Warrior General Games

Problem - AI development is time consuming and not very "new user" friendly. The preset AIs seem to be lacking.

Solution - set up a system where veteran players are encouraged to create AIs for the user base.

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

Step #1 - veteran creates an AI.
Step #2 - veteran fills out form describing AI. Descriptions can include things like "west coast offense" or "blitz heavy defense." Suggested personnel would be included.
Step #3 - veteran uploads AI to database and publishes it for use.

Step #4 - users browse data base. Here they can read descriptions and look for schemes developed by successful veteran coordinators.
Step #5 - users pay 50 flex to download an AI. 25 flex goes to veteran AI publisher.
Step #6 - A feedback system (thumbs up/down) is established where AIs are rated. Purchase is required to enable feedback option.

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

Pros - rookie players can find good AIs and game plans until they are ready to create their own. Cost is minimal and both publishing veteran and GLB profit from the system. The veteran coordinators also get a sense of pride in seeing how many times their schemes are downloaded.

Thoughts?

Great idea, but how many people would really write AI's?

+1

An AI is nothing without it's packages and plays. The implementation of this type of thing would have to encompass those. I think that is where you would see folks less likely to share.

With that said, many online games have had their "shelf life" extended by enabling the users to create and share new content. I know AIs and plays are not exactly new content, but the point remains.

+1 to idea, but probably is much more complex than you think to implement.

Great idea.

T

Originally posted by Plankton
An AI is nothing without it's packages and plays.

You can build the AIs without packages, inputting the actual plays instead. Would definitely make it cumbersome to build, but the reward is there if the scheme gains notoriety and get a lot of downloads (25 reward points per download).

^ that, and there is a reason there is no packages in the existing pre-set AI's.

Originally posted by Plankton
An AI is nothing without it's packages and plays. The implementation of this type of thing would have to encompass those. I think that is where you would see folks less likely to share.

this, I know when catch was creating the preset AI's he had to convert all his packages back into regular %'s.

I do, however, like the idea if we can get over this hurdle. I think some type of rating system (1-5 stars, etc) would be great too so people know how useful the AI is before paying for it.

I'm not a coder, but perhaps there is a way that this can be set up as a combo - AI template and packages. The user would purchase and install both simultaneously.

Might be logistically impossible, but let's leave that to Bort. With 25 flex per download, he has motivation to make it work if the user base wants it.

Somebody would have to test this AI's also, before hand.

Sure the thumbs up/down rating system would be good once an AI has been bought and used a few dozen times, but what about the poor sap that is the 1st to purchase and there are a few glaring problems in the AI that would cause a team to go for it on 4th and 12 from their own 20yard line with 20 seconds left in the game and ball spotted on their own 15. Wasted 50 flex AND blew the game (if failed to pick up 1st down) <---for example

Not a bad suggestion, just would need a lot of work to get right

Originally posted by Kirghiz
Great idea, but how many people would really write good AI's?

to be more precise.

However, great idea regardless

+1

Originally posted by Bane
Somebody would have to test this AI's also, before hand.

I thought about this.... maybe an uploaded AI would be run 100 times on the test server and stats can be posted. W/L record, yards per carry, yards per attempt, sacks, etc.

Maybe the first 10 downloads cost 50 flex points, with option to get that flex back with a feedback report. That way, users will have at least 10 reports to read (and 10 games to watch) before committing to purchase.

It would certainly be possible to have the packages as part of the, um, package, but it would clearly be more work to set up.

But you would need to have custom D plays for this to work for DAI anyway, which would also take work to set up.

Originally posted by markm6770
You can build the AIs without packages, inputting the actual plays instead. Would definitely make it cumbersome to build, but the reward is there if the scheme gains notoriety and get a lot of downloads (25 reward points per download).



You would still need a play database to go along with the AI database, even if you exclude packages, since the system would have to have some way to associate the plays that are used in an AI so that when you chose an AI, you also downloaded the associated plays. Otherwise, you would have an AI full of bad links to plays.

Excluding packages has other ramifications as well. Within the current AI coding, once you pick a play you have no additional options to designate specific players for that play or to determine the rate of auto-adjustment for that package. I think those lost capabilities are significant. In fact, they could be a big part of the reason that Catch's preset AIs suck more than the average AI.

Once again tho, I am not against this idea at all. I just think you have to flesh out an idea pretty thoroughly if you want to have a shot at it being done. Obviously, they know better how much work any of this is and will make their own determination on if it is a good idea and/or worth the time.