Page 1 of 1
AI or change in execution in v28?
Posted: Tue Feb 17, 2026 4:19 pm
by joliverlay
I am playing the official nasty level 1 scenario against the AI. I have played this many times before using the older exe files (mostly 26a or 26b) and saw something I have never seen before with the latest beta, v28. Reinforcements sailing from Cambodia for the Singapore assault in December 41 were intended to go just past the straights next to Singapore (west side) and instead of going past Singapore they (3 unescorted convoys) all went south of Java instead of staying in the south china sea. Since I've not noticed this before in prior versions of the game I'm wondering of the routing was changed in the latest beta for AI controlled fleets before singapore falls. If this is a change the scripts for the scenario will need to be changed. Taking a path that is 10X longer and 3 TFs full of troops getting slaughtered ruins the game way to early. In the past I'd lose a few ships to the minefield, which I could live with in a "nasty" scenario (AI has plenty of ships), but loosing 3 TFs and the loaded troops is much worse than "running the minefield".
Re: AI or change in execution in v28?
Posted: Wed Feb 18, 2026 7:25 am
by Chris21wen
Can you clarify are you Japan and is it turn 1?
I've not played 28b yet only tested stuff on it so I don't know if the routines controlling TF paths have been changed. What I do know is from other versions those routines will make decsions based on what is known about the enemy. This is where the TF routining controls come in handy. If you want to send three unescorted trans TF passed Singas then you should tell the TF to ignore threats using the TF routing controls.
It is entirely possible that it is not these routines that have changed, but the scenario and/or AI scripts for the Allies. If they are now stronger then all the more reason for setting the routing controls to ignore threats.
Re: AI or change in execution in v28?
Posted: Thu Feb 19, 2026 3:13 pm
by joliverlay
In in late December, and I'm Allied against the Japanese AI. The TFs that send additional troops to Malaya from Cambodia that should sail around singapore and land on the w side of the penensula instead whet east of borneo and south of Java. I found them by accident (3 TFs) in late december and slaugtered them with the Dutch navy. So I restarted the turn as Japanese to get the TF IDs and then I opened the scenario from Day 1 as Jap to see where they were supposed to go. The Nasty 1 AI script sent then to malaya and I expect inteneded them to route just 1 hex south of Mayala through the straits to land. Instead the went East of Borneo between Borneo and the PI, sailed south of java between Java and Austraila and got hit by my fleet hiding well away from the action. There routing made no sense unless the system was not allowing Japanese to move throught the straits while Singapore is Allied. If is the case this script should be changed to direct them to land on the east side of the penensula (Mersang for example) rather than on the west side. It really ruined my game to ravage multipe TFS with lots of infantry and no escorts doing something so unrealistic deep in allied territory.
I know the AI is stupid, but I did not see this behavior in the nasty scenarios before v28 so I was wonderng if something changed. And
Re: AI or change in execution in v28?
Posted: Thu Feb 19, 2026 7:07 pm
by BBfanboy
Where is Force Z? If it was in the area of the Bandoeng Strait between Java and Sumatra the AI could have balked at the risk. You did not answer the question of threat tolerances.
Re: AI or change in execution in v28?
Posted: Thu Feb 19, 2026 9:13 pm
by zebrazwo
if you are playing scen 2, then the data and one version of the AI were updated

- scen002 v28.png (30.35 KiB) Viewed 648 times
If you wanted to be sure to have the latest AI, you would have had to delete all of the aei002-02 -16 so the game would only use the -01 version.
Re: AI or change in execution in v28?
Posted: Fri Feb 20, 2026 1:32 pm
by joliverlay
Thank you for the responses.
1. Force Z was out of the area returning to Capetown. It was not part of the issue.
2. I don't know what the threat tolerances were, they were AI controlled and I did not look at that when I peaked. I don't know if the AI even uses threat tolerances.
3. It was scenario 4, not 2, but I did note that two of those scenario files have "old AI' and two have "new AI" I will delete the old ones. Thanks for pointing that out.
I may restart and try again. I have considered playing both sides head-to-head to avoid this problem using another scenario, (bottlenecks) but thought it would be too much of a pain. I'm playing it 2X PBM now, but need something to keep me occupied between turns so I play the AI with my free time. Anyone have a good experience with solatire head to head play?
Thanks again.
Re: AI or change in execution in v28?
Posted: Fri Feb 20, 2026 3:21 pm
by RangerJoe
If you start over, you have to start from scratch since the AI script that was selected was saved and will continue to be used. One option is to remove all but one AI script, play a game, then exchange scripts for another game, then you can have a different game. You can go through all of the AI scripts that way.
Re: AI or change in execution in v28?
Posted: Sat Feb 21, 2026 1:27 am
by Sardaukar
IJ AI is giving me quite tough time in latest 1128, way more than in previous versions.
Re: AI or change in execution in v28?
Posted: Sun Feb 22, 2026 1:30 pm
by PaxMondo
Sardaukar wrote: Sat Feb 21, 2026 1:27 am
IJ AI is giving me quite tough time in latest 1128, way more than in previous versions.
Andy Mac still has his magic mojo!!!
"ALL HAIL THE ANDY MAC!!!!!"
