Breif Idea For Gameplay Of Future Rts Games

[weaver]

I have this idea in my head for a new way oif thinking about a game and the scenarios/missions within the game. Many of tehRTS games I have played basically let you play each side of a conflict and label them campaigns. Usually the only change is that the characters that you control are reversed, the player has a new mission and a new set of maps to play on as the new race or breed depending.

I will take the example of the command and conquer games with the russians and the americans. What if you could play a campaign thru with the americans, defeating increasngly difficult missions and scenarios, etc to a goal like spreading democracy across the globe. Basically some cheesy reasson for fighting the enemy. The new twist comes in when the player wants to then play the russian side against the americans, ie to spread communism across the globe. Usually the player is facing a preprogrammed AI for each of the missions. what of the new campaign adapted itself to the playing style of the user from each of the levels.

For instance, if while playing the americans, the player built up a quick army and rushed the opponent, when the player played the russian side for that same level, he would have to defend against an enemy that rushes him early in the game. This would be more of a philisophical change in the enemy for the level. if the enemy were to exactly mimic the original movements for that level, that player could predict the play. But if perhaps there were several AI scripts describing different player types that the program would pick based on the style of the player 9ie aggressive, defensive, seige based, sneak attack, hit and run, etc). the map would also probably be different to change to the mission so the play would not be predictable.

Basically the player would advance in skill as he/she played the computer ai through the first campaign. and then would have to leap frog over his own skill level to defeat 'himself' when he/she played the other side of the compiagn. this idea could be adapted for an entire game to as small as a twist at the final mission that you could conquer a stronghold, and then turn around and defend against you own original advance.

Let me know what you think of this idea, i am interested in hearing some feedback about this.

[Wijitmaker]

That is an interesting idea. I have heard of AI programs that analyze the players actions and it then calls up from a list of pre-programed reactions to combat the stratagy in the most effective way. For example in AoK if you as a human continually attack the computer with 5 monks and 15 tuetonic knights, the computer would respond to those attacks buy building the appropriate counter unit to defend itself.

As far as AI personalities go, I think it would be best if we followed the example that AoM had where it you could assign random or define the type of player you want to play against such as - Agressive, Defensive, Big Boomer, Rusher, etc... That seemed to work well.

I'm not sure if I understand you exactly, tell me if this is the feedback you were looking for?

[Khazun]

i actually like the idea. I mean, the player would keep getting better, as his strategy would keep being coutnered

[Ulysses S Griffin]

it kinda reminds me of ut2003 where u can choose to have the difficulty stepped up a notch each time u won

[Khazun]

yes

[Licensed Devil]

You pointed out the main problem with this, that being the predictability of it if you are playing a player who does exactly what you did when you played as it, and in order to get round it you would have to write a script that recognized the general pattern of the way the player attacked, which would be a lot more difficult to do, but with advances in programming, who knows!

[Khazun]

yes. You may be able just to make the script read what the player built, when, and even how/when the attacks went (maybye-thats going ou on a limb).

We would have several pre-defined ones, but there are always new strategies, so for patches, perhaps we could add/update the AI's for the current strategies?

Just a suggestion.

[Zeusthor]

I'm in the process of figuring out a way to create this learning AI, but we need to get further in development before I can decide whether it's feasible and possible or not.