Plan for Ted Southard

Bot Muggers... And how not to complicate your AI...

Okay, so I can finally loot the bots in my game! It's a great feeling, though for some reason, they get back up when I start looting them and do a run animation(they probably want to hang on to their stuff). Other than that, it's rather nice.

Working with the database for transferring items was a lot easier than I thought it would be. I had this whole complicated function to copy the item from one spot in the inventory record to the other, and then clear out the former spot, and then it hit me: Why not just change the ID of the "owner" of that inventory record? So, viola, looting just changes that one bit, and it instantly shows up in your inventory the next time you scroll around. Finesse dictates that there be the obligatory refresh function for both inventory and loot being displayed, but everything else is in place, and I just need to get a right-click on world objects working to make it fit in with gameplay completely. Here's a shot of looting a bot in all it's glory(the box around the bot is from the Click n Pick resource while I think of a better selection cursor for the game):



The rest of the GUI is coming along nicely, and I just finished the artwork for the skill GUI, to be followed by the rest of the functionality today through the end of the week. Basically, you can click on one of the named categories of skills, and then scroll through them on the right hand side. This makes it more manageable, seeing that there's 120+ skills in the game. The GUI also makes it possible to expand categories if that need arises:



Now, I've been thinking about the AI lately, and the Artificial Emotion(AE) part in particular. I found a list of 500+ complex emotions on a website a few weeks back, and was only able to narrow it down to about 175. A second round of narrowing would probably bring it down to about 100 or maybe even less, but that being said, it's still too many.

I'm keeping the low-level bots in the game with rules-based AI, for the obvious reason that I don't want to swamp the server with trying to figure out if a rat is feeling wistful at that particular moment. But in the interests of being "light" on network bandwidth, I'm going to take a chainsaw to the AIPlayer class and start cutting out things I don't need. I'll have to do that with the Player class as well, since there's quite a bit that's not going to be getting used(no projectile collision testing, since damage is done via dice rolls, etc). In it's place should be very thin classes that are generic, affording very little services that the server-side scripts won't provide. Or functions that take a "type" variable to decide what kind of AI to apply to that function(ie, rats won't worry about improvising weapons or using healthpacks, whereas high level bots query emotional states and communicate with other bots to do flanking maneuvers and small unit tactics).

KISS rules in this area: Keep It Simple, Stupid. So now that I have my bot in, and he can fight(kinda), there's two things that need to be done: Divide the layers of AI so that I know which rulesets go to whom(as well as other features), and then set up the "action queue" for combat, where the player can click on different attacks and actions, and they enter a queue to be used when the combat timers allow. A monitoring function needs to be written for that in order to see if any "combinations" are successfully performed, and award points for it accordingly.

And lastly, time is rapidly approaching where I need to make a decision on whether or not to split up the Skill Point system. On one hand, it makes sense to keep things simple. On the other, that system is bland and allows people to increase some skills by performing irrelevent acts (like increasing your mechanic skills by killing wild beasts). So, if I go with the seperate Skill Point categories, I'll need to think of as few as possible.

The idea I was playing with was to have something like the skill categories, as well as a "General" category that allowed for a limited overlapping of Skill Points, so that players are not completely boxed in when they get Skill Points. It should be easy enough to follow, as the accumulated Skill Points will be displayed as the player switches categories in the Skill GUI, and thus knows what they can spend(and they can also spend it on the fly, since there are no "levels" in the game).

It's not as simple as other systems, but one of the goals of my game is to get players to role-play naturally, so the rules and gameplay have to accomodate that end organically. If it doesn't do so from the ground up, then the game won't stand a chance of being a true RPG... Much like the current crop of persistent worlds, which are more like MMOFPS' than MMORPG's...