Poll

How are AI Reinforcements in 5.063+?

Too little!
0 (0%)
Too much!
7 (70%)
Just Right (watch out, the bears may be coming home)
3 (30%)

Total Members Voted: 0

Author Topic: Poll in light of Recent changes to AI Reinforcements  (Read 4255 times)

Offline keith.lamothe

  • Arcen Games Staff
  • Arcen Staff
  • Zenith Council Member Mark III
  • *****
  • Posts: 19,505
Poll in light of Recent changes to AI Reinforcements
« on: August 18, 2012, 02:56:48 pm »
A few months ago we did a thorough investigation of the AI reinforcements logic and found it was... well, pretty crazy.  It worked well enough for the most part because we all got used to the result, but there were various problems (notably it getting way more low-cap ships than it should, which I implemented several bandaids to prevent).  So it was made more sane logically.

Over the intervening months, however, it became clear that "sane" produced "very, very little defense" as many players have been rampaging all over AI territory with relative impunity simply due to not finding many ships in there.  So over the past few weeks I've made some other changes to put some meat on the bones.

But I haven't seen a lot of concrete feedback on reinforcements since those changes, and just got a mantis reports of reinforcements being "out of hand".  Single anecdotal data points don't help me a lot, but it did remind me that I need feedback on this, preferably sooner than later so there's time to get things ironed out before 6.0 (which will probably hit in early October with the official AS release).

So, three things would help:

1) Vote your general opinion based on recent play.

2) If possible, find a gameplay situation (i.e. savegame) that exemplifies why you formed that opinion, turn on Reinforcements Logging on the Advanced tab of the Settings window, play in that gameplay situation for at least 15-20 minutes (so some reinforcement cycles actually happen) and post those logs (two files in the RuntimeData directory, they'll have Reinforcement somewhere in the names).  That will help me make sure I've got the relevant details and something like apples-and-apples comparisons (or at least know when it's different, like one player playing vs Turtle, and the other vs Mad Bomber).

3) Post your reasons for your opinion here.  If you don't have the logs that's fine, but generally anecdotal data is going to weigh less with me.

Thanks :)
Have ideas or bug reports for one of our games? Mantis for Suggestions and Bug Reports. Thanks for helping to make our games better!

Offline Cinth

  • Core Member Mark II
  • *****
  • Posts: 2,527
  • Resident Zombie
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #1 on: August 18, 2012, 03:36:43 pm »
I really don't have an opinion either way. I take a lot of time to clean off AI planets and really kill its ability to reinforce many of its planets. Many times the only planets mid-late with any significant numbers of AI ships are major crossroads with lots of wormholes. What I do have are some long ... logs. I've had them enabled over the last week. 10/10 Bully assassin, 10/10 double vanilla, 1/1 sandbox game, and a 10/10 double turtle. In all these games I have no less than 4 planets constantly on alert. And yep these are all with my standard setups. I do have some saves from most of these games too.


 :o files are too big (even zipped) to add here  ::)
Quote from: keith.lamothe
Opened your save. My computer wept. Switched to the ST planet and ship icons filled my screen, so I zoomed out. Game told me that it _was_ totally zoomed out. You could seriously walk from one end of the inner grav well to the other without getting your feet cold.

Offline Hearteater

  • Core Member
  • *****
  • Posts: 2,334
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #2 on: August 18, 2012, 03:42:56 pm »
I'll come back to respond to this once I see how reinforcements work on a higher AIP in my current game.  At < 20 AIP on difficulty 9 I'm finding them just fine.  Certainly nothing to ignore, but not something that becomes impossible to break even after an hour on alert.  However, I do need to be very careful of Hybrids.  I don't know if your changes impacted them, but I had a false start in my current game where a Mark IV next to my homeworld got unbreakable due to insane hybrid buildup at just 30-40 minutes.

Offline keith.lamothe

  • Arcen Games Staff
  • Arcen Staff
  • Zenith Council Member Mark III
  • *****
  • Posts: 19,505
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #3 on: August 18, 2012, 03:49:05 pm »
:o files are too big (even zipped) to add here  ::)
You have text log files too big, when zipped?  These should compress very well  ... You have frightening gamestates, sir.

If you like, go into the log and chop off anything from before your last session, the timestamps should tell you when entries were made.

Or just delete the logs and play some more, then submit what comes from that ;)
Have ideas or bug reports for one of our games? Mantis for Suggestions and Bug Reports. Thanks for helping to make our games better!

Offline Cinth

  • Core Member Mark II
  • *****
  • Posts: 2,527
  • Resident Zombie
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #4 on: August 18, 2012, 03:56:38 pm »
:o files are too big (even zipped) to add here  ::)
You have text log files too big, when zipped?  These should compress very well  ... You have frightening gamestates, sir.

If you like, go into the log and chop off anything from before your last session, the timestamps should tell you when entries were made.

Or just delete the logs and play some more, then submit what comes from that ;)

Its from the 12th thru about 1 am this morning. I'll see if I can chop out the double turtle (last game)
Quote from: keith.lamothe
Opened your save. My computer wept. Switched to the ST planet and ship icons filled my screen, so I zoomed out. Game told me that it _was_ totally zoomed out. You could seriously walk from one end of the inner grav well to the other without getting your feet cold.

Offline Toranth

  • Hero Member Mark III
  • *****
  • Posts: 1,244
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #5 on: August 18, 2012, 03:59:02 pm »
Actually, I was getting ready to prepare a post on this topic myself.

I voted for too much, for two reasons.  One HUGE reason, more than anything else:  The higher reinforcement level in multi-HW games makes the Unity engine unhappy.  In my current 9/9 (Golemite, Radar Jammer) 10 HW, 8 Champion game, I need to shut down and restart AI War about every hour of game time (2-3 hours real time).  Otherwise I start getting Out of Memory errors.  Most unfortunately, they usually don't show up until I try to save.  I know this is an engine issue, and a rather extreme case on top of that.  But it is still a problem.

Issue #2 is also engine related.  Because there are so many more ships, large battles occur more often.  When I've got 5000 ships duking it out with a few thousand AI ships all in one system, the ship targeting and control falls way behind.  In the attached game, ("Bug Report - Too many units, defenders not attacking (Tsih).sav") there is a large battle going on in the "Second Star" system.  However, if you switch to the Tsih system, you'll see 150-ish enemy units running around, attacking my stuff.  There are 20 fortresses and 100+ turrets all within range of the enemy units, but they are not reacting.  They fail to react at all until the battle in the "Second Star" system drops to a small number of enemy ships.  I'm not sure exactly when the ships spawned in the Tsih system, but it's a total of at least 30 seconds of pure passivity by the fortresses and turrets.
I've seen this before - in some of the more absurd Homeworld or Hacking responses.  Under normal circumstances, this didn't come up.  With the new reinforcement logic (it looks like a x6.75 strength increase?) there enough units that I'm running into this problem on a regular basis.

Edit:  As an idea for handling this, would it be possible for the ship targeting thread to run even while the game is paused?


Now, back to the original intent of your question.
Gameplay wise, I'm doing fine.  There've been issues at the very start of the game, when a Mk IV world next to the homeworld gets 100 reinforcements during my initial clearing attack, 5 minutes into the game... but that's (sometimes) survivable, and triggered by me anyway.  Most systems quickly get to 500-800 ships when alerted.  Even a completely neutered system, without even any wormhole guardposts, was getting 100-150 ships.  One Mk IV world received over 400 ships at once.  The net result of this is that when I look at a world I'm about to attack, I can't just say "Oh, this world has 1000 ships, and I know my fleet can takes that."  I also need to say "I know that there will be at least 1 reinforcement during the battle, possibly two.  Do I have enough fleet to take on the original 1000 plus 1000 reinforcements?"
Still doable, but...  annoying.

I've attached the complete Reinforcement logs for my current game (mentioned above) - it's about 6 hours of gametime.  The file was too large, and has been zipped for convenience.
« Last Edit: August 18, 2012, 04:04:02 pm by Toranth »

Offline Draco18s

  • Resident Velociraptor
  • Core Member Mark V
  • *****
  • Posts: 4,251
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #6 on: August 18, 2012, 05:09:44 pm »
I'm going to throw a null vote because it is my game that triggered the mantis report.

However, I do think that there are too many reinforcements, at the current game-state we've claimed...15 systems with another 4 neutral systems (80 planet).

To me this does not feel like a significant amount of player-owned space.  7 "core" worlds creating a backwater for our homeworlds, with a 3-planet arm reaching out to a Mk4 factory.  The remaining 5 (and 2 neutral since abandoned) owned systems are claimed for ARS or golems.   Of those, two have spire leaders and cannot be abandoned.  Two of the remainging 3 are within transport-jump distance of each other, allowing for transportation of ships back and forth if needed (I use warpgates to direct built units, but I still need "supply lines").  The last system is adjacent to the prior system, claimed in order to repair an artillery golem, and is 2 jumps from an AI homeworld (though there are still 2 ARS and the E-generators left to take out before it can be assaulted).

6 hops out in the other direction is a Civ Leader + ARS, adjacent to a second and third, and then a dyson sphere 1 hop out (with an ARS); raid engine adjacent.  2 jumps from that cluster is another CL.  4 more hops to another CL, adjacent to a CL, two hops from the other homeworld (which has one of the four co-processors, all four are still intact).

AIP is at ~470.

From this point, assuming the AI somehow turns completely passive and all of its mobile units completely self-destruct...

6 more command stations should be destroyed (Civ Leaders).
2 systems need to be claimed, one held indefinitely.
1 additional warp gate needs to be popped (in order to secure the civ leader + ARS system).
3 other systems need to be claimed to have transport capability from core space to second AI Homeworld
1 Data Center remains intact

Which means that the final assault is actually quite a ways off.  So how many ships does the AI have that it's currently threatening us with?  About 3000.  And in the time it takes us to clear out one of the planets where those ships are congregating, it fills up another spot.  We have to move the golems carefully, or we'll inadvertently flood 800+ ships into a system that can't handle it, despite two fortresses, turrets, and a golem.  The AI will simply hit the command station with vampire claws, while the rest of its fleet soaks the DPS of the fortresses and golem.

Our only other option is to drop warheads through the wormhole about every 10 minutes.

Ironically, the last wave announced was 125 ships (119 parasites, 2 blockades, and a handful of starships).
« Last Edit: August 18, 2012, 05:11:18 pm by Draco18s »

Offline Cinth

  • Core Member Mark II
  • *****
  • Posts: 2,527
  • Resident Zombie
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #7 on: August 18, 2012, 05:12:41 pm »
Finally got the logs chopped in half.  ::) Anyway, here ya go.

As far as the issues Toranth mentioned, I haven't had a memory error since the sprite pooling was removed. It takes an excessive amount of time (based on recent test games) to hit close to the neighborhood where it might become an issue for me. I have see slide shows with excessive amounts of units in a system, however combat calcs seem to still take place and the lag is primarily graphics related.
Quote from: keith.lamothe
Opened your save. My computer wept. Switched to the ST planet and ship icons filled my screen, so I zoomed out. Game told me that it _was_ totally zoomed out. You could seriously walk from one end of the inner grav well to the other without getting your feet cold.

Offline keith.lamothe

  • Arcen Games Staff
  • Arcen Staff
  • Zenith Council Member Mark III
  • *****
  • Posts: 19,505
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #8 on: August 18, 2012, 05:19:09 pm »
AIP is at ~470.

(...)

Ironically, the last wave announced was 125 ships (119 parasites, 2 blockades, and a handful of starships).
I... how?  How could it send fewer ships than it has in AIP? :)

Not that it matters to this discussion, I just don't know how it managed that, even on ultra-low caps (though I could just barely see it there, which would make that 3000-ship threat very insane indeed).
Have ideas or bug reports for one of our games? Mantis for Suggestions and Bug Reports. Thanks for helping to make our games better!

Offline Diazo

  • Master Member Mark II
  • *****
  • Posts: 1,717
  • I love/hate Diff 10
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #9 on: August 18, 2012, 05:40:39 pm »
I have yet to vote and I'm not sure I should.

I'm playing on 10/10 and find them to be decent at low AIP.

I have not really gotten AIP increased since the changes, keep starting new games for all the new AS stuff.

For reference, I'm making raids into Mk III and IV systems with a starship fleet, if there is less then 200 ships I won't lose a ship, up to about 500 I'll lose a starship or two clearing the system. Have not run into a system with more then that recently.

I'm planning to to take the current game to the end so I'll post my experiences at higher AIP once I hit it, but for now I'm going to say low AIP (10 to 20 or so) feels about right.

D>

Offline Wingflier

  • Core Member Mark II
  • *****
  • Posts: 2,753
  • To add me on Steam, click the little Steam icon ^
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #10 on: August 18, 2012, 05:41:13 pm »
I didn't vote because I tend to think that the current problem with reinforcements is less a question of quantity, and more a question of quality.

It doesn't really matter whether you're attacking any enemy planet with 100 enemy ships or 1,000, you deal with them both in pretty much the same way:  Send your whole fleet, retreat when necessary.  Repair, then send them back in.  Bonus points if they attack you so that your defenses can help clean them up. 

Larger enemy forces may require a tactical missile or Spire Martyr first.  If the planet has an Eye on it, you simply try to avoid alerting the horde by sending a small group to take out the guard posts, then you send your whole fleet.

In other words, regardless of what's on an enemy planet, the solution always up with:  Send your whole fleet.  That's rather boring, and the only real difference between 100 or 1,000 enemy units on an enemy planet is how long it takes to clear them all.

Balance using this method is tricky.  The only real factor at work here is time.  How much time does it take you to clear out important planets (or impending threats)?  Does it give you enough time to keep playing the game, or will you get to a point where you spend all your time (as in Draco's post) attempting to defend the constantly-growing enemy threat?

This is less a question of player skill, and more a question of AI difficulty and patience. 

I'd like to think we can handle AI Planet Reinforcements in a much more interesting and strategic fashion than "my horde vs. your horde" or balancing it around the time it takes to perform the task.

I think this can be accomplished in 2 ways:

1. Buff Guardians massively, reduce their spawn rates to compensate.


When is the last time you even noticed that a Zombie Guardian or group of Zombie Guardians was stealing your units?  When was the last time you had to redirect a large number of your forces to deal with a Vampire Guardian because it was literally healing faster than you could kill it?  When was the last time you were worried an AI Warp Gate Guardian would reach your Homeworld and spawn a massive raid?

The point I'm trying to make here is that Guardians in their current form are simply a force multiplier.  Their unique, individual abilities don't really shine through that much (except in extreme cases like with the Gravity Guardian or EMP Guardian), and you deal with them all in a similar way.

Zombie Guardian - send in your whole fleet
Flak Guardian/Lightning Guardian - send in your whole fleet
Heavy Beam Guardian - send in your whole fleet
Sniper Guardian - ignore until you're close, then send in the fleet
Tractor Guardian - send in the fleet
etc.

Guardians really only act like a force multiplier, and therefore are dealt with all in a similar way.  Enemy reinforcements are bolstered by these things, but they simply encourage the same kind of monotonous player behavior as the schizophrenic fleet ships spawned on an enemy planet.

What if we buffed Guardians to the point that you FEARED to activate them or ignore their presence?

Zombie Guardian - Attack it with ships that are immune to reclamation
Flak Guardian/Lightning Guardian - Attack it with heavier ships that can withstand the aoe damage
Heavy Beam Guardian - Separate your fleet into several balls, or risk taking massive losses
Sniper Guardian - Kill with long-range snipers of your own, or send a quick interception force to destroy it before you take heavy losses
Tractor Guardian - Attack it with ships immune to tractor beams or kill it before it can drag half your army away
etc.

The point is that if Guardians were a lot more powerful and interesting, you wouldn't WANT to put the entire enemy planet on alert.  If you alert all the Guardians at once, you would be in serious trouble.

Which brings me to my next point:

2. Make each planet favor a certain type of Fleet Ship archetype (Fighter, Bomber, Frigate, etc.)

Instead of each AI planet having a mess of everything, populating each AI planet with a specific type of threat would make countering each planet a lot more interesting.  If a planet was filled with Fighters, you could send primarily your Frigates to remove the majority of the enemy threat, then deal with the Guardians on your own terms without alerting ALL OF THEM (like you would if you sent everything).

In other words, the current problem with planet reinforcement mechanics is that to deal with the schizophrenic enemy threat of weak Guardians and random spawns, you have to send your whole fleet.  Adding a greater or lesser number to the enemy reinforcements doesn't fix the problem of enemy planets being easy or non-interesting to conquer, it just makes it take longer.  We're trying to put a band-aid on a bigger problem.


If dealing with the enemy force became more interesting and methodical, you would have a reason to send a certain type of fleet ship and not send everything.  With this method, the amount of reinforcements on each enemy planet would be significantly lower, but dealing with each planet would take much more thought and effort from the player.  I like to think these solves both problems at once.

I realize that this solution won't be very popular or well-liked, but once again, I don't think the problem lies in how many reinforcements the AI is or isn't getting, I think the problem lies in how boring and uninteresting it is to deal with said reinforcements.
"Inner peace is the void of expectation. It is the absence of our shared desperation to feel a certain way."

Offline Draco18s

  • Resident Velociraptor
  • Core Member Mark V
  • *****
  • Posts: 4,251
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #11 on: August 18, 2012, 06:31:34 pm »
I... how?  How could it send fewer ships than it has in AIP? :)

That's what we said.  I checked the computation logs, didn't see anything...bizarre.

Quote from: AIThreadWaveComputationLog.txt
8/18/2012 1:56:47 PM (5.063)
-----------------------------------
Starting CreateHomogenousWaveToPlanet at Game Time: 10:17:40 ; Player.AIType: Feeding_Parasite ; Player.AIDifficulty: 7.3 ; AIProgressionLevel: 447 ; AITechLevel: 2
WaveSize = MultiplierFromWaveInterval * MultiplierFromHumanHomePlanetAndChampionCount: 1.23 * 1 = 1.23
aiTypeBasedAIPIncrement : 0
since difficulty < 8, effectiveAIP = AIProgressionLevel
workingShips = ( effectiveAIP * player.AIDifficulty ) / ( 13 - player.AIDifficulty ) : 572.44
workingShips *= FInt.FromParts( 0, AILoop.Instance.AIRandom.Next( 800, 1100 ) ) : 475.59
workingShips = Max(workingShips,34 * handicap_multiplier) :475.59
Inside AdjustNumberShipsFromAIType, multiplier: 1.25
after AdjustNumberShipsFromAIType call, workingShips :594.48
numberShips = workingShips.IntValue :594

So it's going to spawn ~600 ships

Quote from: MainThreadWaveComputationLog.txt
8/18/2012 1:56:47 PM (5.063)
-----------------------------------
Performing first CheckWave with size factor of 0.61 on wave at Game Time: 10:17:41

CheckWave: populating count of ParasiteII with base magnitude of 689
numberUnits = kv.Value * this.WaveSize : 423.56
after applying the ship-type-specific cap multiplier (which includes the unit-cap-scale multiplier), numberUnits : 52.94
after applying UsefulnessInAIWaveMultiplier if any, numberUnits : 52.94
after applying Mark-based multiplier if any, numberUnits : 47.65
after applying at-least-one rule, numberUnits : 47.65
after applying difficulty-based multiplier (if <= 3 then 1, <= 4 : 1.5, <= 5 : 1.75, <= 6 : 2, <= 7 : 2.25, <= 9 : 2.5, <= 9.3 : 2.75, <= 9.6 : 3, <= 9.8 : 3.8, 10 : 4.5), numberUnits : 119.11
numberUnitsInt = numberUnits.IntValue : 119

CheckWave: populating count of LeechStarshipII with base magnitude of 2
numberUnits = kv.Value * this.WaveSize : 1.23
after applying the ship-type-specific cap multiplier (which includes the unit-cap-scale multiplier), numberUnits : 1.23
after applying UsefulnessInAIWaveMultiplier if any, numberUnits : 1.23
after applying Mark-based multiplier if any, numberUnits : 1.11
after applying at-least-one rule, numberUnits : 1.11
after applying difficulty-based multiplier (if <= 3 then 1, <= 4 : 1.5, <= 5 : 1.75, <= 6 : 2, <= 7 : 2.25, <= 9 : 2.5, <= 9.3 : 2.75, <= 9.6 : 3, <= 9.8 : 3.8, 10 : 4.5), numberUnits : 2.77
numberUnitsInt = numberUnits.IntValue : 2
it's a starship so only adding one rule : 1 (had you going there, didn't we)

CheckWave: populating count of ZenithStarshipII with base magnitude of 1
numberUnits = kv.Value * this.WaveSize : 0.61
after applying the ship-type-specific cap multiplier (which includes the unit-cap-scale multiplier), numberUnits : 0.61
after applying UsefulnessInAIWaveMultiplier if any, numberUnits : 0.61
after applying Mark-based multiplier if any, numberUnits : 0.55
after applying at-least-one rule, numberUnits : 1
after applying difficulty-based multiplier (if <= 3 then 1, <= 4 : 1.5, <= 5 : 1.75, <= 6 : 2, <= 7 : 2.25, <= 9 : 2.5, <= 9.3 : 2.75, <= 9.6 : 3, <= 9.8 : 3.8, 10 : 4.5), numberUnits : 2.5
numberUnitsInt = numberUnits.IntValue : 2
it's a starship so only adding one rule : 1 (had you going there, didn't we)

CheckWave: populating count of AIBeachheadII with base magnitude of 1
numberUnits = kv.Value * this.WaveSize : 0.61
after applying the ship-type-specific cap multiplier (which includes the unit-cap-scale multiplier), numberUnits : 0.61
after applying UsefulnessInAIWaveMultiplier if any, numberUnits : 0.61
after applying Mark-based multiplier if any, numberUnits : 0.55
after applying at-least-one rule, numberUnits : 1
after applying difficulty-based multiplier (if <= 3 then 1, <= 4 : 1.5, <= 5 : 1.75, <= 6 : 2, <= 7 : 2.25, <= 9 : 2.5, <= 9.3 : 2.75, <= 9.6 : 3, <= 9.8 : 3.8, 10 : 4.5), numberUnits : 2.5
numberUnitsInt = numberUnits.IntValue : 2

CheckWave: populating count of DreadnoughtII with base magnitude of 1
numberUnits = kv.Value * this.WaveSize : 0.61
after applying the ship-type-specific cap multiplier (which includes the unit-cap-scale multiplier), numberUnits : 0.61
after applying UsefulnessInAIWaveMultiplier if any, numberUnits : 0.61
after applying Mark-based multiplier if any, numberUnits : 0.55
after applying at-least-one rule, numberUnits : 1
after applying difficulty-based multiplier (if <= 3 then 1, <= 4 : 1.5, <= 5 : 1.75, <= 6 : 2, <= 7 : 2.25, <= 9 : 2.5, <= 9.3 : 2.75, <= 9.6 : 3, <= 9.8 : 3.8, 10 : 4.5), numberUnits : 2.5
numberUnitsInt = numberUnits.IntValue : 2
it's a starship so only adding one rule : 1 (had you going there, didn't we)

CheckWave: populating count of BomberStarship with base magnitude of 1
numberUnits = kv.Value * this.WaveSize : 0.61
after applying the ship-type-specific cap multiplier (which includes the unit-cap-scale multiplier), numberUnits : 0.61
after applying UsefulnessInAIWaveMultiplier if any, numberUnits : 0.61
after applying Mark-based multiplier if any, numberUnits : 0.92
after applying at-least-one rule, numberUnits : 1
after applying difficulty-based multiplier (if <= 3 then 1, <= 4 : 1.5, <= 5 : 1.75, <= 6 : 2, <= 7 : 2.25, <= 9 : 2.5, <= 9.3 : 2.75, <= 9.6 : 3, <= 9.8 : 3.8, 10 : 4.5), numberUnits : 2.5
numberUnitsInt = numberUnits.IntValue : 2
it's a starship so only adding one rule : 1 (had you going there, didn't we)

Wave total ships: 125
TypesForDirectAdd count by type:
ParasiteII => 119
LeechStarshipII => 1
ZenithStarshipII => 1
AIBeachheadII => 2
DreadnoughtII => 1
TypesForCarrierAdd count by type:
BomberStarship => 1

So this.wavesize = 0.61?

Offline keith.lamothe

  • Arcen Games Staff
  • Arcen Staff
  • Zenith Council Member Mark III
  • *****
  • Posts: 19,505
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #12 on: August 18, 2012, 06:47:15 pm »
That's what we said.  I checked the computation logs, didn't see anything...bizarre.
Yea, there's a few bits of the log missing there (I'm not sure how it got from numberShips = 594 to base magnitude of ParasiteIIs of 689) but between the low roll on the random step in the AI thread calc, the nearly-min-time wave (wave size of 0.61) and the ship-type-specific cap multiplier of 0.125 (parasites start at 0.5 on high caps, * 0.25 for low caps = 0.125), it makes sense that it would get that low.

On the original topic: sure, reinforcements could stand to be more interesting (and I do want to look at making specific posts and/or planets "specialize" to reduce the "every AI defense is the same mix" problem), here I'm primarily concerned with the overall scale of said reinforcements.  If they're off by an order of magnitude in either direction then the "flavor" is of reduced importance.
Have ideas or bug reports for one of our games? Mantis for Suggestions and Bug Reports. Thanks for helping to make our games better!

Offline Draco18s

  • Resident Velociraptor
  • Core Member Mark V
  • *****
  • Posts: 4,251
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #13 on: August 18, 2012, 07:20:41 pm »
That's what we said.  I checked the computation logs, didn't see anything...bizarre.
Yea, there's a few bits of the log missing there (I'm not sure how it got from numberShips = 594 to base magnitude of ParasiteIIs of 689) but between the low roll on the random step in the AI thread calc, the nearly-min-time wave (wave size of 0.61) and the ship-type-specific cap multiplier of 0.125 (parasites start at 0.5 on high caps, * 0.25 for low caps = 0.125), it makes sense that it would get that low.

I could send you the unadulterated logs, but I have a feeling that it's not actually a problem, just an anomalous result that is within the bounds of accepted normality.  Rare and weird, but valid.

Offline keith.lamothe

  • Arcen Games Staff
  • Arcen Staff
  • Zenith Council Member Mark III
  • *****
  • Posts: 19,505
Re: Poll in light of Recent changes to AI Reinforcements
« Reply #14 on: August 18, 2012, 07:33:55 pm »
I could send you the unadulterated logs, but I have a feeling that it's not actually a problem, just an anomalous result that is within the bounds of accepted normality.  Rare and weird, but valid.
I agree.
Have ideas or bug reports for one of our games? Mantis for Suggestions and Bug Reports. Thanks for helping to make our games better!