pattern will trick it into dying.


About chokepoints... It was already possible to have multiple chokepoints before. I don't see your point. Actually, the inherent advantage behind going to a single chokepoint is even higher than before because a single chokepoint spawns higher waves. MOAR SALVAGE.


Quote
I love hacking, but hacking both Advanced producers plus even half the Fabricators is impractical - It'll normally run about 650 HaP to get both Mk IVs plus a mere 5 Fabricators.  That doesn't include hacking ARSs, or K-raiding, or design download/corruption, or sabotage, or the SuperTerminal, or... anything else, really.  HaP is always tight, if you are taking advantage of it.  Saying it should have been the way to cover for the fact that it was impossible to protect your uniques seems like solving the wrong problem.  I like being able to do some much more, different, stuff with Hacking that I would hate if I had to spend it all on just acquiring a few 'capturable' uniques.

What are you speaking about ???

I just stated that hacking provided a solution to being able to "protect" advanced constructors and that the higher turret cap did not do it. If an exo targets your unique it's dead. I personally like the fact that you need to choose. But that's another subject entirely.

Quote
As for normal fortresses, I have to disagree with kasnavada.  If a planet gets overrun, your fortress will absolutely die

I tend to place them at chokepoints. Then, if it's overrun, I'm dead. Otherwise I agree with you actually - if you don't use them as a resource which spells your doom if killed, it's a waste of K and metal.

Offline Histidine

  • Hero Member
  • *****
  • Posts: 581
Re: Per-planet Turrets vs Guard Posts
« Reply #52 on: June 06, 2015, 03:35:12 am »
But the AI got stronger waves and more ships.
Wait, where? (Specifically, what buffs and when did the AI get that occurred after the turret change?)

Offline kasnavada

  • Hero Member Mark II
  • *****
  • Posts: 986
Re: Per-planet Turrets vs Guard Posts
« Reply #53 on: June 06, 2015, 04:25:44 am »
Check the patch notes. In AI war there is a constant power creep on both sides.

Offline Yavaun

  • Newbie Mark III
  • *
  • Posts: 40
Re: Per-planet Turrets vs Guard Posts
« Reply #54 on: August 21, 2015, 06:19:37 am »
http://www.arcengames.com/mantisbt/view.php?id=16821

Does anyone use the per-planet turrets in any way other than "pile them all up as closely as possible in the right spot"?

Yes. Yes I do. There is a lot of effeciency and fun to be gained from the right placement and I love to fiddle with these things. Please don't streamline what is essentially half the game (defending) into one autoplace-able structure! -.-