January 26th, 2011 at 4:57 pm Posted by TheOtherWinter

I’m wondering if your reliance on Armory updates to indicate activity is misguided, unless you’re using the bots who can look as us hourly to validate the activity (or lack thereof). With Cata’s release, I’m not triggering as many Armory-tracked achievements while I’m leveling even my primary toons (never mind the alts), and I’m definitely not dropping raid bosses nearly every night as I did at the end of Wrath. I’m consciously aware that there are nights I play for hours yet leave no trace on the Armory… out in the field questing, maybe run a Normal or even an Heroic but get no drops, go picking weeds for flasks for the one toon undertaking the 10man raids with a guild that has dropped Magmaw exactly once (while I wasn’t there of course). Comment?

January 26th, 2011 at 5:15 pm Posted by Nick Yee

Great questions, TheOtherWinter. And I definitely could have made this more clear in the post.

For hours, we’re using the in-game bots for the estimates rather than the Armory.

And with the Armory, the interesting thing is that it updates the “last played on” field even if you just log on and not do anything. So even if you don’t trigger an achievement, the Armory notes that you logged on, and our data scraper will collect the XML. So as long as you logged on, we can still reliably use the Armory to count days played.

January 27th, 2011 at 10:30 am Posted by Chris

How can you know the person’s gender through the armory… I hope it’s not based on characters gender. Were these people aware of that “tracking” and volunteered in it or is it just random picks without any validation?

January 27th, 2011 at 11:34 am Posted by Nick Yee

Chris – This was part of a study of 1000 WoW players who completed a survey and consented to being tracked longitudinally. See this: http://blogs.parc.com/playon/2010/07/23/welcome-to-playon-2-0/

January 27th, 2011 at 7:48 pm Posted by Mike

I’m not really sure I believe that in a 5 month period somebody played for 140 days. That is basically never logging out for the entire time. I don’t think there is a way you can accurately estimate played time. Even an average of 65-70 days /played is excessive. Over 5 months (150 days) that is about 10 hours a day.

Sorry, but this doesn’t mean much.

January 27th, 2011 at 8:13 pm Posted by Nick Yee

Mike – “Days” counted in the Armory are “unique days” not “24 hour days”. I.e., 140 days = someone logging in 140 unique days, but they could have just spent 10 seconds in game each day.

“Hours” as counted by the in-game census bot has intervals of one-hour, and is a much more accurate measure of time played.

That’s why we mention both, and what’s important is that both measures are consistent with each other.

February 5th, 2011 at 2:58 pm Posted by Changing Guilds | PARC PlayOn 2.0

[...] both regions, men were more likely to switch guilds than women. Consistent with the days played analysis, we find that women are more loyal and engaged to both the game as a whole and to social [...]

March 30th, 2011 at 12:59 pm Posted by Christine

Keeping in mind you’re only getting the players non-casual enough to look up extra things for WoW, like blogs, wherein they found out about and signed up for this study.

April 9th, 2011 at 4:38 pm Posted by erik

Unfortunately the armory data could be significantly overstating the amount of days played, depending on how and when it updates. I give an example.

Armory updates at 0000h (midnight) for every day.

If I play from 2200 to 0200 it counts as two days, but it actually is just one night of playing, even though it happens on two days.

I am not sure when the armory updates though.

cheers
erik

April 13th, 2011 at 12:21 am Posted by Nick Yee

Armory updates around 4am. So there’s likely noise, but just a little.

But the in-game census data (which doesn’t have this problem) is consistent, so this also suggests that the “break problem” has low impact on data quality.