build 5517 w/ up to date \default directory content.
Some AI feedback.
AI feels medium to medium-easy difficulty wise
AI does build a lot of ships, and that's scary... but it doesn't seem to utilize that fleet strength well
AI does detect areas of enemy pressure, and positions sufficient fleet strength in appropriate positions opposite the enemy pressure.
- either that, or it's detecting choke points... but I get the sense it's more in response to enemy pressure
- this results in usually 2 or 3 system pairs where ships accumulate in large numbers across from each other. "stand offs"
- I've been breaking this by timing the production of something like 20 new ships, then suddenly massing most of my ships onto one of these enemy standoffs. This effectively halves the enemy's fleet strength, and then I run rampant, invading systems every 2-3 turns.
- The AI does not do this, despite the lack of pressure my re-positioning of ships causes. (perhaps it is holding onto a choke point that it deems valuable?). In the picture below it there are several (at least 3) relatively very underprotected systems by which to enter and pillage my empire that it does not take advantage of (aka: a trading bases strategy OR a draw me away from my invasion strategy)
- The AI also does not respond adequately to the invasion. It either needs to back off smaller offensive fleets (don't let me piece meal the AI) OR combine them to have sufficient numbers/firepower to oppose the the big fleet running around its back yard. OR it needs to take it's other large picket line fleet and come hunt my invading fleet.
- I will occasionally pick off small groups of ships, just to "trim the grass" - keeping the AI's ship count down - and the AI lets me do this too much.
-AI over builds Neutronium Forges; it only needs them on planets with shipyards (and I see the AI building them almost on all planets) and only if you've got a Neutromium Extractor AND only if you've got ship designs that utilize Neutronium Armor (and I'm not seeing ships that use that).
- While I think that offensive troop ships (aka: those with weapons) makes sense in the late game when there's a clear ship/tech advantage, I don't think an offensive colony ship makes sense (it's going to a place where there's no enemy presence). AI could probably recoup some PP by not investing in weaponry on colony ships, and waiting for weaponry on troop ships until much later ('cause you lose the whole ship when you colonize or invade - a loss of PP).
Picture notes
* Merope had ~140 blue AI ships holding ground at turn 135, when I brought to bear 90% of my ships on that location.
* Levitt also had ~140 blue AI ships at this time.
* Over the next 30 turns, I followed the 1 - 6 route, invading as a went. (the AI has apparently not researched all growth tech, as they've not colonized a number of planet types)
* The AI fleet at Levitt hasn't moved
AI Feedback (rev 5517+)
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.
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.
AI Feedback (rev 5517+)
Code released under GPL 2.0. Content released under GPL 2.0 and Creative Commons Attribution-ShareAlike 3.0.
Re: AI Feedback (rev 5517+)
Cool, it's very helpful to get thisyandonman wrote:build 5517 w/ up to date \default directory content.
Some AI feedback.
Ya, that's about where I rate it also. I have also seen a couple of requests for it to be toned down a little bit though, and I think I've figured out the basics of how to implement a pseudo-difficulty setting for the AI -- I'm thinking four categories: Turtle, Defensive, Aggressive, and Maniacal. Partly these would vary aggressiveness/risk-taking, as the names suggest, and the most defensive one might not research (or deploy) all the best offensive ship technology. The current AI is probably somewhere between Defensive and Aggressive, closer to the latter but I'm planning on making it more so. I do have some ideas on how to make it both more aggressive and more difficult for ManiacalAI feels medium to medium-easy difficulty wise
Yup, still working on thatAI does build a lot of ships, and that's scary... but it doesn't seem to utilize that fleet strength well
It's not yet detecting choke points. It had prioritized territorial protection & capture goals, and in pursuing that it is responsive to, but not driven entirely by, enemy pressure. There is certainly still plenty of room for improvement.AI does detect areas of enemy pressure, and positions sufficient fleet strength in appropriate positions opposite the enemy pressure.
- either that, or it's detecting choke points... but I get the sense it's more in response to enemy pressure
The accumulation of opposing forces in a temporary standoff seems right and proper to me; the question is all about when to try breaking it.- this results in usually 2 or 3 system pairs where ships accumulate in large numbers across from each other. "stand offs"
- I've been breaking this by timing the production of something like 20 new ships, then suddenly massing most of my ships onto one of these enemy standoffs. This effectively halves the enemy's fleet strength, and then I run rampant, invading systems every 2-3 turns.
Well, I think that if you pay attention you will see the AI breaking these now and then (it does have directives to do so) but it is probably still too defensive here & I'm trying to get it better at this in general.- The AI does not do this, despite the lack of pressure my re-positioning of ships causes. (perhaps it is holding onto a choke point that it deems valuable?). In the picture below it there are several (at least 3) relatively very underprotected systems by which to enter and pillage my empire that it does not take advantage of (aka: a trading bases strategy OR a draw me away from my invasion strategy)
Hmmm, it does have some flexibility to rebalance deployments, but you've convinced me that it needs better capability there.- The AI also does not respond adequately to the invasion. It either needs to back off smaller offensive fleets (don't let me piece meal the AI) OR combine them to have sufficient numbers/firepower to oppose the the big fleet running around its back yard. OR it needs to take it's other large picket line fleet and come hunt my invading fleet.
Hmm, good enough pathing planning to prevent you from doing this is a long ways down the road. A sooner goal would be to be able to do the same to you much better than it does now.- I will occasionally pick off small groups of ships, just to "trim the grass" - keeping the AI's ship count down - and the AI lets me do this too much.
That should have been fixed at least by the code I committed earlier today. The problem was that a 'not' had slipped into an evaluation of where to build the Forges, and it was only building them where it did NOT have shipyards (at least at the time it started construction of the forge) -- Ouch!! That's why you never actually saw ships using the armor - it does have the designs, & you should start seeing it show up now.-AI over builds Neutronium Forges; it only needs them on planets with shipyards (and I see the AI building them almost on all planets) and only if you've got a Neutromium Extractor AND only if you've got ship designs that utilize Neutronium Armor (and I'm not seeing ships that use that).
In my own games I use both bare colony ships and armed colony ships, because I'm better at assessing risk and guarding against it than the AI is. Although I do have the AI try to secure colonization targets first, it's still far from doing this well enough to really ensure the safety of the colony ship. I do carefully consider any nonessential components I put on. With a cheaper, single-unit colony ship I don't think I put on any more than a single mass diver unit, at cost of one PP. For the triple colony ships, which cost more and take longer to make, I do put on some heavier weaponry. The AI's colonization planning suffers from a greater time lag than yours does, and I think an extra 10% increase in ship cost is worth the extra reliability of the colonization effort.- While I think that offensive troop ships (aka: those with weapons) makes sense in the late game when there's a clear ship/tech advantage, I don't think an offensive colony ship makes sense (it's going to a place where there's no enemy presence). AI could probably recoup some PP by not investing in weaponry on colony ships, and waiting for weaponry on troop ships until much later ('cause you lose the whole ship when you colonize or invade - a loss of PP).
Hmm, they should be researching it all (well, currently not doing terraforming nor Gaia), & have it done well before they're building Neutronium Forges. I think the real problem is with my prioritization balancing -- I still don't have it working really well when the military forces all get scaled up really big. I have some thoughts on improving it I'll try to get implemented.Picture notes
* Merope had ~140 blue AI ships holding ground at turn 135, when I brought to bear 90% of my ships on that location.
* Levitt also had ~140 blue AI ships at this time.
* Over the next 30 turns, I followed the 1 - 6 route, invading as a went. (the AI has apparently not researched all growth tech, as they've not colonized a number of planet types)
Hmm, in that map it is not at all clear to me why a big AI force would stay stuck at Levitt. I'll certainly work on improving responsiveness to military movements.* The AI fleet at Levitt hasn't moved
If I provided any code, scripts or other content here, it's released under GPL 2.0 and CC-BY-SA 3.0