View Issue Details

IDProjectCategoryLast Update
0023826AI War 2Bug - GameplayDec 14, 2020 11:32 am
ReporterSuzera Assigned ToChris_McElligottPark  
Status feedbackResolutionopen 
Product Version2.602 Transports Fixed In Multiplayer 
Summary0023826: Global threat doesn't seem to go down unless you destroy threat ships on a planet you own after a conversion to attack
DescriptionIf I go out and destroy threat on nearby planets, the global threat meter does not decrease. If they snipe the CC, global threat will go back up (they get added back into global pool) and not decrease if you blow them up after since you don't own the planet anymore. Often when clearing adjacent planets, global meter increases a little because the non-threat ships get released and/or the AI assigns extra threat due to the attack, see screenshots.
TagsNo tags attached.

Activities

Suzera

Sep 28, 2020 10:39 pm

reporter  

before.png (1,667,974 bytes)
noescape.png (1,528,382 bytes)
after.png (1,626,417 bytes)

Suzera

Sep 28, 2020 10:42 pm

reporter   ~0058737

A save game too if you want it.
Autosave.17100.save (510,688 bytes)

Suzera

Sep 28, 2020 10:43 pm

reporter   ~0058738

It might only be when ships start running away they no longer give kill credit regarding threat (and they are very often running away). I am not sure if the global threat meter is slow to update or not sometimes.

Chris_McElligottPark

Sep 29, 2020 4:22 pm

administrator   ~0058759

Badger, I can take a look at this longer term, but do you have any ideas on this?

BadgerBadger

Sep 29, 2020 5:48 pm

manager   ~0058769

First, most of this threat is in the Hunter Fleet. There was a bug with the hunter which I have just fixed where it was making really bad choices for lurk planets, so possibly the problem is that the Hunter was doing too much chilling.

The "threat" number in the resource bar there is some sort of moving average over time (It's been like that since day 1, though I'm not really sure why it was done this way).

If I go and blow up threat with the "Scrub Foes" command then the numbers seem to respond correctly just fine (though they won't be immediately reflected because of the moving average thing).

I will note that at difficulty 9 the hunter fleet is getting lots of income from the base AI, so I could imagine that if you are only clearing out a few threat every so often, the AI is just getting Hunter at the same rate.

Suzera

Sep 29, 2020 5:55 pm

reporter   ~0058773

Sometimes if I crunch through a 200 strength sentinel threat fleet or a 50-100 hunter fleet the global number also does not seem to move. Unless the AI is immediately replenishing that as they get destroyed as well. I do notice that SOMETIMES the threat number goes down after I blow up a bunch of clearly threat ships. But it never seems to after they start retreat logic at least which is I wonder if something is happening with that there. Presumably scrub foes would destroy them before they start retreat logic. Going to see about using a scrub foes before and after retreat logic real quick.

Not counting warden->hunter conversions that immediately die, where the number does not seem to go up or down, and they also tend not to retreat.

Suzera

Sep 29, 2020 6:02 pm

reporter  

scrub grove.save (1,126,925 bytes)

Suzera

Sep 29, 2020 6:02 pm

reporter   ~0058774

I tried scrubbing the planet Grove in this save. It has 103 threat listed. Threat does not change on scrubbing, even after waiting a few game minutes. Screenshots before/after and save included.
scrubgroveafter.png (869,928 bytes)
scrubgrovebefore.png (1,581,855 bytes)

Suzera

Sep 29, 2020 6:03 pm

reporter   ~0058775

In this case, it seems like a lot of the threat fleet was busy moving places instead of waiting.

Suzera

Sep 29, 2020 6:09 pm

reporter   ~0058776

I tried scrubbing every hunter ship on a planet with more than 5 strength of hunter fleet and the global threat went up for a bit but went down after a few minutes. It was about 30 strength total the number went down by about that much. Though a lot of warden and praetorian fleet got caught in the scrub so there might be budget allocation shenanigans happening.

I don't know that a moving average would be the cause because CPAs bump it a dramatic amount immediately.

Suzera

Sep 29, 2020 6:11 pm

reporter   ~0058777

Based on that, destroying hunter fleet might be working, but sentinel fleet threat seems to not reduce if destroyed in at least some kind of reproducible scenario.

BadgerBadger

Sep 29, 2020 6:16 pm

manager   ~0058778

The game displays the highest threat number seen in the last bunch of seconds. So it will immediately go up but very slowly go down.

Lets say I have 5 data points for threat strength, from latest to most recent.
10 10 10 10 100

it will display 100 immediately.

If I have 10 10 10 10 0, then it will show 10 until all those 10s are cleared, so I'd need to wait 5 more interavals till its
0 0 0 0 0

Suzera

Sep 29, 2020 6:17 pm

reporter   ~0058779

Also counter to "most of the threat is in hunter fleet", in the diff9 game in that save using watch all, hunter fleet seems to account for about 15-20% of the threat, so that assumption might break on difficulties with higher base budget.

BadgerBadger

Sep 29, 2020 6:18 pm

manager   ~0058780

Per my logging the hunter accounted for 90 of the 150 or so threat strength.

Chris_McElligottPark

Sep 29, 2020 6:22 pm

administrator   ~0058781

I'd be quite happy to have that rolling highest number reduced by a lot in terms of how long it keeps a history.

What was happening, and why I think it was I who put that in there, was that the strength counting was producing inconsistent results that flashed by really fast, or perhaps it was due to battles. But I think actually it was something like every few seconds the number would drop dramatically lower, then reset to the proper amounts, etc.

If this is truly an issue in the strength counting code, then probably we need to actually see that. It may be that the rolling average is not needed at all anymore, thanks to improvements made since then. We don't see problems on individual planets in the galaxy map with numbers fluctuating, so I'd wager that whatever the old problem was has been fixed.

Suzera

Sep 29, 2020 6:40 pm

reporter   ~0058786

Is the global threat counter not in strength? Because there's 494 global counter in the save, and that seems to be about the total threat strength listed in the map. If the threat strength is 494, then your logging may not be accounting for something that the ui counts as threat if your logging says 150 strength. If the threat strength is supposed to be 150, then something is getting counted as threat that shouldn't be.

Trying a couple more things to compare vs scrub.

BadgerBadger

Sep 29, 2020 6:43 pm

manager   ~0058787

The counter says 150 in the save

Suzera

Sep 29, 2020 6:44 pm

reporter   ~0058790

One bit of info. Hunter fleet is about 30% of the total threat according the threat breakdown with all watch.
hunterpercent.png (35,120 bytes)   
hunterpercent.png (35,120 bytes)   

Chris_McElligottPark

Sep 29, 2020 6:46 pm

administrator   ~0058791

Maybe that's a count of units? I don't know. That is a really old piece of UI, compared to the rest of the UI.

Suzera

Sep 29, 2020 6:46 pm

reporter   ~0058792

I was looking at the second save I posted (scrub grove). So yeah, that makes sense.

If I scrub grove in the second save while leaving the game paused the threat does seem to go down around the expected amount, but I don't see the same effect if I leave the game running.

BadgerBadger

Sep 29, 2020 6:56 pm

manager   ~0058793

If I scrub grove on that save then wait about 30 seconds, the number becomes reasonable.

Suzera

Sep 29, 2020 7:06 pm

reporter   ~0058794

If I go knock down Grove with regular units I get about what I'd expect for global threat counter decrease if I pause immediately after clearing most of it. If I don't pause after it doesn't seem to be nearly as much, and the decrease seems to correspond strengthwise with attacks that start in that timeframe. I would expect that if I scrub 100 strength and then get attacked by 100, the threat should go down 200 and not just 100ish. Is that how it's supposed to work?

BadgerBadger

Sep 29, 2020 7:19 pm

manager   ~0058795

Last edited: Sep 29, 2020 7:19 pm

If I have the code stop using the MaxIntValueOverTimeList data structure then the threat values do fluctuate significantly. But using it causes the game to be very slow in responding to threat changes.

You can't actually "knock down grove" with regular units, since all the threat ships on it are going somewhere else; by the time you get there there's only a handful of units.

I don't see any evidence of code problems here except for the fact that shown values are significantly delayed, and I'm going to bow out of this discussion and leave Chris to deal with it.

Chris_McElligottPark

Sep 29, 2020 7:20 pm

administrator   ~0058797

Makes good sense -- thank you for all the help on this!

Suzera

Sep 29, 2020 7:21 pm

reporter   ~0058798

Alright, I got a run where few attacks happened for a while so it's cleaner. It seems like threat did go down from scrubbing grove, and the total decrease does about add up to grove's strength and the small attacks' total. It might just be that attacking a planet incurs enough extra threat so cleaning small threats may not reduce threat, and/or the ai kept deciding to pump hunter fleet through some of the grove scrubbing runs. Plus the threat counter making it unclear what's happening until 30 seconds later doesn't help.

Consider this a non-issue for now, other than maybe the global threat counter timer could be tightened up some. I'll keep an eye out for more oddities here just in case given the above information. I'm still not feeling 100% that there's nothing weird happening here, but I think I at least have a better idea of how to look for that if it is happening.

Suzera

Sep 29, 2020 7:22 pm

reporter   ~0058799

My "knock down with regular units" run I used watch-all and chased them down to adjacent systems to about 5-10 strength error since my ships were much faster.

Chris_McElligottPark

Dec 14, 2020 11:32 am

administrator   ~0059955

Is this still a thing in the latest versions? A lot has changed.

Issue History

Date Modified Username Field Change
Sep 28, 2020 10:39 pm Suzera New Issue
Sep 28, 2020 10:39 pm Suzera File Added: before.png
Sep 28, 2020 10:39 pm Suzera File Added: noescape.png
Sep 28, 2020 10:39 pm Suzera File Added: after.png
Sep 28, 2020 10:42 pm Suzera File Added: Autosave.17100.save
Sep 28, 2020 10:42 pm Suzera File Added: Autosave.17100.savemet
Sep 28, 2020 10:42 pm Suzera Note Added: 0058737
Sep 28, 2020 10:43 pm Suzera Note Added: 0058738
Sep 28, 2020 10:44 pm Suzera Description Updated
Sep 28, 2020 10:45 pm Suzera Description Updated
Sep 29, 2020 4:22 pm Chris_McElligottPark Assigned To => BadgerBadger
Sep 29, 2020 4:22 pm Chris_McElligottPark Status new => feedback
Sep 29, 2020 4:22 pm Chris_McElligottPark Note Added: 0058759
Sep 29, 2020 5:48 pm BadgerBadger Note Added: 0058769
Sep 29, 2020 5:55 pm Suzera Note Added: 0058773
Sep 29, 2020 5:55 pm Suzera Status feedback => assigned
Sep 29, 2020 6:02 pm Suzera File Added: scrub grove.save
Sep 29, 2020 6:02 pm Suzera File Added: scrub grove.savemet
Sep 29, 2020 6:02 pm Suzera File Added: scrubgroveafter.png
Sep 29, 2020 6:02 pm Suzera File Added: scrubgrovebefore.png
Sep 29, 2020 6:02 pm Suzera Note Added: 0058774
Sep 29, 2020 6:03 pm Suzera Note Added: 0058775
Sep 29, 2020 6:09 pm Suzera Note Added: 0058776
Sep 29, 2020 6:11 pm Suzera Note Added: 0058777
Sep 29, 2020 6:16 pm BadgerBadger Note Added: 0058778
Sep 29, 2020 6:17 pm Suzera Note Added: 0058779
Sep 29, 2020 6:18 pm BadgerBadger Note Added: 0058780
Sep 29, 2020 6:22 pm Chris_McElligottPark Note Added: 0058781
Sep 29, 2020 6:40 pm Suzera Note Added: 0058786
Sep 29, 2020 6:43 pm BadgerBadger Note Added: 0058787
Sep 29, 2020 6:44 pm Suzera File Added: hunterpercent.png
Sep 29, 2020 6:44 pm Suzera Note Added: 0058790
Sep 29, 2020 6:46 pm Chris_McElligottPark Note Added: 0058791
Sep 29, 2020 6:46 pm Suzera Note Added: 0058792
Sep 29, 2020 6:56 pm BadgerBadger Note Added: 0058793
Sep 29, 2020 7:06 pm Suzera Note Added: 0058794
Sep 29, 2020 7:19 pm BadgerBadger Note Added: 0058795
Sep 29, 2020 7:19 pm BadgerBadger Assigned To BadgerBadger => Chris_McElligottPark
Sep 29, 2020 7:19 pm BadgerBadger Note Edited: 0058795
Sep 29, 2020 7:20 pm Chris_McElligottPark Note Added: 0058797
Sep 29, 2020 7:21 pm Suzera Note Added: 0058798
Sep 29, 2020 7:22 pm Suzera Note Added: 0058799
Sep 29, 2020 7:27 pm Suzera Severity major => tweak
Dec 14, 2020 11:32 am Chris_McElligottPark Status assigned => feedback
Dec 14, 2020 11:32 am Chris_McElligottPark Note Added: 0059955