Wow.

What can I say, besides:

Hail the frog!

----------------------------

Though that makes me wonder, how much of a computing infrastructure will such a system require? Will stardock go bankrupt because all single player people will suddenly go online?

And is there an option to play only with computers online? (I.e. for people who prefer single player but would like an always up to date AI that provides a good challenge.) Or short: Will there be an ability to discrimate against humans?

There won't be an unlimited number of these robots on-line.  We're thinking only around 200 to 300 of these robots.  And once they're all involved in games, they're involved.  But it automatically gives the multiplayer ecosystem a population.

As for discriminating against humans. I really hadn't thought of that.

They won't be able to create thei rown AI personalities from scratch but I would likely welcome the opportunity to let others create derivative ones or even host their own AI personalities that play on-line.

Let me describe what we've got here a bit:

So you would have a directory called say \BILLY

Inside \BILLY would be three files along with a subdirectory called \LOGS.

The three files would be ElementalRobot.exe, AIInfo.XML and AITactics.xml

AIInfo.xml contains the Impulse account information.  You would have to create an Impulse account for your robot  with UserID and password.  This info goes into the AIInfo.xml file.

The AIInfo.xml also has the parsing stuff in there. You'll be able to put in key words for the AI to look for and type in hard-coded responses. 

Example:

<PlayerQuery>Are you there?</PlayerQuery>

  <Response>Sure am, just want to play!</Response>

AITactics.XML would have a bunch of data on how they will play the game.  Note that this isn't a scripting language. Our AI's are written in C++ and are, sadly, closed source.  But the XML file would have a lot to do with how well the AI plays the game (most of my work in GalCiv II's AI at this point is just modifying variables that could have been put into an XML file).

The \LOGS directory includes game/chat info from each game so it would be full of files from each game the robot has played.

You would run ElementalRobot.exe (or you could rename it to Billy.exe) and off he goes, connecting on-line to find games to play and joining games, chatting, whatever.

Like I mentioned before, we would probably run 200 to 300 of these robots ourselves and let others add their own if they want.  

We haven't decided whether we're going to let the robots start their own games otherwise you might have tons of robot on robot battles which kind of defeats the purpose.

Very cool. This system sounds righteous and given it is pretty much open to the community base, makes it wicked if it works as intended.

I hope we will be able to create video avatars and logos for the AI bots as well. I can see the day when we could download or even purchase, insidiously well developed custom AI opponents. Hell toss in some unique units and we could have wholesale faction add-ons. A mini cottage industry if you will.

All hail the frog!

You mentioned you wanted the AIs to behave like real people playing the game, but will there also be AIs that behave like real leaders in the game, that is, an AI that tries to roleplay as a channeler?

In multiplayer games, the player generally only has one goal, win the game. All the rest comes secondary. Some players however, especially in singleplayer mode, try to roleplay a leader with a type of personality. For example, an honorable one that will not stab you in the back. Or a xenophobic one that trusts noone and will not make an alliance, since he is convinced you will betray him. Or a peacefull one that will not start a war, but will also not give in to threats, and when attacked will respond by trying to destroy your entire civilization, to set an example, and make others think twice before doing the same.

Now, having AI opponents that have such personalities will make them seem more real, and make them for some more fun to play against. But against players that play to win at all cost, and try to game the system, they will stand little chance. As such there also need to be AIs that play just like those players do, win the game at all costs, even if it means that your 'good' civilization will have to commit genocide without any good reason.

So, will we have both types, and if so, can we select which types can join the games?

Thats a nice initiative. I like the idea to have both Player and AI in matchs (I often did that when playing some games) because its fun to see the AI interfering. Simply because, the AI do not have friend preferences (unless you are going to ingrate it) so you need to win his grace towards you somehow; and also do not have preferences to whom to atack in most games, so its just fun to see it interfering. There really are good things about AIs that a human player can not substitute.

Both have their positives and negatives; a player can be more friendly and you can chat with (which is great), but like Luckmann said, what can be a negative side of it for some (it is for me, in a way), is that it can be very competitive (or a jerk). So mixing both might be the thing.

I think mixing them might integrate players more to online games and to play together, thats why I think its a nice initiative.

I'm getting that, if there is a need for the AI to answer to players, it means the player won't be able to know what is AI and what is a player trough Info, because if they knew they would probably do not say anything to an AI. If I'm not undertanding it wrong, won't that lead to people frustratingly asking between players and bots which is what, and also having to keep in mind every match which is a player and what not, so that what they say to it would be valid?

That was fun reading, but true during a match might feel different. Thats exacly it, I worry that in that way the players would be starting chats to discover, or most probably, would always be asking arround who is Human and having to remember that trough the matchs to know to whom they should chat. Maybe in the moment the match starts it could be informed.

A good side to don't indenfy AI's: I guess there is also a deal that there are shy players (that might be afraid to play poorly (or don't show enough skill) and be recognized for that) could just don't say much and then he is considered an AI for a while (he can play cloaked lets say).He probably won't hear any agressive words.

A bad side: Soon might be a list with the AI nicks on the internet. Unless the AI nick could change and vary.

EDIT: I also have an idea. Some players make another account for playing (probably because of the reason I stated above) I suggest a feature where the player could (if he wants to play invisible anytime) be able to choose a different nick before entering the lobby.

Infractional Indentify problem: If that player act with bad behavior, there could a report that would still link to his account with the reason of it and a feature that would send a screenshot of the act direct to moderators (the accusing player selects the infractor during play and write the reason) that reveals the clocked ID so that only moderators could see which ID is. Actualy the screen goes with a little archieve that shows which IDs was at that match and what nick they were using.

Am I deviating much? (I feel I am) What I know is that even in IRC, and many games some people do like to be cloaked at least sometimes and doing another account is not practical and limited. I think something like that could encourage online playing even more, I'd do it at least sometimes.

I mean, how to not feel confortable when in shooter games for exemple, you can just put on a nick and beggin to play? Or even the IRC? That could be a feature. To avoid offensive nicks, than there could be pre-seted ones to choose from. That put less pressure on players or at least a group of them, and pressure unless from the gameplay itself is not a good incentive.

So how long until this cloud of AI's become self aware, take over the internet and then try to eliminate the human race?

Rolling your own AI and watching what the logs bring in could be a game in itself.

Will you have control over which AI populate your game?  I mean, I don't want some impossible AI joining my game under some innocuous name like "Fred" and taking everyone to school.

Only fair way is the order in which people preordered. FIFO style.

Damn, why did i wait a few days to preorder?

If people cannot play the beta as soon as it's available, they are going to complain. A lot. Let's hope that Karma doesn't gives privileges for the beta...