AI friendly senarios

[Nomad]

Has anyone done any AI friendly senarios? I remember in PacWar there were some made that were designed to make the AI perform a lot better and I was wondering if anyone has done that for WitP yet.

[rockmedic109]

Scenarios that are specifically geared for the AI, I don't know. I am not sure there are any.

My latest idea is to modify CHS. Add Resources and Oil and HI to Japan. Up Nakajima and Mitsubishi engine production. Add small repair facilities at Truk and Cam ranh Bay {AI seems to need a little help in repairing ships}.

I have a list of changes I will make and House Rules I will use {I can post them if you'd like}.

I will up the IJA/IJN pilot pools by 200-300% and increase pilot production by about 50%. A good attrition strategy will strip the Japanese of pilots by april or may of 42....A little too early. Since the AI will not use any of the standard tricks for pilot training, I think giving them more pilots can have a major effect.

Continuous replay of first turn till at least two BBs are sunk at PH and at least 3 have greater than 50% damage {Having a very fast computer makes this not so daunting a task}.

If you or anyone else has any ideas on how to make an AI game better, post it! Always looking for new ideas.

[Ron Saueracker]

Smoke, drink and eat to excess, have a stroke, and give it a go.[:D] Other than that I think the entire excercise would be a waste of time and energy. How do you keep the AI from impaling itself? If you have the time to bother attempting to prop up the AI via the editor, why not just play against a PBEM opponent.

[Mifune]

Now that CHS and RHS are reverting back to the standard slots. The A.I. is as good as it going to get with user-made mods.

[pad152]

I created a scenario non-CHS for the AI, giving Japan more ships, planes, airgroups, armor for all planes, etc. the imporved planes made it hard to win air battles until late 1942, bottom line it still doesn't stop the AI from sending ships to stupid places to be slaughtered peice meal, the game still ended in 1943.

[Nomad]

ORIGINAL: Ron Saueracker

Smoke, drink and eat to excess, have a stroke, and give it a go.[:D] Other than that I think the entire excercise would be a waste of time and energy. How do you keep the AI from impaling itself? If you have the time to bother attempting to prop up the AI via the editor, why not just play against a PBEM opponent.

Damn Ron, sorry I asked. [:(][:@] Unlike some of my PBEM opponents, at least the AI will send me turns. I was just wondering if anyone was doing anything in this area and if thye had any success. So sorry to step outside the PBEM bounry. I guess I should know better than ask any question that might go against the great PBEM philosophy.

[Drongo]

I don't think there's much you can do with the AI in situations where it begins on the offensive and then has to cope with defending once its initial forces are beaten.

I thought it was worth trying to skip the first bit.

I'm in the process of final testing with a modified stock scenario 12 ('43-'45) for Allied vs Jap AI play. The original objective was to create an Allied player practice scenario where the AI wouldn't just hand the game to the Allies after 6 months, thereby giving the Allied player time to try out all his later war toys.

As I was test playing the original mod, I found that the Japanese AI on the defensive was good enough to justify the time to try modifying a scenario that played to the AI's strengths and minimised its weaknesses so that there would be less need for a human player to "handicap" himself in order to get a contest.

The scenario is now probably close to its final form. It has a similar starting postion to scenario 12 but the Japanese now begin with New Guinea and the Solomons firmly under their control. Major changes have also been made to the Japanese economy, force deployments, units and bases to better suit AI play. Wherever possible though, starting quality and quantity of Japanese forces was kept to what would be reasonably attainable had a human opponent (successfully) run the Japanese up to that point in the game.

The objective for the Allied player is to achieve auto victory at some point in the game. Any strategy can be chosen but to achieve auto victory before the end of the game will most likely entail him having to fight through a series of strong AI defense zones.

I'm on my third historical difficulty play test (the other two only lasted 6 months before revisions) and it's now the end of '44 in that game. The AI player has performed very well and is (to this point) yet to make any major "throw away" naval or land moves.

I've got one more set of revisions to make and then a final test run. If it all hangs together, I'll make it available in a few weeks (somewhere) for anyone interested.

Cheers

[Ron Saueracker]

ORIGINAL: Nomad

ORIGINAL: Ron Saueracker

Smoke, drink and eat to excess, have a stroke, and give it a go.[:D] Other than that I think the entire excercise would be a waste of time and energy. How do you keep the AI from impaling itself? If you have the time to bother attempting to prop up the AI via the editor, why not just play against a PBEM opponent.

Damn Ron, sorry I asked. [:(][:@] Unlike some of my PBEM opponents, at least the AI will send me turns. I was just wondering if anyone was doing anything in this area and if thye had any success. So sorry to step outside the PBEM bounry. I guess I should know better than ask any question that might go against the great PBEM philosophy.

Ruff ruff! [:D]

[el cid again]

Has anyone done any AI friendly senarios? I remember in PacWar there were some made that were designed to make the AI perform a lot better and I was wondering if anyone has done that for WitP yet.

The whole point in converting RHS over to using stock slots was to get AI to behave as well as it will. This cost a delay - and the conversion process was very painful (literally) - too much hand typing. Eventually we will address this in mini-scenarios with special AI.

[akdreemer]

ORIGINAL: Nomad

ORIGINAL: Ron Saueracker

Smoke, drink and eat to excess, have a stroke, and give it a go.[:D] Other than that I think the entire excercise would be a waste of time and energy. How do you keep the AI from impaling itself? If you have the time to bother attempting to prop up the AI via the editor, why not just play against a PBEM opponent.

Damn Ron, sorry I asked. [:(][:@] Unlike some of my PBEM opponents, at least the AI will send me turns. I was just wondering if anyone was doing anything in this area and if thye had any success. So sorry to step outside the PBEM bounry. I guess I should know better than ask any question that might go against the great PBEM philosophy.

Boy, been there, had that done to me... PBEM only owkrs with dedicated opponents and is very slow. Playing against the AI i can usually process quite a few turns a night.

[Nomad]

I love my PBEM games, I have 4 of them going. But it would be nice to have an enjoyable game that only depended on me having time to do turns and on some days getting a lot of turns done. I am hopeful that drongo is on the right track, let the AI handle a mostly defensive battle for Japan, gives it interior lines and maybe it wont screw things up too bad. [:)]

[Ron Saueracker]

How can one even attempt this though given the fact that the ability to script scenarios is not part of the editor (although it looks like it was considered)? Exactly what are people doing that is possibly going to alter the script?

[Drongo]

ORIGINAL: Nomad

I love my PBEM games, I have 4 of them going. But it would be nice to have an enjoyable game that only depended on me having time to do turns and on some days getting a lot of turns done. I am hopeful that drongo is on the right track, let the AI handle a mostly defensive battle for Japan, gives it interior lines and maybe it wont screw things up too bad. [:)]

One can but hope.

An additional benefit I've found to having the AI on the defensive is that it is far less prone to expose its all important CV forces to the Allies as easy kills.

Its obviously no "smarter" than in any other scenario but I've been very pleased so far in the latest test game to see the AI's CVs operating well back from the front lines and appearing only infrequently (and unpredictably) on the edge of the perimeter.

In the first 12 months of my latest test play, the AI's CVs initiated battle on seven occasions. Five of them involved attacks against contested forward bases while another two involved CV vs CV battles where my CVs covering an invasion were attacked.

In all seven attacks, there had been no warning of their approach as it had been made from well within the AIs defensive zone (where I had little or no search capability other than submarines). The CVs would approach and launch full strikes at maximum (torpedo) range for one day and then be gone the next, disappearing again behind the defensive barrier. Attempting to follow them with Allied CVs in these circumstances always meant having to deal with a defence zone's LBA as well. Since the bulk of the battles were fought in forward zones, the AI's CVs were also able to operate in conjunction with friendly land based fighters and bombers during the battle itself.

These "hit and run" tactics always cost the AI's CVs the bulk of their aircraft (it is to be expected in '43/'44 battles) but on each occasion that I next encountered them, they would be once again fully equipped with aircraft (but naturally their replacement pilots weren't much chop).

Suffice to say they inevitably got in some good hits. More importantly, playing a game where the AI's CVs remain an elusive but ever present threat adds a lot to scenario enjoyment.

I just hope it's a consistant feature of these types of "defensive" modified scenarios. The encouraging aspect is that all three of my test plays have shown the same AI CV behaviour.

Time will tell.

[Drongo]

ORIGINAL: Ron Saueracker

How can one even attempt this though given the fact that the ability to script scenarios is not part of the editor (although it looks like it was considered)? Exactly what are people doing that is possibly going to alter the script?

Player modifiable "scripting" is available for the smaller campaign scenarios. The large campaign scenarios appear to use a combination of scripting and hard coded moves (based on OOB slots) but as you mentioned, the scripting in these is not directly modifiable by the player.

This may change in the future but it doesn't mean you can't alter the way these larger scenarios play out now.

I'm not sure about what others have experienced but broadly speaking, I found there are three categories of units, ones that are hard coded to do specific actions, ones that aren't hard coded but the AI HQs will make use of them to take/hold their objectives and finally, units that the AI won't use or move. The last is important as you know that if you put them at a location (ie to stay and defend a coastal hex), they will remain at that location. Its a good way of covering the holes that the AI player often creates with both its objective driven and hard coded moves, etc.

The way I used to identify unit categories was to modify the starting unit deployments of the chosen scenario to resemble the situation I wanted and then run the game for a short period. Then I'd go in and see what a mess the AI had made of my nicely ordered situation. The units that hadn't stayed where I started them were either hard coded to be somewhere else or the AI's decision process had decided they should be.

If those units moves undermined what I had intended to achieve with my original deployment then they would be replaced with other units that (hopefully) would not. And then the game would be started again to see if everything held together. If it did, I'd play on and repeat the checking at regular intervals.

You can also make use of OOB slot swapping. If say, you want a particular division to be part of force used in a hard coded move or one preferred by the AI for a particular action (ie to retake a lost objective) but the AI player won't use that division, you can copy the division to the slot belonging to a unit that is hard coded or "AI preferred" and swap back the other unit to the divisions original slot.

I ended up using several different ways of trying to get the AI to make proper use of my deployments but I also tried to make full use of the existing hard coded and AI objective driven actions wherever it wouldn't compromise what I was trying to achieve. Its a lot easier to work with it than against it.

As I mentioned previously though, my mod was based on an existing defensive scenario so that I didn't have to worry about "scripted" offensives being part of the AI game play.

Cheers

[Ron Saueracker]

Whoa, that sounds very labour intensive Drongo. Glad to see you are apparently having some success.

[Grotius]

I look forward to trying it, Drongo. I love PBEM too, but my schedule is unpredictable, which makes me a suboptimal PBEM opponent. And yeah, I can do a half-dozen turns against the AI in one evening, maybe more. If I do the "historical" thing -- attack thru the Solomons and CentPac as the Allies, for example -- the AI gives me a decent game, at least for a while. OK, once in a while it exposes its ships foolishly, and I pretend not to notice.

[el cid again]

How can one even attempt this though given the fact that the ability to script scenarios is not part of the editor (although it looks like it was considered)? Exactly what are people doing that is possibly going to alter the script?

Actually, there are ways to alter the script. It is difficult unless you understand what the code is trying to do - but when you understand you may want to.

What happened in fact is that modders - I am the worst offender of all since I hate disorganized files and try to make everything be "in order" - but Andrew writes CHS found accidentally that they had incorporated a number of changes with undesireable effects and recommended using stock slots as much as possible -

what happened is that modders - constrained by too few slots and with other motives - have broken the prohibition in the manual not to move anything - ever. Doing this in ignorance of the effects "breaks" the AI - although one might (and I have) do it on purpose - only if you KNOW (mainly by testing) what the effect will be and you WANT that effect.

Thus - if AI runs the Allies - almost every unit in the Philippines (or Luzon if not playing RHS) runs to Bataan - period - every time. You don't like that ? [I don't] - you CAN change the Bataan hex to unused - and wa la - the AI actually "thinks" about wether a unit should move - or not. I did NOT make this change for RHSCVO - because it is supposed to be almost scenario 15 - or CHS155 - like in every possible way. But I WILL make this change for other scenarios. In my view the Philippine Army should at least fight a little before running - under conditions AI normally would run.
And running to a malarial swamp with too few supplies is not my first choice either. Since Fort Drum cannot run and does not run - you don't need to pretend that Bataan needs to hold to prevent use of Manila Bay. In fact, the same troops in Manila will hold it better even than Fort Drum will without help.

[el cid again]

Player modifiable "scripting" is available for the smaller campaign scenarios. The large campaign scenarios appear to use a combination of scripting and hard coded moves (based on OOB slots) but as you mentioned, the scripting in these is not directly modifiable by the player.

This may change in the future but it doesn't mean you can't alter the way these larger scenarios play out now.

The theory that the big scenarios cannot be scripted is one I am uncertain of. There IS a switch - IF we throw it - MAYBE it works!

[Drongo]

ORIGINAL: el cid again

Player modifiable "scripting" is available for the smaller campaign scenarios. The large campaign scenarios appear to use a combination of scripting and hard coded moves (based on OOB slots) but as you mentioned, the scripting in these is not directly modifiable by the player.

This may change in the future but it doesn't mean you can't alter the way these larger scenarios play out now.

The theory that the big scenarios cannot be scripted is one I am uncertain of. There IS a switch - IF we throw it - MAYBE it works!

Maybe.

Just to be clear, I was referring to the fact that the full map scenarios do not have any entries in their player accessible AI modules. This means AI game play is utilising only the game's "default" scripting (which is not player accessible). It does not mean that that particular scripting could not be over-ridden through the use of player programmable modules as is done in the smaller map scenarios.

However, it had been indicated to us during beta that the "Use scenario AI" switch was to be left "off" for the full map campaign scenarios and that was in response to a direct question about modifying those scenarios. For what its worth, the editor manual also mentions switching it "on" only in reference to small map scenarios.

Maybe that advice came from knowing all the complexities a modder would have to face with full map scenarios in attempting to mesh his own programmable modules with the existing (but inaccessible) pre-set scripting.

Or maybe it came from knowing that pre-set scripting and programmable modules are just plain incompatible in full map scenarios.

Or maybe the advice is wrong.

I certainly don't know as, since I've been able to either work around or with the existing scripting so far, I've had no need to go against the above mentioned advice.

Cheers.