Thoughts on Spirecraft

[Commiesalami]

6) Attritioner: well, it IS good for creating threat, but just like with the Artillery, you can't usually see the results of its work with your own eyes. Who knows, maybe it's actually insane on defence of in a large battle!
Suggestion: don't know how good it is, but feels like it should stay the way it is for now.

Attritioner: I would agree that their primary effect is not very visible. 500 damage per second that they get at Mk. I is not very much. Heck, even the 2500 damage per second that they get at Mk. V is not very much. The stationary attritioners that the AIs sometimes get suffer a similar weakness.
However, their secondary effect, their ability to "echo" damage to spread it around is actually quite useful.

On Normal caps, the Attritioners can do a total ok 10,500 damage per second.  That's 5 * 300/s + 4 * 600/s + 3 * 900/s + 2 * 1200/s + 1 * 1500/s.  Additionally, they do a 21% "feedback" damage pulse on the death of any enemy ship in system.  That's 5 * 0.5% + 4 * 1% + 3 * 1.5% + 2 * 2.5% + 1 * 5%.
Recently, as part of my Hacking experiments, I did a SuperTerminal raid for 10,000 AIP reduction.  There were 500 ships Core spawning every second, and about 500 dying every second, leaving a constant total of about 5,000 ships in the system.  The full cap of Spire Attritioners was therefore doing about 5,000 * 10,500 = 52,500,000 damage per second (!).
Core ships average about 770,000 HP.  500 dying per second resulted in an additional 161,700 damage per ship, or another 80,850,000 damage per second (!!).  Yet because the "feedback" damage is divided equally amongst all ships present, it was doing 16,170 damage per ship.  Combined, that's 26,670 damage per second per ship.
Seems pretty incredible, right?  But this was under some of the most absurd circumstances you're going to find.  500 ships spawning per second, 5,000 total present... And a full cap of everything allied (including a quadruple Spire capital fleet) to keep the 500 kills per second going to keep the "feedback" damage high.

The problem is that, although the "feedback" numbers look high because of the high number of kills, the damage is so distributed amongst the large number of ships that it becaomes insignificant.  What's a one-time shot of 16,170 damage to a Core ship?  Yet, while fewer ships means the damage per ship is higher, it also means the total damage being distributed is significantly lower.  In the minimal case, 2 identical ships, killing one will do an instant one-time shot of 21% damage to the second.  That's 4 shots from a single Mk V Implosion Artillery - yet the Mk V Implosion only requires Adamantite, while the Mk V Attritioner requires Titanite.  Finally, Attritioners lack any sort of initial-strike (alpha) capacity, which means they do little to reduce the amount of damage your fleet takes.

Basically, I think the "feedback" damage is either too little to matter (in the case of large numbers of ships) or unneeded (in the case of small numbers of ships).  Combined with the fact that the Mk V Attritioner require Titanite and provides 15% of the attrition damage and 25% of the "feedback" damage, means that in general I find the Attritioners are lacking in anything but niche cases:  Grav Drill worlds, heavy heavy defenses, drawing threat from an entire system. 
Which is too bad, since I love the Attritioners in concept.

What kinda computer are you running if you dont mind me asking? Running that kinda task is …. intense to say the least.

[Ranakastrasz]

Why not make martyrs deal reduced damage to higher tiers of ships? Like 50% to +1 tier, 25% to +2 tier?

[zoutzakje]

Why not make martyrs deal reduced damage to higher tiers of ships? Like 50% to +1 tier, 25% to +2 tier?

noooo, you better not nerf my martyrs

[Toranth]

I did a SuperTerminal raid for 10,000 AIP reduction.

I think the AI just turned in its resignation.

Well, the likelihood of a "Short range Warp Jump" roll approaches certainty when you have 150-200 rolls, and all the warp jumps were spawning in just two locations.  This made it easy to relocate my defenses to where the ships were spawning from.  It would have been much harder if the spawn location had been truly random.

I could change the "feedback" damage to hit the first target with all of it, and if it dies deduct the amount done from the total and hit the second target, and so on.  I suspect this could be OP, and perhaps generating bug reports, but it could be entertaining.

As awesome as that idea sounds, I would agree that it seems overpowered.  Killing a Mk I H/K would be a lot of damage.  21% of 100,000,000 = 21,000,000 = 770,000 * 27 = instant kill of 27 Core fleet ships.
My biggest complaint about the Attritioners is actually just the asteroid cost.  Titanite and Adamantite are so hard to come by that I rarely get the chance to build Mk IVs and Mk Vs.  On top of that, I never dare risk them except against an AI Homeworld, because they're pretty fragile. 

What kinda computer are you running if you dont mind me asking? Running that kinda task is …. intense to say the least.

I've got an i5 2500k overclocked to 4.2GHz, and I'm running AI War off an SSD.  I was also running at +10 speed, and on the "Extremely Low" performance profile.  It still took hours.

[Lee]

Am I the only one who feels like the Siege Towers aren't that great? They seem pretty decent on paper, but I usually see them move out, fire a couple shots at their target, then sit there for the 15second reload time getting pummeled.

On the other hand, they make playing against the gravity drill AI bearable (slightly).

[chemical_art]

Maybe its just me, but I notice waves take a noticeable performance hit when there are multiple attritioners around...

and I still hear complaints about how they aren't noticeable...


Idea:

Make it rather then x damage per second, make it ( x * 3) damage per 3 seconds.

EDIT: This won't effect the damage mechanics I think, because the on death effect of extra damage occurs regardless of attritioner caused it or not. The effects of damage will occur in bursts, making it more noticable even if not more damaging, and it will reduce the load of cpus of attritioners by 2/3.

[Nodor]

I did a SuperTerminal raid for 10,000 AIP reduction.

I think the AI just turned in its resignation.

Doesn't the superterminal raise the floor?

i.e. if you have 10K AIP reduction, don't you also have a floor of 5k?  If not, why not?

[zoutzakje]

I believe that would give you a floor of 2k (since the floor goes up with 0.2 for every 1 point AIP increase), but that's still a lot.

[keith.lamothe]

Well, the likelihood of a "Short range Warp Jump" roll approaches certainty when you have 150-200 rolls, and all the warp jumps were spawning in just two locations.  This made it easy to relocate my defenses to where the ships were spawning from.  It would have been much harder if the spawn location had been truly random.

Good point, just fixed the randomness (I had made a silly mistake).

The main thing is that it needs to start trading strength for bigger ships rather than just capping it at 500, so that there's not such an artificial ceiling to the challenge output.

And if I made it able to bounce spawns to adjacent planets like the k-hack and sd-hack rolls, that would have gotten out of control kinda quick   Of course you could have just taken all the surrounding planets first.

[Toranth]

I did a SuperTerminal raid for 10,000 AIP reduction.

I think the AI just turned in its resignation.

Doesn't the superterminal raise the floor?

i.e. if you have 10K AIP reduction, don't you also have a floor of 5k?  If not, why not?

The AIP Floor capped at 300.  I went looking for info on that, and eventually found that this was a change made back in release 3.060.  Getting rid of or raising the AIP Floor cap (only for the SuperTerminal?) would probably do a lot to prevent people from using it too much.

[Hearteater]

If I'm doing my math right, assuming you build up your AIP as high as possible while still being able to ride a ST until both your current AIP and floor hit 300*, you can build up to 750 AIP before using the ST.  I think a higher cap, 500-800, is probably warranted.

* In other words, pretend there is no 300 cap on the AIP floor, and you want to get as much AIP as possible and then use a ST to get yourself back down to exactly 300 AIP.

[keith.lamothe]

Don't worry, by the time I've got the ST where it should be; if you get 750 AIP reduction out of it and can keep going you'll have earned the privilege.

[Toranth]

Maybe its just me, but I notice waves take a noticeable performance hit when there are multiple attritioners around...
[...]

Am I the only one who feels like the Siege Towers aren't that great? They seem pretty decent on paper, but I usually see them move out, fire a couple shots at their target, then sit there for the 15second reload time getting pummeled.

Curiously enough, I just finished an experiment trying to get some practical number on the effects of Attritioners, and the unit I compared them to was the Siege Towers.
This time I used cheats to (attempt to) control exactly what was going on.  I set the AIP to 1000, coated everything in GravTurrets, built a fleet and put them on FRD next to the only wormhole. Then I waited for a wave countdown to trigger.  At this point, I saved this as my default. 
Fought the wave 5 times with just fleet ships, recording each time how many ships remained.
Then I added a cap of Mk I Attritioners, repeated 5 times.
Then Mk I + Mk II Attrtioners, repeated.  Etc.

Here's the wave I faced:
AI Wave:
Cruiser => 1000
LeechStarshipIII => 1
BeamStarship => 1
TypesForCarrierAdd count by type:
Cruiser => 331
BomberIII => 1330

Here's my base fleet:
1000 Mk IV Fighters
1000 Mk IV Bombers
1000 Mk IV Frigates


The results were... confusing.  The differences between individual runs at the same level were larger than the differences made by adding ship types.  The randomness involved in FRD is so great, I was getting losses of 500 ships one run, then 100 the next.  Given enough time, I could do enough runs to reduce the variance to something reasonable, but for right now I'm stuck with sample size of 5.  Individual runs had a standard deviation of about 85.

In the end, I did a linear regression against mark included, and came up with:
Attritioners:
y = 39.2x + 2438, r-squared of .85, where y = number of ships surviving (of the initial 3000), and x = highest mark of the included attritioners (x=0 means none included).
Siege Towers:
y = 38.6x + 2436, r-squared of .76, where y = number of ships surviving (of the initial 3000), and x = highest mark of the included towers (x=0 means none included).
StdDev for these lines was about 75.

Statistically, I cannot declare that the number of type of Spirecraft included made any significant difference the results of the battle. 
More practically, I think the slope is about correct.  Each mark of Spirecraft was worth about 30-40 ships. 
However, the effects were vastly outweighted by the pure random factors involved.  FRD, targetting decisions, how ships appeared from the wormhole, etc.  I was really surprised by how much of a difference pure chance played in the results of these battles.


I've attached my spreadsheet in case anyone else feels like doing more or different analysis.

[keith.lamothe]

If you want a somewhat more consistent situation, I suggest a pile of human basic turrets versus a pile of AI fighters or something like that.  And then some turret vs bombers, and then some turret vs missile frigates, etc.

But very interesting in general

[PokerChen]

It might be wiser to work out what Attritioners can do along with Siege Towers. The splash-damage-from-kill effect is very non-linear in terms of its outcome on the battle. STs may not se so hot with the 15sec cool-down, but all of its 000,000s of damage is deal in the first seconds, which combines well with Attritioners to reduce the time taken to kill of other ships - more attrition damage -> etc.

IMO the main advantage of attritioners is that they reduce the effective HP for all ships.