Minor fixes to Union AI and some events.
Good night š
Archive for September, 2012
Among all reports about unbalanced battle results, I ‘ve read this explanation by a beta tester:
What you should remember, though, is that :
– You wonāt beat legions (or Caesarās veterans, or Sulla’s crack troops) by number alone, you will beat them by number IF you can find a place where you can bring your superior number AND a leader with a strong skill
– Clear terrain with no rain is a must if you are playing number against quality
3 remarks:
– Varo’s legion defeat in 9BC? Terrain was at least heavily wooded. Yet it was a large-scale ambush, but I highly doubt the current AJE ambush system to compensate bonuses legions will get from reduced frontage. Unless they give Arminius a bunch of abilities compensatingĀ that ( superior frontages). But in this case, these bonuses will apply too for unrestricted terrains;
-I don’t understand how heavy infantry forming legion should prefer to fight in wooded/hill regions when the real battles occurred on flat ones often, just because heavy infantry never feel easy in restricted terrain…
– A simple designer tip: count on the AJE map the number of clear regions in Italy, Spain, Greece, Illyria…Pretty fascinating.
Conclusion: frontage is an important notion.It doesn’t make all for historicity. The main advantage of the legion was its flexibility, allowing rotation of lines,Ā and then greater sustainability in combat against formations less agile and more quickly exhausted. Add to this a 100,000 men Army wasn’t maneuverable due to the poverty of C3 at these times, and so unable to maneuver on more than 2 to 3 kilometers front wide at best. I begin to think the current battle system in AJE is just working the contrary of reality.
In September 1861 Union has launched his first amphibious operation around Mobile, Alabama.
No one of my events has been activated and this first incursion is blamable on many points. However, it verifies my first guess: for amphibious operations Union AI needs fleets. As it doesn’t know to build fleet, this point hasd to be solved first…
Let’s see in the next months how my events will perform.
1) SVF 2.0: I should release a quickfix tonight updating a few events. Union and CSA AI are now under testing by me, and you….I expect to update the mod once a week. The AI and design features are in, now bug fixing, improvements to AI and game balance will depend of the reports and their rate.
2) ROP AI design will begin the next week. Expect release at worst in the first days of November 2012. We will then be 2 to be able to build better AI experience.
3) FY) This week; the new manual. In a few weeks, it should be available a FY 1.075, incorporating the new guerilla model of SVF 2.0 and a better representation of the role of the cavalry during operations of pursuit after battle.
Later the 1.08 version of FY will deliver an enhanced diplomatic model, with Finns, Freikorps, Balts, Galicians, Poles, Anarchists, and possibly Trasncaucasian factions. The building of factories will be depicted too possibly by regional decisions rather than options.
This version is fully compatible with on going games. It fixes a few bugs and makes a few changes to Union AI.
A significant error in the files in the last version ( today) of SVF 2.0 has been fixed by a new version, fully compatible with games begun with the last version. Sorry for this.
This new version should improve Union AI economic performance. Open beta still.
I should release Sunday a new version of SVF 2.0 beta, tacking the Union AI economic trouble.
I’m yet hopingĀ to release the enhanced FY manual š During the next months, I will update FY to include new partisan system.
BTW, FY being now very mature, It will be from now released with a password. You will have to ask one at fatalyearssup@yahoo.fr to open the zipped file. I’m expressing all my gratitude toward all who have helped me to achieve this hard task, that is inspiring some of the core AGEOD team:
http://www.ageod-forum.com/showthread.php?26092-Dream-scenario-for-RUS
Those beta testingĀ a game for several weeks are the real heroes š Those who are playing SVF 2.0 currently will be a very important part of the success of this mod too š
ROP work should begin next week Expect a open beta before the beginning of November.
BTW, I’m certainly not so away from the truth about AI agents:
http://www.ageod-forum.com/showthread.php?26159-Unlocking-Allies&p=249012&viewfull=1#post249012
We’ll see if some Numidian units will not go yet however in Italy…
thanks to the reports, I’ve done new changes for Union AI. One of its main remaining problem is tied to the abysmal use of economic and conscription options by the AI. I’m not yet to the FY level where the AIs pay the largest part of the new units, but the current SVF 2.0 beta will allow to determineĀ how the AI performs with the right number of units, before tweaking it to force Union AI to spend what’s required.
Reports, the best form of support.
Interesting case, as any player is waiting the AI to attack on Manassas or at least march toward Richmond.
Should it be always the AI stance? No, because anything predictable may become a trap, especially for AI which is deprived of any real memory of the last games. Let’s add the AI is more right in its computation of forces than Union high Command in 1861 and is then very reluctant to fight the Beauregard’s Army.
SO Union AI in SVF will adopt very different behaviours during the Summer months of 1861. It may attack, bypass, stay immobile, target smaller CSA stack in the Shenandoah or yet switch between these plans… I purpose, CSA player can’t be sure of Union AI actions.
Example: in August 61, seeing Union AI yet immobile, I decided to launch Beauregard’s Army in Maryland by Loudun.
But these time, McDowell moved the same turn at Loudun, and the battle turned to Confederate defeat
Union AI doesn’t read my moves before computing its own. But looking at the files, the move was indeed a defensive one to protect Loudun. It could have been otherwise and turned into major Confederate incursion. However, the main benefit for me remains I can’t know what my AI will do the next turn. The simple possibility of such a move forces to plan it a bit better and prepare to the unexpected, to not say the worst.
Of course, such AI moves must follow a scheme avoiding grossly errors, like an undefended capital.
Here the situationat the end of the turn:
As you can see, Harper’s Ferry and Washington are conveniently garrisoned, preventing easy exploit of an hazardous McDowell movement.
Let’s look a bit larger:
Union AI is present in force in WV. I’ve taken one turn before Tyler region, but the next turns Union AI counterattacked here thanks to the forces present in WV. Assault has been repulsed, but my supply line is very assured, to say the last.
When I say AI designing must precede and not follow game design, I’m sure some are just thinking about like a sort of madness. There are generally thoseĀ wondering why the AI performs badly. Not because the AI is stupid. The AI is stupid indeed, but few know AI is playing with penalties as many rules remain unknown to it. It’s easy to build events based on capture or defense of an objective, and rewarding this with NM or VP. The art consists the AI able to cope with. If the AI doesn’t, it will play with a penalty.
That’s why I built first Union AI, tested it, before implementing today the rule giving VP for UNion when controlling WV. Because I know now Union AI able to face this new rule.
Modder Corner 