Comments on: Live at the IndieGamesCon (day 1) http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/ A group blog about computer narrative, games, poetry, and art. Wed, 13 Jan 2010 15:44:40 -0800 http://wordpress.org/?v=2.8.5 hourly 1 By: andrew http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/comment-page-1/#comment-73631 andrew Mon, 10 Oct 2005 18:50:36 +0000 http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/#comment-73631 Nate — it was an outspoken guy named Steven Snow of NCSoft; looks like he is/was <a href="http://www.mmorpg-gamer.com/modules.php?name=Reviews&rop=showcontent&id=5" rel="nofollow">a producer on an MMO called Auto Assault</a>. He was speaking off the cuff without slides; any facts or figures he said were based on his own experience / anecdotal evidence / gut feeling. Nate — it was an outspoken guy named Steven Snow of NCSoft; looks like he is/was a producer on an MMO called Auto Assault. He was speaking off the cuff without slides; any facts or figures he said were based on his own experience / anecdotal evidence / gut feeling.

]]> By: Raph http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/comment-page-1/#comment-73567 Raph Mon, 10 Oct 2005 00:04:26 +0000 http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/#comment-73567 The commonest thing is to log in a bunch of agents to do simple random walks invoking collision, whilst chatting randomly and playing emotes. Just doing this sort of thing with a few thousand agents load testing the server is going to hit the most basic "smoke test" sort of thing. Simple automated testing suites to try every command to make sure they all function, that sort of thing. The commonest thing is to log in a bunch of agents to do simple random walks invoking collision, whilst chatting randomly and playing emotes. Just doing this sort of thing with a few thousand agents load testing the server is going to hit the most basic “smoke test” sort of thing. Simple automated testing suites to try every command to make sure they all function, that sort of thing.

]]> By: Grand Text Auto » Live at the IndieGamesCon (day 2) http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/comment-page-1/#comment-73566 Grand Text Auto » Live at the IndieGamesCon (day 2) Sun, 09 Oct 2005 21:20:56 +0000 http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/#comment-73566 [...] 9, 2005 Live at the IndieGamesCon (day 2) by andrew @ 5:14 pm Adding to yesterday’s coverage, I’m continuing to blog live today at the [...] [...] 9, 2005

Live at the IndieGamesCon (day 2)
by andrew @ 5:14 pm

Adding to yesterday’s coverage, I’m continuing to blog live today at the [...]

]]> By: Dirk Scheuring http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/comment-page-1/#comment-73565 Dirk Scheuring Sun, 09 Oct 2005 21:17:06 +0000 http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/#comment-73565 <blockquote> [I]if your simulated player just randomly takes allowed actions from the list of possible player actions, is this enough to discover design and technology bugs, or is it important to have the distribution of player actions look like real human distributions? Maybe you can find a Zipf-curve-like power law that maps input frequencies to the list of allowable player actions. If so, that could be a shortcut to that artificial "70-percent-player".</blockquote>

[I]if your simulated player just randomly takes allowed actions from the list of possible player actions, is this enough to discover design and technology bugs, or is it important to have the distribution of player actions look like real human distributions?

Maybe you can find a Zipf-curve-like power law that maps input frequencies to the list of allowable player actions. If so, that could be a shortcut to that artificial “70-percent-player”.

]]> By: Nate Combs http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/comment-page-1/#comment-73564 Nate Combs Sun, 09 Oct 2005 21:11:41 +0000 http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/#comment-73564 Hi Andrew - "a tool that does 70% of what a standard user does is so valuable" “if the simulated player AI does about 50% of what real players do, that’s already extremely valuable”. I would be interested in further detail here - can you cite who said this or website. I just gave a talk on this topic involving a lot of the usual conjecture and supposition. Would enjoy inputs from those in these trenches. Detail about what they actually can simulate (vs. what is too hard) is of special interest. -nate Hi Andrew –

“a tool that does 70% of what a standard user does is so valuable”

“if the simulated player AI does about 50% of what real players do, that’s already extremely valuable”.

I would be interested in further detail here – can you cite who said this or website. I just gave a talk on this topic involving a lot of the usual conjecture and supposition. Would enjoy inputs from those in these trenches. Detail about what they actually can simulate (vs. what is too hard) is of special interest.

-nate

]]> By: andrew http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/comment-page-1/#comment-73563 andrew Sun, 09 Oct 2005 20:21:25 +0000 http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/#comment-73563 I think the guy said something like, "if the simulated player AI does about 50% of what real players do, that's already extremely valuable". btw, there's a show-off center here with 25 machines (20 PC's, 5 Macs), where people can install their indie games on them. I installed Facade on 8 of the machines, and it's been getting a lot of play and good buzz. A bit to my surprise, almost no one here had heard of Facade before seeing in here — which just goes to show how the word of Facade's existence still isn't out yet. That suggests its download numbers have a lot of room to grow, I think. I think the guy said something like, “if the simulated player AI does about 50% of what real players do, that’s already extremely valuable”.

btw, there’s a show-off center here with 25 machines (20 PC’s, 5 Macs), where people can install their indie games on them. I installed Facade on 8 of the machines, and it’s been getting a lot of play and good buzz. A bit to my surprise, almost no one here had heard of Facade before seeing in here — which just goes to show how the word of Facade’s existence still isn’t out yet. That suggests its download numbers have a lot of room to grow, I think.

]]> By: michael http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/comment-page-1/#comment-73562 michael Sun, 09 Oct 2005 18:58:46 +0000 http://grandtextauto.org/2005/10/08/live-at-the-indiegamescon/#comment-73562 I'm intrigued by the need to have AI to simulate MMO players during the development cycle; that this is "the most important tool you'll ever make." I wonder how important it is to closely match real player behavior in order to make such a tool useful. That is, if your simulated player just randomly takes allowed actions from the list of possible player actions, is this enough to discover design and technology bugs, or is it important to have the distribution of player actions look like real human distributions? If the latter, seems like machine learning and data mining techniques would be very useful. I’m intrigued by the need to have AI to simulate MMO players during the development cycle; that this is “the most important tool you’ll ever make.” I wonder how important it is to closely match real player behavior in order to make such a tool useful. That is, if your simulated player just randomly takes allowed actions from the list of possible player actions, is this enough to discover design and technology bugs, or is it important to have the distribution of player actions look like real human distributions? If the latter, seems like machine learning and data mining techniques would be very useful.

]]>