OS Stats Removed From Google's Zeitgeist

Kelly McNeill writes "Google's Zeitgeist service is sometimes used by news sources as a resource to generate install-base (don't call it market share!), statistics for operating systems. osViews contacted Google to bring some clarity to questionable aspects of the OS statistic, to which Google said that Zeitgeist is only a fun search inquiry resource and should not be used to generate statistical information. A couple days after that inquiry, we found that Google has since removed the OS stats from the Zeitgeist service."

Comments

Browser stats also gone

[friedegg]

I know a lot of people were waiting to see the July browser stats to see if Internet Explorer share dropped off after the vulnerability announcements last month.

Re:Browser stats also gone

[Njovich]

I use the user agent string Googlebot/2.1 (+http://www.googlebot.com/bot.html). It gives access to quite a few sites that only give full access to subscribers and Google...

Re:Browser stats also gone

[DaveJay]

Better still to change your browser string to get yourself into the site, then (once you've found everything works fine) send a note that says

"Hey, just so you know, I surfed your site with (browser) with a hack to fool your site into thinking it was IE, and your entire site worked fine. So, your site is compatible with (browser). You can safely remove your "your browser is incompatible" message for this browser."

They might do it, they might not, but in this case you've done the work for them -- if you don't validate the site, some site-maintaining wonk has to convince their boss to pay for the new browser testing -- and many bosses won't do that.

Re:Browser stats also gone

[Adam Wiggins]

I actually emailed a site (planetxusa.com) like this, mentioning that IE is not available for my platform and that the site works fine in my browser, except for the stupid warning box. Their webmaster wrote me back a detailed message - he had never heard of a platform where IE was unavailable (i.e., he only knew of Mac and Windows) and was really curious about it. I answered his questions, and he replied saying that they would take down the warning message. About a weak later it was gone.

So - sometimes it works! :)

I can see why...

[TWX]

If people are citing Google's "just for fun" figures as something to base critical decisions on, Google could be subject to liability for the accuracy of the figures. Granted, it's not likely that a lawsuit would succeed, but simply having to defend against one wouldn't be very good.

What I don't get

[r.jimenezz]

That Google doesn't want to talk to the press now, I understand.

That the OS/browser stats would not be too reliable (I assume they are computed similarly, via the User Agent String) I can also easily understand.

That they took the stats off Zeitgeist, however, that's what I don't get. Wonder if they are now a bit paranoid about all things media after their recent faux pas?

BTW, those who don't like reading the articles would wish all stories were like this ;)

Accuracy

[Anonymous Coward]

The big problem with the Zeitgeist stats, from what I've heard, is that they only recorded the same IP address once. For people who are more likely to have a broadband connection, which is probably true of Mac and Linux users, they get counted less because their IP address changes less frequently. As Google said, it's just for fun.

Re:Accuracy

[Aero Leviathan]

Here, I'll take you through this logically.

Premise 1: Mac and Linux users are more likely to have broadband connections than Windows users. (I'm not saying that it's true, but for the sake of argument, we will assume it is.)

Premise 2: Users with broadband have IP addresses that rarely change; users with dialup have IP addresses that change frequently.

Premise 3: Google counted OS usage by the number of IP addresses that used them.

Step 1: Premise 1 + 2 implies that Mac and Linux users are more likely than Windows users to have IP addresses that rarely change.

Step 2: Step 1 + Premise 3 implies that any given Mac or Linux user is likely to be counted by Google's statistics fewer times than any given Windows user.

Now you see how the results would be skewed in favour of Windows, given the three premises (the first of which supplied by the grandparent). I think I did this right... feel free to correct me if I have erred.

What about slashdot.org statistics?

[gmajor]

I, for one, would be very interested in seeing the browser and OS breakdown on Slashdot. IIRC, slashdot has not given out this info in the past?

Meanwhile, up north

[Otter]

Lesson learned from this morning's "Lunix's desktop share is triple MacOS's!" "No, dumbass, Google Zeitgeist has it at 1%! "No, you M$ astroturfer, that 51% for Windows XP is all Linux users with spoofed browser IDs!" argument:

Google has pulled OS stats from the US Zeitgeist but Canada still has them. And Lindsay Lohan has pulled ahead of Avril Lavigne.

Re:Google Cache?

[Matheus Villela]

Google cache is good, but web.archive is best as you can see here

Re:Do People From GOOGLE Read /.?

[Suppafly]

Google has GoogleGuy who responds on a lot of forums. From what I understand, it mostly one guy, but it's a generic name so that anyone that works at google can step up and fill in for him.