Thanksgiving Metrics Teaser
Here is a quick teaser for the metrics I’m analyzing from the playtest data from both PAX and NYU. The counts below are the number of times the Spy player picks a given character, out of 601 games across both playtests.
I find it interesting that the guy in the obvious plaid jacket gets picked almost twice as often as the James Bond-esque dude at the bottom. What kind of Spy would wear that jacket? A lot of them, it turns out.
More soon! Metrics I’m going to mine include types of wins, characters most likely to be shot correctly versus incorrectly, last mission attempted before getting shot, etc. Also, some charts over time to see if standing in line really did help the Snipers.
 |
74 (12%)
|
 |
72 (11%)
|
 |
68 (11%)
|
 |
62 (10%)
|
 |
55 (9%)
|
 |
52 (8%)
|
 |
47 (7%)
|
 |
45 (7%)
|
 |
43 (7%)
|
 |
42 (6%)
|
 |
41 (6%)
|
I need to find a better graph dispay system than these horrible HTML tables. Any suggestions are welcome.
Edit: I figured out a slightly better HTML table approach, and it seems like the most portable at this time, so we’ll see how it goes.
Interesting that the most “Spy Looking” avatar is the least shot.
This is the most/least chosen, not shot. I’ll get the “who got shot” data up here as well soon.
I was surprised that the James Bond guy was last, but also that the General wasn’t chosen often!
I can understand that, I would feel like a sitting duck walking with the General, since he stands out naturally.
People can’t control everything there brain does, so I think some characters might naturally be given more attention. Be it because of colors or appearances.
Interesting.
Processing.js is probably your best choice for charting if you’re doing relatively straightforward charts such as barcharts with custom axis labels (portraits, in this case).
There is a program called R http://www.r-project.org/
It’s free, I’ve not used it but one of the data viz websites I read recommended it.
So that means that to most people an ugly plaid jacket (apologies to anyone who owns one) is the most antithetical to a spy. Very interesting. :)
Chris, in the final version will all of the people continue to look so different or could it progress in difficulty so that you were eventually looking at people in identical tuxedos and differentiating them by their hair and features?
Hi,
Just wanted to point out that the one most chosen also is the one standing out the most, imo. He has a bright red/weird jacket, whereas most of the others seem to be shades of gray/bland. Probably why he is picked most.
I picked the Bond-esque character exactly because it seemed to be “too obvious,” so people wouldn’t expect it! As spy, I won when my opponent shot the wrong character as time ran down. I think I also fooled some people watching from the sniper side, because someone said (paradoxically), “huh, he picked the most obvious character.” Anyway, I like that there was no “random” option, so that even choosing your character can be a psych-out.
That reminds me of the thing you said about someone tapping his controller in sync with the actions of an AI, to bait the sniper. So this only works head-to-head, but I wonder if you could do something like have a little indicator that could act like the remote version of that; something that subtly changes (or another audio cue?) when the spy touches the controller. Banana bread.
Oh, and I played “Dueling Machines” (works on Wine!). Pretty cool, but it is freaking hard to pick out the hunted form the crowd.
Looks about right to me, no one really expects the guy that is vibrant and stands out to be the spy, plus the james bond guy is pretty obvious, my friend actually got shot within 30 seconds of playtime because he chose the james bond guy and the 12 year old he was playing (who didn’t really pay attention to the rules) thought he just “looked like a spy”
GOD DAMN IT.
And it looks like he’s still mad about it.
I like Google Charts. It’s pretty simple to use, and you can just point people at a URL on the Google servers, where the URL contains ALL the data. Main gnarl is the normalization step.
http://code.google.com/apis/chart/
At the NYU test I was considering choosing the African American fellow with the hope that some kind of subconscious fear of racism might prevent my opponent from singling him out. I’m looking forward to the shot correctly/incorrectly metrics …
I’m sure that you don’t have the data, but in live play I wonder how much each spy player’s actual appearance correlates to whom they choose and to whom their sniper opponent suspects.
Political correctness be damned. He’s at a spy party, how do I know that he’s American?
[From The Venture Brothers]
Jefferson Twilight: Yes, I only hunt blaculas.
Guild Candidate: Oh, so you only hunt African-American vampires?
Twilight: No, sometimes I hunt British vampires. They don’t have “African Americans” in England!
Candidate: Oh yeah, huh, good point.
Twilight: So I hunt blaculas.
Candidate: I was just trying to be…
Twilight: Man, I specialize in hunting black vampires, I don’t know what the P.C. name for that is!
We’re probably all racist on a unconscious level. At least that’s scientific findings point at.
I bet it’s far less likely for white people to play a black NPC, and vice versa, for black people to play a white NPC.
Unless of course you consciously choose to do so, knowing the general bias.
Hey Chris, you ever tried out epic mafia?
http://www.epicmafia.com
It’s a browser based game where you try and fool the town into thinking you’re town when you’re actually a mafia. Hard to explain, but it’s good!
I’ve played live-action Mafia (actually, Werewolf in my case, but it’s the same game), but not this one, I’ll check it out, thanks!
Thanks for the tip, I registered. In-game name is Loptr.
Chris, SpyParty made it into PCGamers most exciting indie games for 2011!
SpyParty was just announced a finalist for the Seumas McNally Grand Prize in this year’s Independent Games Festival. Can’t wait to play it!