AI Sly getting nice supply reach by turn 50

Describe your experience with the latest version of FreeOrion to help us improve it.

Moderator: Oberlus

Forum rules
Always mention the exact version of FreeOrion you are testing.

When reporting an issue regarding the AI, if possible provide the relevant AI log file and a save game file that demonstrates the issue.
Post Reply
Message
Author
User avatar
Oberlus
Cosmic Dragon
Posts: 5715
Joined: Mon Apr 10, 2017 4:25 pm

AI Sly getting nice supply reach by turn 50

#1 Post by Oberlus »

Playing test version compiled from master (build 2018-04-02.02f4a84).

I'm in a game as human vs 9 AIs, two of them are Sly.

One of the Sly have completely overcome its no-supply range by turn 51 (maybe earlier, I wasn't paying attention to the map for the last 10 turns).

It has this techs researched:

Code: Select all

Turn02	The Physical Brain 
Turn05	Algorithmic Elegance 
Turn06	Planetary Ecology 
Turn09	Subterranean Habitation 
Turn13	Nascent Artificial Intelligence 
Turn14	Mass Driver 2 
Turn16	Planetary Bunker Complex 
Turn16	Fighters and Launch Bays 
Turn17	Robotic Production 
Turn20	Military Robotic Control 
Turn21	Orbital Construction 
Turn23	Mass Driver 3 
Turn25	Mass Driver 4 
Turn28	Microgravity Industry 
Turn34	Symbiotic Biology 
Turn36	Zortrium Armor Plating 
Turn38	Planetary Defense Network 1 
Turn40	Spatial Flux Drive 
Turn42	Fusion Generation 
Turn47	Orbital Generation 
So it does not have exobots yet (I do, haha).

Yet it has this nice supply reach on turn 51:
Sly_with_nice_supply_turn51.png
Sly_with_nice_supply_turn51.png (432.93 KiB) Viewed 1630 times
I find quite interesting how well is performing this Sly AI in this game. Not only regarding normal (non-stealthy) expansion, but also it has the best PP/turn and it's only slightly below the best RP/turn among all empires.

Alkes is the AI's homeworld (last seen turn 20). Rin alpha (last seen turn 17) only had those Ancient Guardians, so AI got no natives from there.
I assume ('cause I'm not enabling Super Testers) that it got some natives from the unexplored systems nearby Alkes.

But, I've checked the "Planets Invaded" Graph and only two empires (that I have not met yet; purple and blue, none of the Sly, which are yellow and ochre). Unless this graph does not show native planets invaded, this AI must have Sly population only. And it does not have Architectural Monofilaments (so no Space Elevators).

So, how can it have that nice supply reach?

I attach save game and all logs for this turn, in case @devs find this something interesting to look at.
Attachments
FO_AI_Sly_nice_supply_turn51_AI6-9_logs.zip
(2.21 MiB) Downloaded 117 times
FO_AI_Sly_nice_supply_turn51_AI1-5_logs.zip
(2.64 MiB) Downloaded 100 times
FO_AI_Sly_nice_supply_turn51_save&fo_logs.zip
(1.24 MiB) Downloaded 110 times

User avatar
Oberlus
Cosmic Dragon
Posts: 5715
Joined: Mon Apr 10, 2017 4:25 pm

Re: AI Sly getting nice supply reach by turn 50

#2 Post by Oberlus »

Oberlus wrote:Unless this graph does not show native planets invaded
I've checked in-game that this graph does seem to note native invaded planets (dark blue player, an Etty, invaded one Silexian planet next to my homeworld and next turn its graph showed its first invaded planet as well as a discrete increase in its "Colonies" graph; I then invaded that planet and saw the corresponding decrease in both graphs for that AI player).

Ophiuchus
Programmer
Posts: 3433
Joined: Tue Sep 30, 2014 10:01 am
Location: Wall IV

Re: AI Sly getting nice supply reach by turn 50

#3 Post by Ophiuchus »

Well if thats a Sly-only empire, a hidden gas giant in Rin alpha would even not work after moving the palace there. (orb.con +1, GG -1, bad supply -1, imperial palace +2). So the only conclusion is there are hidden Sly systems everywhere with a supply of 1 because of lots of imperial palaces uummm.

Or there is a dying Sly population on Rin alpha V with a palace (orb.con+1, bad supply -1, palace +2) and some other dying sly in e.g. alkaid beta.

Or there is a native species with >=5 supply planet beyond your detection range; thats really easy
Orb.Con +1 , Tiny planet +2, Imperial palace +2, Great supply +2
:D

If i look at the scripts actually there is no decrease for supply anymore (BUUUG!). So moving the palace should give some lasting impression. But i dont think the AI does move the palace.

This tells me on one hand I should relocate my palace more often for exploration and also that probably there is some confusion..
Any code or patches in anything posted here is released under the CC and GPL licences in use for the FO project.

Look, ma... four combat bouts!

User avatar
Oberlus
Cosmic Dragon
Posts: 5715
Joined: Mon Apr 10, 2017 4:25 pm

Re: AI Sly getting nice supply reach by turn 50

#4 Post by Oberlus »

Ophiuchus wrote:a hidden gas giant in Rin alpha[...]
From the explored systems in the screenshot, only Alkes (the homeworld) and Alkaid beta had GGs (two, one already colonised by the Sly when my scout got there in turn 19). Alkaid alpha and Rin alpha had no GGs.

But I think I got it. The Sly has outposted (or colonised, died out and revert to outpost) Rin alpha IV. Check the screenshot at turn 68:
Sly_nice_supply_68.png
Sly_nice_supply_68.png (410.17 KiB) Viewed 1615 times
Rin alpha IV has pop 0 and a supply of 2, constant over time. The "Predicted Next Turn 1.00" information is false or buggy, because what you can see in that screen stays the same for turn 69 and following.

I guess that's a bug in there? I mean, supply should be 2 and predicted next turn 2, or 1 and predicted next turn 1, right? No idea where that "Unknown +1.00" comes from.

User avatar
Dilvish
AI Lead and Programmer Emeritus
Posts: 4768
Joined: Sat Sep 22, 2012 6:25 pm

Re: AI Sly getting nice supply reach by turn 50

#5 Post by Dilvish »

Oberlus wrote:Rin alpha IV has pop 0 and a supply of 2, constant over time. The "Predicted Next Turn 1.00" information is false or buggy, because what you can see in that screen stays the same for turn 69 and following.

I guess that's a bug in there? I mean, supply should be 2 and predicted next turn 2, or 1 and predicted next turn 1, right? No idea where that "Unknown +1.00" comes from.
I think I have this partly figured out, but only partly. My best guess would be that the Predicted value being off is probably because it doesn't know the source of that Unknown +1, and so it doesn't know to include it in the prediction (the current meter value +2 is provided by the server). But the initial predicted value provided by the server probably should have been correct and so this would only be thrown off it the client were updating it, which it doesn't seem to me like it really should in this situation.

Also, the "Unknown +1.00" comes from the small planet supply adjustment (which in turn is specified in SHP_GAL_EXPLO.focs.txt). It's strange that the client can know the source of the Orbital Construction bonus but not know the SHP_GAL_EXPLO bonus-- it's not like the client does not know it's a small planet. And then if I either (i) research sensors and park a scout near their Capital, or (ii) build a supertesters takeover at your Capital to get the detection boost, then all of a sudden my client realizes that this extra +1 is from the SHP_GAL_EXPLO small planet adjustment. It should not normally take visibility of the enemy Capital in order for the client to properly process their techs (as it did not for their Orbital Construction tech). So this aspect does seem a bit buggy to me. If you would, please make an Issue with this savegame and pic
If I provided any code, scripts or other content here, it's released under GPL 2.0 and CC-BY-SA 3.0

User avatar
Oberlus
Cosmic Dragon
Posts: 5715
Joined: Mon Apr 10, 2017 4:25 pm

Re: AI Sly getting nice supply reach by turn 50

#6 Post by Oberlus »

Dilvish wrote:please make an Issue with this savegame and pic
Done. Issue 2064. Not sure if I made the right description, tho.

User avatar
Dilvish
AI Lead and Programmer Emeritus
Posts: 4768
Joined: Sat Sep 22, 2012 6:25 pm

Re: AI Sly getting nice supply reach by turn 50

#7 Post by Dilvish »

It's fixed now, at least this particular problem, which stemmed from superfluous activation clauses citing objects for which the human client did not have sufficient visibility and so it could not determine that the tech was applying. There is still some similar cleanup needed elsewhere, which I discussed in the Github Issue.
If I provided any code, scripts or other content here, it's released under GPL 2.0 and CC-BY-SA 3.0

Post Reply