Open Play?

[hmssws]

I was wondering if Open Play will be set up the same as the Mad Minute releases?

Particularly, I found it aggravating that the AI would not use all of its available troops except on meeting engagements.....Will there be an "All In" option on this?

Thanks,
Hank

[Hancock the Superb]

That is actually apart of the game (TC2M).

I have looked at the armyplays, orders, and the other sheet of strategies, and all the engine does is generate a random number, and use the events associated with that one (in OP). Often times, generals wouldn't pitch in the whole way.

Finally, it seems to me that most of the attacks happen from 30 minutes to and hour out, so in a 30 minute game, there isn't going to be too many complex manovers.

[norb]

The AI is it's own beast and as Hancock pointed out, it's based on many random factors. Holding some in reserve, etc. Now it might have had bugs because I don't think it was tested well enough, but we all feel that SOWGB will be a better experience.

[pcelt]

I used Open Play a lot --usually with the Long Time Option ---and frequently(though not always) found the whole or close to the whole of the enemy force employed against me during the course of the battle whatever the "type" of encounter selected. Certainly if the battle went very badly for the AI any reserve brigade might well not be sent to its likely doom --but this seems an acceptable and desirable military decision.
There were certainly some AI issues--my major concern being the tendency for the defending AI to often attack out of its good defensive position when spotting the approaching enemy------but I am sure Norb et al will have nailed all these AI "Achilles heels"for Gettysburg !!!!!

[norb]

I am still working on that. The way I originally approached this problem was for the AI to attack based on the officer's personality, then return to their hold position when they were done. But I have found that this is not an acceptable solution for anyone. So I have to work harder at getting something that the rest of the team is happy with.

[hmssws]

What bothered me was that I could usually take the objective without the AI using but half of their troops.....I could be badly outnumbered in the scenario, yet never have to face the majority of the force the enemy had at hand, while concentrating my entire force and easily winning.....While yes this type of thing did happen, there was usually some reaction by the force being attacked to get their reserves into the battle.....I'm looking forward to multiplayer....even the best AI is not up to human competition....I did think the AI was as good as I have seen.....Overall, my compliments.....

[Gfran64]

I think some of the problems with the AI in open play/no VP's are 3 fold. Firstly, the AI does not seem to recognize the significance of certain terrain features on the map. Most likely high ground, road intersections, railroad cuts, sunken roads for their defensive significance. Secondly, it does not generally concentrate it's forces before beginning an attack. Thus brigades are sent in piecemeal. Lastly, the AI will almost always attack you no matter how strong your position as opposed to moving to your rear/flank. That said, it is still the best out there by a long shot.

Regards,

Greg B)

[Hancock the Superb]

There are no defensive commands for the AI, as far as I've seen.

Just for knowledge:

I was looking in the orders.csv file, and found that the Hold (not Hold_D) command disables any wheeling! This means, that if you can figure out how to enable the Hold (for brigade command only), instead of the Hold_D, and line up your brigade upon a stonewall or whatnot, the brigade will not rotate until attacked.

This leads to guesswork here, though.

Once attacked, the brigadier can send his troops anywhere, just like in the Hold_D function. This is common. Officers rarely kept their men behind a fence or whatnot, because they are easy to be flanked. In GB Day 2, one of the generals in the salient (forgot who it was, in charge of the Pennsylvania Brigade, 1st Division) sent some regiments across the road (away from the fences) to try to halt Barksdale's drive. (Though this may be to cover the cannon nearby).

After the officer repulses the attack, he returns to his spot, as ordered by you. Then, if my guesswork is still correct, he will rotate to the original direction, and if you have it so the normal line is Aform:Brig_Line_FIGHT (instead of Aform:Brig_Line), the regiments will resume their original positions along the stonewall.

However, in an army command, this means that one must deal with EVERY brigade.

[Davinci]

norb wrote:

I am still working on that. The way I originally approached this problem was for the AI to attack based on the officer's personality, then return to their hold position when they were done. But I have found that this is not an acceptable solution for anyone. So I have to work harder at getting something that the rest of the team is happy with.

I don’t know if it’s possible, but I would like the AI to have the ability to read the TGA file.

A special column could be added into the map.csv file, and that color could be used in the tga file.

Now, depending on the value of that color - the AI would have the fore-sight \ ability - to move to the best defensive position, and stay there.

This would mean that the map makers would use this color for the best defensive positions on the map, and not break the color up, which might confuse the AI thinking process.

davinci

[norb]

There is a lot that can be improved in the AI. I am always working on making it better. There AI does have code in it for defensive terrain, but it's obviously not enough. I wish that I could just focus on one part of the code, but it's just not possible. I will keep working on it and keep listening to the comments and fix or improve what I can.