Archived

This topic is now archived and is closed to further replies.

fup

GDC 2004 Impressions?

Recommended Posts

quote:
Original post by Anonymous Poster
quote:
Original post by Ferretman
I was quietly surprised at the lack of the middleware discussions that there have been in the past. Either developers are evaluating the products out there and keeping it quiet, or they''re just not interested at present.....



I wonder if there would be more discussion on the middleware topic if there were more producers or leads in the room rather than AI programmers (who might feel as though the advent of AI middleware could hurt their careers).




Well, it''s a good question, but honestly we''ve had quite a few of both in the room in the past. My experience (and this is not intended as a slam on anybody) is that producers don''t know enough to know if a package would be useful or not, and that developers essentially have to be convinced that any learning curve is worth the time/cost/energy.

A couple of years back the potential of middleware was hotly discussed (just look at my previous roundtable reports) and eagerly debated, but I think that the first generation of middleware fell a little flat with most developers. Since then there have been some companies come and go, and the current crop of toolkits (which Eric Dybsand did a fantastic job of reviewing a while back in Game Developer) have matured a great deal.

My general impression, when the subject came up, is that most developers who had experimented with middleware weren''t particularly interested in the current crop--it will take somebody new, and/or somebody with some academic background, to take the reins on this one.

Development of hardware and/or some common standards would help a lot too, come to think of it.




Ferretman

ferretman@gameai.com

From the High Mountains of Colorado

GameAI.Com

Share this post


Link to post
Share on other sites
I went to many of the AI roundtables this year, and was admittedly one of the "more vocal" people that some of you are referring to. Without being accused of wanting to "take over" this thread too, I would like to offer two small points in self-defense.

One, it is the roundtable moderator''s job to rein in the conversation if he/she thinks it is getting out of hand; any failure to do so can easily make the participants think that everything is going just fine.

Two, even with our very vocal minority, there were plenty of situations in every single session I was in where the conversation significantly lulled. Any of the non-vocal majority who wanted to jump in at those times could have easily done so, but for the most part they did not. Heck, they didn''t even have to jump in; just raising a hand would''ve sufficed, but that still didn''t happen most of the time. If most people want to spend most of their time at a roundtable just listening, that''s fine, but that means that a few people have to do a whole lot of the talking to make up for it, and that''s exactly what happens. Roundtables are discussions, not lectures.

Next time, if you don''t want only a small number of folks holding the conch all the time, raise your hand and hold it for yourself once in a while.

Share this post


Link to post
Share on other sites
Hey Chris... good to see you here.

One thing you said at a roundtable that rang a bell with me is when you brought up the 4 types of intelligence (strategic, tactical, logistical and diplomatic). Someone knows their personality and temprament theory, eh? Straight out of David Keirsey''s work!

I actually submitted a proposal for a lecture that covered personality types and game design. It didn''t get through... this year!

@Ferretman... have you guys floated the idea of having an all day AI roundtable? There is far too much to discuss in one hour. We are usually just getting going by the end of the hour... and then the next day there is a different mix of people. If you were to disquise it as an all-day tutorial, it would be great. Break the day up into some different sections by type of AI, select some people to do some presentations on cutting edge work they have done, then open it up to discussion for a while on that topic. That way, there is some structure to the day and it doesn''t just wander... but it isn''t strictly a series of lectures or a panel discussion either.

By making it a tutorial (as defined by GDC), there is an extra cost for attending it - but that also makes sure that it is the attendees are either real AI programmers or people who are serious about getting into it.

Damnit... we all bitch about how AI isn''t getting the attention it deserves - and yet the graphics guys get all day sessions to do stuff. Why not the AI folks?

Dave Mark - President and Lead Designer
Intrinsic Algorithm -
"Reducing the world to mathematical equations!"

Share this post


Link to post
Share on other sites
Very good idea Dave. It might be better though to make the format more of a conference workshop than a tutorial. A tutorial normally has the expectation of interactive learning, directed by one or more tutors, on pre-defined problems. That might be exactly what you were thinking, but since many of the attendees would have little or no implementation experience, it might prove a fruitless endeavour...and I don''t think you meant you actually wanted to teach techniques.

A workshop on the other hand is typically intended as a forum in which people can present information and then this information can be discussed by the group as a whole; a sort of ''think tank''.

I would suggest - for an all day workshop - inviting ~6 people to present information on advanced problems in Game AI and possible solutions they have considered for these problems. Have each speaker present about 15-20 minutes on the problem and the solutions they have considered (along with the problems they found with the solution methods) and then have about half an hour discussion on each problem. Break the day up into 4 sessions along the lines of

Start 9:00 am
Session 1: 1.5 hours (2 speakers)
Morning Tea: .25 hours
Session 2: 1.5 hours (2 speakers)
Lunch: 1.0 hours
Session 3: 1.5 hours (2 speakers)
Afternoon Tea: 0.25 hours
Session 4: 1.5 hours (further open floor discussion of any of the days topics or anything not covered)*
Finish: 4:30

*This gives people the chance to raise questions or provide ideas after having had some further time to think about the problem.

Timkin

Share this post


Link to post
Share on other sites
I was using the word "tutorial" in the way the GDC uses it: a full day session on one of the first two days of the conference. It is not meant to imply format. However, it would be more like a workshop sort of environment or simply a very large roundtable. There could be general interest periods and then smaller, genre-specific sections such as the guys already do now.

Dave Mark - President and Lead Designer
Intrinsic Algorithm -
"Reducing the world to mathematical equations!"

Share this post


Link to post
Share on other sites
Well I suggested to Steve at GDC that at least a 2 or 3 hour would be good. Full day sounds even better

Perhaps if the groups get too large (even 15 active speakers may be a bit much for a good discussion) that the groups break up into smaller work-groups.

- - -

Might be fun to have an ''It Came From GDC'' paper written after each GDC describing briefly a wacky new idea that comes out of discussion groups.



regards,

GeniX

www.cryo-genix.net

Share this post


Link to post
Share on other sites
quote:
Original post by GeniX
Well I suggested to Steve at GDC...
Damnit! I''m trying to figure out who you are! This is killing me!



Dave Mark - President and Lead Designer
Intrinsic Algorithm -
"Reducing the world to mathematical equations!"

Share this post


Link to post
Share on other sites
Yeah, I went there in an effort to find out... but I am horrible with names. I would likely recognize the face. *sigh*

Dave Mark - President and Lead Designer
Intrinsic Algorithm -
"Reducing the world to mathematical equations!"

Share this post


Link to post
Share on other sites
quote:
One thing you said at a roundtable that rang a bell with me is when you brought up the 4 types of intelligence (strategic, tactical, logistical and diplomatic). Someone knows their personality and temprament theory, eh? Straight out of David Keirsey''s work!

Cool, someone else here has read Keirsey. Yeah, I''ve done a bit of research into psychology and personality/temperament stuff; it''s a definite benefit when designing believable game AI.

quote:
I actually submitted a proposal for a lecture that covered personality types and game design. It didn''t get through... this year!

Keep trying; if it doesn''t make it on a game design track then a programming track might be another avenue to pursue. It''s definitely something I''d attend, that''s for sure.

Share this post


Link to post
Share on other sites
quote:
Original post by Chris Hargrove
Cool, someone else here has read Keirsey. Yeah, I''ve done a bit of research into psychology and personality/temperament stuff; it''s a definite benefit when designing believable game AI.
A bit of reading? Laurie (my wife) gave me ENTP license plates for my birthday - I responded a month later with INFJ ones for her. Yeah, we know it inside out.



Dave Mark - President and Lead Designer
Intrinsic Algorithm -
"Reducing the world to mathematical equations!"

Share this post


Link to post
Share on other sites
quote:
Original post by Chris Hargrove
I went to many of the AI roundtables this year, and was admittedly one of the "more vocal" people that some of you are referring to. Without being accused of wanting to "take over" this thread too, I would like to offer two small points in self-defense.

One, it is the roundtable moderator''s job to rein in the conversation if he/she thinks it is getting out of hand; any failure to do so can easily make the participants think that everything is going just fine.

Two, even with our very vocal minority, there were plenty of situations in every single session I was in where the conversation significantly lulled. Any of the non-vocal majority who wanted to jump in at those times could have easily done so, but for the most part they did not. Heck, they didn''t even have to jump in; just raising a hand would''ve sufficed, but that still didn''t happen most of the time. If most people want to spend most of their time at a roundtable just listening, that''s fine, but that means that a few people have to do a whole lot of the talking to make up for it, and that''s exactly what happens. Roundtables are discussions, not lectures.

Next time, if you don''t want only a small number of folks holding the conch all the time, raise your hand and hold it for yourself once in a while.




Chris makes a good point. I think that many of the folks who are new to the roundtables and/or GDC don''t feel like they can hop in enough. When I''ve talked to them offline they usually are so new that they don''t want to "seem dumb" or "ask stupid questions", but that''s part of the point of a roundtable of course--to talk about things. There really aren''t any dumb questions, IMO.

I didn''t really think you got too out of line or anything Chris--your points were all good and thoughtful. As Dave says you brought up some great topics. Heck, I had one session a couple of years back where virtually all the talking was done by Will Wright...and frankly nobody cared, because everybody wanted to hear about The Sims .





Ferretman

ferretman@gameai.com

From the High Mountains of Colorado

GameAI.Com

Share this post


Link to post
Share on other sites
quote:
Original post by GeniX
Well I suggested to Steve at GDC that at least a 2 or 3 hour would be good. Full day sounds even better

Perhaps if the groups get too large (even 15 active speakers may be a bit much for a good discussion) that the groups break up into smaller work-groups.




We did indeed talk about it, and several folks besides yourself suggested it, as it turned out.

While I think an all day session would be too much in a roundtable format for the reasons Dave mentions, doing a half-day session could work with a break or two in between. Alexander Nareyek has done very well with 2-hour sessions to work through the IGDA AI SIG issues; perhaps we can do this too. I'll talk to the guys about it.

What other suggestions do you guys have? This is a great time, frankly, to consider changes to the way we are doing the roundtables.


quote:


Might be fun to have an 'It Came From GDC' paper written after each GDC describing briefly a wacky new idea that comes out of discussion groups.

GeniX

www.cryo-genix.net



We've sort of done that with our Moderator's reports after each GDC. Neil and Eric have actually already done theirs for the 2004 sessions (they're on my site, while mine is still in progress.

A couple of years back we were talking heavily about doing a "Tales from the Roundtable" book on things that have cropped up over the years...perhaps it's time to consider this again? Would anybody out there be interested in something like that?




Ferretman

ferretman@gameai.com

From the High Mountains of Colorado

GameAI.Com


[edited by - Ferretman on April 20, 2004 12:40:34 AM]

Share this post


Link to post
Share on other sites
quote:
Original post by InnocuousFox

I actually submitted a proposal for a lecture that covered personality types and game design. It didn''t get through... this year!




I didn''t know you submitted a proposal like that, Dave--did you mention that to me? If you did I apologize for spacing it...sounds like a neat idea to me. You should resubmit it again for next year!

quote:


@Ferretman ... have you guys floated the idea of having an all day AI roundtable? There is far too much to discuss in one hour. We are usually just getting going by the end of the hour... and then the next day there is a different mix of people. If you were to disquise it as an all-day tutorial, it would be great. Break the day up into some different sections by type of AI, select some people to do some presentations on cutting edge work they have done, then open it up to discussion for a while on that topic. That way, there is some structure to the day and it doesn''t just wander... but it isn''t strictly a series of lectures or a panel discussion either.




My biggest problem with an all-day thing like you suggest is that it gets awfully big. The IGDA sessions where people break up into a dozen sub-groups comes to mind, and I''m not sure all that much ever really gets accomplished there. Still, the focus would be different for our stuff so that''s not without merit...we''d be talking about all kinds of stuff not trying to formulate policy.

Question: Are there enough AI people in the industry to reasonably drive one of these all-day things? A lot of folks don''t show up for the first couple of days so I''m worried about attendance....

quote:

By making it a tutorial (as defined by GDC), there is an extra cost for attending it - but that also makes sure that it is the attendees are either real AI programmers or people who are serious about getting into it.



True...good points.

quote:

Damnit... we all bitch about how AI isn''t getting the attention it deserves - and yet the graphics guys get all day sessions to do stuff. Why not the AI folks?




I think there''s a lot of value in what you''re suggesting, though we''d definitely need to line up a couple more folks to help with the break-into-groups moderation. Hmmmmmm.....




Ferretman

ferretman@gameai.com

From the High Mountains of Colorado

GameAI.Com

Share this post


Link to post
Share on other sites
quote:
Original post by InnocuousFox
quote:
Original post by GeniX
Well I suggested to Steve at GDC...
Damnit! I''m trying to figure out who you are! This is killing me!



Heheheheheh...I know who he is....





Ferretman

ferretman@gameai.com

From the High Mountains of Colorado

GameAI.Com

Share this post


Link to post
Share on other sites
quote:
Original post by Ferretman
While I think an all day session would be too much in a roundtable format for the reasons Dave mentions, doing a half-day session could work with a break or two in between. Alexander Nareyek has done very well with 2-hour sessions to work through the IGDA AI SIG issues; perhaps we can do this too. I''ll talk to the guys about it.

The problem with that is it may eclipse other sessions during that time period. Some serious planning would have to be done to make sure that it didn''t. If you are not careful, you are going to lose a lot of high-level AI programmers to other programming roundtables and lectures. At that point, you efforts at maximizing the quantity end up shaving from the quality.
quote:
What other suggestions do you guys have? This is a great time, frankly, to consider changes to the way we are doing the roundtables.

You and I talked about putting the beginner session on the first day rather than the third. That way, people can take what they learn there and decided whether or not to attend the following sessions.

Also, it may be better to have a more scripted approach to that session - something that covers all the basics concepts (but not implementations) of FSMs, pathfinding, etc. Make it a definition of terms sort of thing and then let it spin from there. There is too much ''wandering'' going on in the beginners session.

Also, while I like the idea of breaking up into smaller, genre-based groups... that first day (under your current scheme) is kind of upsetting in that you don''t know who you are going to end up with. One discussion in room A may have been really boosted by someone in room B, etc. If at all possible, the roundtables should be blended.

We have already noted the fact that a lot of people don''t even WANT to share - or may not posess enough knowledge to even ask a question. Let''s assume that about 20% of the people in the rooms are the ones with lots of knowledge (the "answerers") and another 40% are those that are seeking more (the "questioners") and the remaining 40% are those that are just soaking (the "listeners"). With the crowds that we have had in each room (~30), that means that about 6 people are "answerers", 12 are "questioners", and 12 are "listeners". If we combine the 3 rooms, we now have 18, 36 and 36.

If you put that into a half-day session, you may loose some of the "listeners" to other sessions. That''s OK, they may be blown away by the higher level of discussion. (Let''s assume then that the "listeners" go to the beginners session during a different time.) That leaves us with about 20 "answerers" and ~40 "questioners". During a 3-4 hour session, there is a LOT of sharing that can go on!

The way I see it, the questioners would tend to start a thread and then the "answerers" would run with it for a while - occasionally being redirected by follow-up questions. The "answerers" would get a lot of knowledge sharing out of the process as they exchange their ideas and approaches and the "questioners" would almost see it like attending an open panel session - making it very educational for them as well.

quote:
A couple of years back we were talking heavily about doing a "Tales from the Roundtable" book on things that have cropped up over the years...perhaps it''s time to consider this again? Would anybody out there be interested in something like that?

This is something that will be far more feasible with longer sessions. You won''t get enough meat with the single hour stuff. Also, this will have to be based on the premise that you are recording the roundtables - which we have also discussed. A lot of the knowledge sharing that goes on in those rooms tends to evaporate once you leave. It is hard to take with you.


Dave Mark - President and Lead Designer
Intrinsic Algorithm -
"Reducing the world to mathematical equations!"

Share this post


Link to post
Share on other sites
quote:
Original post by Ferretman
Heheheheheh...I know who he is....

Yeah, he contacted me via email and pointed me to a picture... I feel a lot better now!



Dave Mark - President and Lead Designer
Intrinsic Algorithm -
"Reducing the world to mathematical equations!"

Share this post


Link to post
Share on other sites
quote:
Original post by Ferretman

quote:


Might be fun to have an ''It Came From GDC'' paper written after each GDC describing briefly a wacky new idea that comes out of discussion groups.

GeniX

www.cryo-genix.net



We''ve sort of done that with our Moderator''s reports after each GDC. Neil and Eric have actually already done theirs for the 2004 sessions (they''re on my site, while mine is still in progress.



I''ve had a read through Neil and Erics reports and those are not exactly what I mean.

My idea, however impractical it may be, is that some interested parties either get together to have a good talk around some wacky algorithm or implementation idea (and stay reasonably focused on it for 30 mins or so) and then do a Game Programming Gems length write-up on it post-GDC. Not that the AI field is in need of creativity (in fact it seems to be full of it!) but to present GDC as an opportunity to have a think-tank.

Some off-the-top of my head wacky examples (fairly poor, but you get the idea):
- decide on 3 key personality traits, and use matrix/quaternion math for interpolation. Use projectin and skew transformations and see if anything comes of it.
- use the players well-travelled paths in an rpg for NPC''s to use (maybe through scenting system, allow the player to unknowingly define the paths - what implications would this have?)
- dynamic navigation mesh working off probabilities where the bot (in an fps) can try things and update the mesh accordingly.





regards,

GeniX

www.cryo-genix.net

Share this post


Link to post
Share on other sites
quote:
Original post by Ferretman
I think there''s a lot of value in what you''re suggesting, though we''d definitely need to line up a couple more folks to help with the break-into-groups moderation. Hmmmmmm.....
Are you recruiting?



Dave Mark - President and Lead Designer
Intrinsic Algorithm -
"Reducing the world to mathematical equations!"

Share this post


Link to post
Share on other sites
quote:
Original post by GeniX
quote:
Original post by Ferretman

quote:


Might be fun to have an ''It Came From GDC'' paper written after each GDC describing briefly a wacky new idea that comes out of discussion groups.

GeniX

www.cryo-genix.net



We''ve sort of done that with our Moderator''s reports after each GDC. Neil and Eric have actually already done theirs for the 2004 sessions (they''re on my site, while mine is still in progress.



I''ve had a read through Neil and Erics reports and those are not exactly what I mean.




I think I get it now....interesting thought.

quote:

My idea, however impractical it may be, is that some interested parties either get together to have a good talk around some wacky algorithm or implementation idea (and stay reasonably focused on it for 30 mins or so) and then do a Game Programming Gems length write-up on it post-GDC. Not that the AI field is in need of creativity (in fact it seems to be full of it!) but to present GDC as an opportunity to have a think-tank.



You''re dead-on right about the industry being filled with great ideas, but I''m wondering about the practicality of this one (which you yourself even doubt). A GPG-length article is going to take a fair amount of work for something that is essentially a free GDC writeup, if nothing else, and I wonder about how folks might feel about an idea of theirs being posted for free in the GDC papers after the fact. At least when somebody is putting together a presentation it''s generally on stuff they''ve already (hence the presentation of Doom and Quake engines) or based on roundtables where everybody expects ideas to be exchanged. What you''re suggesting seems one step further than that to me, which means it might limit our getting participation.

Neat thought though....

quote:

Some off-the-top of my head wacky examples (fairly poor, but you get the idea):
- decide on 3 key personality traits, and use matrix/quaternion math for interpolation. Use projectin and skew transformations and see if anything comes of it.
- use the players well-travelled paths in an rpg for NPC''s to use (maybe through scenting system, allow the player to unknowingly define the paths - what implications would this have?)
- dynamic navigation mesh working off probabilities where the bot (in an fps) can try things and update the mesh accordingly.



Interesting suggestions--more of an "AI workshop" than a roundtable per se from the look of them. This might be a good way to build on the basic approach of the roundtables but focusing on individual problems--something I''ve seen the IGDA do well (and which I kind of wish the IGDA AI Standards SIG would do).

Our biggest problem might be to come up with good "sample problems" to really capture the imagination and interest of attendees, but that''s not impossible. We could possible poll people ahead of time...hmmmmm.

Neat idea!






Ferretman

ferretman@gameai.com

From the High Mountains of Colorado

GameAI.Com

Share this post


Link to post
Share on other sites
quote:
Original post by InnocuousFox
quote:
Original post by Ferretman
I think there''s a lot of value in what you''re suggesting, though we''d definitely need to line up a couple more folks to help with the break-into-groups moderation. Hmmmmmm.....
Are you recruiting?



Dave Mark - President and Lead Designer
Intrinsic Algorithm -
"Reducing the world to mathematical equations!"


WELL NOW , anything is certainly possible....if we did make change to the roundtables such as those being kicked around here I think we''d need more moderators if only to help with control. A tutorial-sized session in which we break up into smaller groups would need a moderator for each group, I should think.....




Ferretman

ferretman@gameai.com

From the High Mountains of Colorado

GameAI.Com

Share this post


Link to post
Share on other sites
Okay, there has been a of interesting discussion here about the possibilities of changing the GDC AI roundtables, and I think some good ideas have been kicked around. It just might be time to consider changing the format of the roundtables again....

...so, to that end, I've put up a new poll on the GameAI.Com site about this very issue. How, if at all, would folks hereabouts like to see us change the GDC AI roundtables for next year?

I put up the most likely options but if I've missed something or you have another suggested, just add a comment...I can't cover everything. Sorry.

I'm very interested to see what crops up here....




Ferretman

ferretman@gameai.com

From the High Mountains of Colorado

GameAI.Com


[edited by - Ferretman on April 22, 2004 12:05:29 AM]

Share this post


Link to post
Share on other sites
quote:
Original post by Anonymous Poster
Sorry, but where exactly is the poll? I can''t find it.

Click on the link in the post above, then find the place where it says "poll".

Dave Mark - President and Lead Designer
Intrinsic Algorithm -
"Reducing the world to mathematical equations!"

Share this post


Link to post
Share on other sites
quote:
Original post by Anonymous Poster
Sorry, but where exactly is the poll? I can''t find it.


My apologies; I didn''t make it very obvious. It''s in the list of news items towards the middle of the page, where it says "What do you want the 2005 GDC sessions to be like?"

I need to clean up that page badly.....not enough time, not enough time...




Ferretman

ferretman@gameai.com

From the High Mountains of Colorado

GameAI.Com

Share this post


Link to post
Share on other sites