[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Public WebGL] WebGL Stats

Note that even defining "unbiased" is nontrivial. It could mean at least 3 different things:
 1. representative of the whole population of Web users
 2. representative of the subset of Web users who try to access WebGL content
 3. representative of a more specific population i.e. if you're a game developer, you're interested in the population of visitors of your game's website.

For Firefox users, I have some precise data about 2. and more indirect data about 1. on this page:

The blue curve on these graphs gives the answer to 2. for Firefox users. It says it's currently just under 50%, about to hit 50% soon.

The orange curve says how many % of people have accelerated compositing. We know that the % of people who can access WebGL content is significantly higher than that (because our blacklist rules for compositing are particularly conservative). Currently the orange curve is at 27%, so it's safe to say that the answer to 1. for Firefox users is somewhere between 30% and 35%.


I am not sure how biased this is. If I compare these numbers with our internal tracking they seem low for coming from a WebGL background. In general we just use 2d canvas and offer WebGL only as experimental option. But we track all people that hit our game.
In the last 30 days we had 500.000 unique users. There was no WebGL support at all for 22% (mostly IE). Around 15% have a WebGL compatible browser but there was a Problem creating a context. And final for 63% we were able to create a WebGL context. I know that this is biased too as PC gamers are most likely update their browser and drivers more often then other users. 
Anyway I think the details are the more interesting part as it helps to know what could be used if someone made the decision to use WebGL at all.

Von: owner-public_webgl@khronos.org [owner-public_webgl@khronos.org] im Auftrag von Florian Bösch [pyalot@gmail.com]
Gesendet: Montag, 9. April 2012 09:07
An: Mark Callow
Cc: Mo, Zhenyao; Ben Vanik; public webgl
Betreff: Re: [Public WebGL] WebGL Stats

On Mon, Apr 9, 2012 at 4:23 AM, Mark Callow <callow_mark@hicorp.co.jp> wrote:
I hate to rain on the parade but the data is being collected from WebGL applications which gives an inherent bias to the "Who has WebGL answer." I'd use that particular piece of information with caution.
There is a certain seed bias but not because the information is being collected from WebGL applications (it isn't just), but because those interested to embedd the tracker are from a WebGL background, who tend to have web properties that attract a WebGL centric crowd.

That being said, the tracker is now included on 34 different sites, it has 146k visitors and 230k impressions. You have to start somewhere if you want to get data.