1.5.16 - is enemy cavalry bit too keen to funnel?

[Benedict151]

Finally managed to fight the 3rd battle in my French vs Arabs (chosen as there should be good numbers of cavalry to check how the AI handles reserves). This was a rearguard action for my French (the previous 2 battles all appeared to play through fine).

There were no real issues with this 3rd battle either but I am wondering if the recent AI tweaks to make the use of reserve cavalry more aggressive might be resulting in a bit of a 'funnel effect' in certain battles? The AI seems keener to push its reserve cavalry into funnels where the liklihood of success is low at best. Its very possible that this might always have been there and I've not noticed it before and I'm just seeing it now due to there being a) more cavalry around in the 'Wolves' DLC and a couple of atypical battlefields.
I previously mentioned this during the 3rd and final battle in my Byzantine vs Bulgar campaign fought in a mountain pass (using 1.5.15, see previous post where the AI threw its numerous horse through the pass) and it was apparent in this rearguard battle. I confess that my deployment did slightly shamefully err towards the "camping" (in part due to previous rearguard experiences!) but I certainly left one hanging flank as this screenshot might show. The right is 'anchored' on a wood but there is plenty of space to come around behind it.

French deplyoment

Arab v French rearguard start.jpg (1018.49 KiB) Viewed 2569 times

The Arab AI certainly tried to commit all its cavalry to battle but as 2nd screen shows (see 2nd post) it poured them in against the French shieldwalls resting on the woods (even the 2 Arab horse above the wood turned to fight in the woods rather than circling in behind), even a couple of enemy cavalry units circling around would probably have completely unhinged my position and resulted in a suitably heroic massacre (and there was time enough for the cavalry to go around and arrive before end of battle).
Of course one could say that the AI in both cases could not have known the flanks were open and so its quite right that it should funnel which is very fair (and one can see accusations of 'cheating' if the AI had gone for the outflank) but I suspect players would probably have tried this move ....

This is not a big thing and I've only noticed it in 2 out of 7 battles so far and these were in atypical battles/battlefields so the actual incidence of this is probably much lower. Perhaps the issue is (if there is one) that in some cases the AI would have been better not committing its cavalry reserve so aggressively? This would be in the case in the Byzantine v Bulgar battle, but not of course in this rearguard battle.

Just something to possibly ponder on?

regards
Ben

[rbodleyscott]

Finally managed to fight the 3rd battle in my French vs Arabs (chosen as there should be good numbers of cavalry to check how the AI handles reserves). This was a rearguard action for my French (the previous 2 battles all appeared to play through fine).

There were no real issues with this 3rd battle either but I am wondering if the recent AI tweaks to make the use of reserve cavalry more aggressive might be resulting in a bit of a 'funnel effect' in certain battles? The AI seems keener to push its reserve cavalry into funnels where the liklihood of success is low at best. Its very possible that this might always have been there and I've not noticed it before and I'm just seeing it now due to there being a) more cavalry around in the 'Wolves' DLC and a couple of atypical battlefields.

I think so. I don't think the changes would cause this, except in so far as previously the cavalry might have stayed back and done nothing at all.

Of course one could say that the AI in both cases could not have known the flanks were open and so its quite right that it should funnel which is very fair (and one can see accusations of 'cheating' if the AI had gone for the outflank) but I suspect players would probably have tried this move ....

The flanking cavalry are likely to bipass the flanks of the infantry, because they are scripted to go past before turning in.

The central (reserve) cavalry, however, have no obvious way for the AI to make them take a significant detour. (And which direction should they take?)

Just something to possibly ponder on?

Yes.

Thanks Ben.

[MVP7]

Does rear guard battle AI have cavalry reserves or flanking groups? At least on some occasions in non-rearguard battles before 1.5.15 I have had major enemy cavalry/chariot groups circle around major forest areas to attack my rear. I don't think I have had any such situations during 1.5.15 beta but I'll pay extra attention if I get any.

[rbodleyscott]

Does rear guard battle AI have cavalry reserves or flanking groups?

No