Fighting against the AI, IMHO

[simovitch]

Even considering the inherent flaws of the AI, more and more I feel that playing against the AI gets a bad rap.

Sure, the occasional vehicle will move off a perfectly good road for some 4-wheeling back and forth through the trees, and squads will leave the protection of a perfectly good trench and venture like zombies to cross an open field under fire, but my argument is that these odd behaviors can be reduced by the nature of the scenario via the scenario designer.

First, If a game is designed for 2 player, then I don't have any grounds for lamenting the substandard performance of the AI, period. I also feel that if I add +XX% forces and +XX experience, it's no longer the scenario that was designed. I may as well play a QB, ugh! (Somewhere I likened the QB's to "Combat Mission Beyond Thunderdome"...)

So, on to scenarios specifically designed for the AI. Yes it can be done, in perhaps a limited fashion, but it can be done for both AI attacking and defending.

I am still learning the subtleties of designing a good scenario VS the AI, such as:

AI attacking: Give the AI open avenues of approach, good quality troops, limited vehicles, and numerical superiority. design the map knowing the AI will advance toward the defender under any cover it can find. Give the AI artillery spotters with good inital LOS, and some locked TRP's at good, probable defensive locations. Keep the starting locations close to the defensive MLR. keep the map size small, or relative to the number of troops the AI has to command. Position flags in quantities, sizes and locations that will direct the course and outcome of the battle. Spread out reinforcements so they don't get jumbled and jammed.

AI defending: Keep artifical terrain like trenches and foxholes in the trees or brush; it does seem that the AI won't then be so anxious to leave it's positions. Set up a good default defensive position and tell the player to stick to it (although sometimes the computer will generate a pretty good defense, but I don't count on it often). Position flags in quantities, sizes and locations that will direct the course and outcome of the battle.

There is no substitute for playtesting, even if you do it yourself. FOW will effect AI play!

Last, I stopped applying troop and quality bonuses when playing scenarios designed against the AI, and as such I don't design scenarios with the application of bonuses in mind. If you are an experienced player, you might just win. If you're a noob, you will likely lose but hey, you might just learn some tactics. Kind of like real life. Points off for battles designed to be played against the AI that are easy Total Victories.

Also, I have to throw this in: why would you design a scenario best played VS the AI, and suggest adding +50% to the AI forces??? When I see that, and I've seen it more than once, I'm too baffled to even contemplate playing it.

Really these are just general suggestions that may be obvious to seasoned designers. I would like to hear some more tips on designing scenarios VS the AI.

[junk2drive]

I have seen some scenarios designed to be fairly balanced for two players with the advice that if you insist on playing vs AI to give a bonus just to make it challenging.

I have seen some without bonus vs AI that I can't win.

I am an amateur at design but I test and retest my scenarios, with full knowledge of the enemy, until I get around a draw. I figure an average player blind should be able to get a draw or if interested enough will retry with different tactics. Or decide it sucks and move on.

[Fredrock1957]

I agree with never adding a Play Balance Percentage to any vs. AI especially with all the customization within the parameters of 'designed' scenarios. As far as giving the AI an experience bonus of +1, 2, or 3, sometimes when creating Historical or Semi-Historic battles (that might have been one-sided in real life) an addtional punch to the AI is necessary to make the battle more enjoyable... The other way of doing that is to add Bonus Points which can counter some of the one sided affect. Also as Junk2Drive has mentioned in a game designed for H2H play an experience bonus could help game play when playing that scenario vs the AI... Interesting stuff Simovitch would like to hear what others tend to do...

That is all...

[Larry Thorne]

Hi

Some scenarios I have made are claimed to be too hard, and asked to have reduced the ai skill levels. If I am making a scenario to be historical or semi-historical I tend to make as they were back in the days, which usually means very tough. As I personally don´t like walkthrough scenarios when the fun is just destroying the enemy. If I get a draw or minor vic I usually post them for playtest. If not tough I try to make scenarios with variable units that all have certain tasks so that the AI could be beaten, these have also been claimed to be boring or too hard as I guess some players just wanna charge the troops blindly ahead instead of making battlefieldlike commands with the units available at the terrain they are in. Like using on map mortars to take out entrenched gun positions etc.

I guess these are the reasons I haven posted too many scenarios to TSD yet and I am learning my lessons as I haven´t been making scenarios for release that long time. I have tried to listen and learn from other designers and playtesters of variable skill levels and come to a simple result. Designers just cannot please everyone. That ultimate scenario is somewhere out there that works every way possible..

As I think I have gotten pretty good reviews for my fictional stuff I am getting into the world of trying to make a good historical battle. As I still tend to pick battles that somehow draw my attention I can only hope someone likes them.

I have also found out that people like to play H2H more than against AI..correct me if I am wrong..so multiople playtesting is mandatory to get into somehwat enjoyable results for playing H2H. If one creates a challenging battle against Ai it is 99% sure that it is just tyoo easy for the other player in H2H.

Just my 1.5 cents from my short career as a designer. This is one good topic for discussion and I agree mostly on the things said by Simovitch Fred and J2D. Thanks to you guys my designing has gone from bad to little better :cool:

Cheers

-LT

[Fredrock1957]

Originally posted by Larry Thorne:

Thanks to you guys my designing has gone from bad to little better :cool:

Cheers

-LT

Don't sell yourself short my friend... yours are some of the Better ones I have played... ... Even though I wish I could beat you in a PBEM on a level battlefield.....

That is all...

-FR

[junk2drive]

I've been playing some CMBO scenarios. Amazing how far we've come.

[Mad Russian]

I found a surprising thing when I released two versions of the same scenario on the same day. I released a "vs AI" version and a "H2H" version

(multiplayer) of HSG KC Major Bake on 16 December 2004.

The AI version today has 393 downloads on the Scenario Depot while the multiplayer version has 127. In other words, for all the pomp and ceremony put forth about how players don't like playing AI scenarios it appears to be the more downloaded version of the two. Now I will readily admit that this is a single scenario sampling.

The scenario in question is also part of a sceries that has been exclusively playtested and balanced for vs AI play in the past. This was the first attempt to include a H2H version of a Knight's Cross scenario.

As far as playbalance for a vs the AI scenario I routinely playtest my vs AI scenarios with a CEB of +2. That will allow for less experienced players to use either no CEB (computer experience bonus) or a +1 CEB and still get some enjoyment out of the scenario. It will allow them a chance to win or get a close game. At the same time it allows more experienced players an opportunity to apply a +2/3 CEB to get a more challenging fight.

I never use a force % increase to balance my scenarios. I have used Bonus Points for playbalance but not as a general rule.

I try to follow the KISS principle.

Keep It Simple Stupid.

Otherwise I get in trouble. IMHO, too many variables can spoil the scenario and make it next to impossible to balance.

I agree that playtesting is the key. I had always heard that the AI wouldn't attack but I believe the KC series has proven that statement to be false. If you keep your map/forces/objectives balanced I think you can get the AI to do a pretty good job of attacking.

I think that the smaller the map/forces/objectives the easier it is to get the AI to do what you want with it. I am eagerly awaiting the CMx2 AI to see what it will be capable of.

There are now six Knight's Cross scenarios and four of those have the AI attacking. They are some of my most downloaded scenarios and have gotten very good reviews at the Scenario Depot and The Proving Grounds.

HSG has other scenarios that have the AI attacking that do an equally good job. I was only using the Knight's Cross series as an example because it is designed to be played vs the AI.