Idea: Intelligent Autocalc

**SigniferOne** · 04-14-2005, 22:53

Umm, can we modify anything at all related to auto calc? Or is this all just wishful thinking? What file do we have to modify?

**Red Harvest** · 04-14-2005, 23:34

Originally Posted by dsyrow1

Umm, can we modify anything at all related to auto calc? Or is this all just wishful thinking? What file do we have to modify?

AFAIK, it is hardcoded. And it is not something we have any definitions for from CA to my knowledge. I'm just throwing it out as a concept. Sometimes things like this take root, sometimes not. Not many of us seem satisfied with autocalc, and we all seem to dislike what it does in the campaign game. This is what makes rebel armies so annoying...it is forcing us to waste time in an unfun lopsided battle. Same thing happens when facing other small AI stacks.

**Julius_Nepos** · 04-15-2005, 00:02

I've never understood Audo-calc dynamics in RTW. Even when outnumbering the enemy 800 to 100 I may only get a 2-1 strength advantage. I've had entire armies destroyed and routed on auto-calc because the defenders were one unit of uber-Cretian Archers, and another unit of uber peasents. I think the whole thing would benefit from a reality check.

I don't know how something like that would be implimented but it does seem there needs to be more consideration given to the various factors that effect who wins a battle. The sea-battles, which must use auto-calc suffer from similar problems. I only really use the auto-calc feature when my strength ratio is better than 10:1, otherwise I'm asking to not just lose alot of men but to lose the battle, no matter what my numeric superiority is. If anything it's the concept of "relative strength of armies" that ruins the ability for auto-calc to work properly. You could have 10,000 men, but if the enemy's 300 men are "rated" as having a 2:1 strength advantage you'll lose if you don't play the battle yourself.

So I agree the entire system needs to be revamped, taking more variables into account, and not necessarilly basing the whole outcome of a battle on a perception of "quality" of the armies.

**SwordsMaster** · 04-15-2005, 00:12

So you basically want to record "log" files. The thing is, to calculate results you have to store them all in one single file and then you have to produce some actual code to deal with it, although that would be probably doen in a script kinda format.

The thing is, the players victories depend greatly on the players army composition, so unless you have a record of all units available on the battlefield your results wont be that accurate anyway. I mean, if you are fighting Numidia with an all hoplite army, chances are you are going to lose even of you have the advanage of numbers. So a pretty detailed composition of the army would be needed, and including some 200 units would be quite a bit of work...

But if you can dedicate it enough time I think ot would be a nice feature.
Good Luck

**Red Harvest** · 04-15-2005, 01:01

Not quite recording a log file. More a scoring of the log file and only saving characteristics that emerge from the scoring. More detail than won-loss, but far less than a battle log. The only thing that would be saved are whatever factors would be fed into the autocalc engine. Ideally, I see this as adding only a few lines to a save file. Or optionally, it could be done as a seperate small text file, with different factors and abbreviated logs for each difficulty level, that way it could span across campaigns. Otherwise, you have to repopulate the factors whenever a new campaign begins.

It is something a developer would have to do since it is hardcoced (or some folks working with the developer could do a fair bit of grunt level testing to tune it.) Most importantly, it is going to require substantial playtesting in conjunction with coding. With a little tuning it should be possible to "model" typical human battle performance...even among different humans and on different levels.

**Oaty** · 04-16-2005, 03:13

While on the subject of autoresolve I think the A.I. gets punihed too. What I've noticed is A.I. against A.I. is both side get slowly dwindled down. At least it seems that way but I play with fog of war on so my observations are limited.

The attacker loses close to the same casualties as the defender

I hate those rebel battles too along with small A.I. stacks. The easy way to make this more enjoyable is the more you outnumber them the more likely they will auto disband without a fight. As far as the small A.I. stacks maybe give them a good chance to retreat to the nearest city. As that is what most small garrisons would do retreat to a stronger army or fortification and hope they don't get caught by the enemy army.

**Red Harvest** · 04-16-2005, 03:25

I should also add that since the player is likely to run a high net positive in his/her autocalc scoring, these factors would NOT apply to naval combat or any thing else that can only be autoresolved.

The idea is merely to make autocalc a more natural extension of the campaign, where players can fight battles because they want to, rather than being compelled to for every battle no matter how trivial.

**SwordsMaster** · 04-16-2005, 23:53

Originally Posted by Red Harvest

Not quite recording a log file. More a scoring of the log file and only saving characteristics that emerge from the scoring. More detail than won-loss, but far less than a battle log. The only thing that would be saved are whatever factors would be fed into the autocalc engine. Ideally, I see this as adding only a few lines to a save file. Or optionally, it could be done as a seperate small text file, with different factors and abbreviated logs for each difficulty level, that way it could span across campaigns. Otherwise, you have to repopulate the factors whenever a new campaign begins.

It is something a developer would have to do since it is hardcoced (or some folks working with the developer could do a fair bit of grunt level testing to tune it.) Most importantly, it is going to require substantial playtesting in conjunction with coding. With a little tuning it should be possible to "model" typical human battle performance...even among different humans and on different levels.

Yep, thats what I thought, you probably cant make the game access some files other than the ones its using now unless you write actual code. And you technically arent allowed to do that. Thats why I suggested a script... But if the variables need to be stored then a script is probably not going to do the job.