This is typically one of those things that could start to lead a life of its own if it doesn’t get handled properly from the get-go. As it turns out, Canonical is interested in tracking Ubuntu OEM installs, and it has decided to do so by installing a tracking package on OEM machines running 10.04.
The package is called canonical-census
and is installed on Ubuntu machines from OEMs only (i.e., Dell and such); in other words, regular manual installations will not receive this package. The package contains a Python program which pings Canonical on a daily basis so that they can keep track of the amount of OEM machines.
Send an “I am alive” ping to Canonical. This is used for surveying how many original OEM installs are still existing on real machines. Note that this does not send any user specific data; it only transmits the operating system version (/var/lib/ubuntu_dist_channel
), the machine product name, and a counter how many pings were sent.
The program is added to the daily cron jobs. This way, Canonical can keep track of the number of OEM installs, as well as keep an eye on how many of those installs remain active over time – or if they’re wiped clean in favour of something else (Windows, another Linux distribution). This information is obviously of value to OEMs too.
Obviously, you can remove the package through a simple sudo apt-get remove canonical-census
. While the package looks benign enough, a dialog informing of this phoning home would be nice.
…even tho’ it seems innocent enough.
I don’t really like the idea of my computer performing any acts of inter-computer communication without my knowing it or my approval.
That being said, our computers are actually pretty verbose and chatty on the network to begin with… so I suppose I am just being silly (altho’ the general network chattiness is not sending specific information to a specific server like this).
At first, I thought Canonical was going to be tracking OEM installs; kind of confusing as any business would already be tracking it’s inventory. Direct downloads remain outside the stats due to the same issues every other distribution has.
This is more like Debian’s Opt-In Popularity-Contest package for tracking what packages are installed and updated. If Canonical can do this in an anonimous way then I’m all for it due to the “market share” issue also cased by distributions outside of MBA clean retail channels.
But yeah, any information given out will inevitably be used for something it was not intended for. Canonical and *nix can benefit from the resulting stats if Canonical can do this right.
basically what it will tell us and the OEMs exactly how much of a hold linux is starting to build.
I have no problem with them being sent the data as its realistically the only way to get accurate information.
I would be interested in seeing netbook/desktop/server separation as I think the netbook market hasnt gone as well as people ‘think’ it has for linux
Yeah, you have to weed out the ones that install pirated windows from the stats.
I think it would be very interesting to see the results, certainly when the package would be installed as default on all systems, unless the user chooses to disable/remove it.
I know that would be a difficult point for a lot of OSNews readers, but I can’t think of another way to get ‘reliable’ numbers. Maybe you would even get nice numbers to show hardware vendors to convince them to share good drivers with the community.
Seriously couldn’t it be handle in updating process? I mean all machines will be updated some point, why not add simple “I am Ubuntu OEM” in that process. I mean it’s much better since
a) You can tell that they are unique machines since who crazy installs same update twice (well unless you reinstall whole system)
b) It’s user prompted, not automatic.
I’m not paranoid that they will collect IP addresses(Q:how else would you make difference between machines? A: SERIAL CODE) or something like that, but there would be better ways.
How is update a better way? There are hundreds of independent (as in not controlled by Canonical) Ubuntu mirrors, probably just as many private mirrors for corporations. Some people are using apt-p2p to update their machines at home to save some bandwidth.
There is also a problem of people hiding behind the NAT, how to count them reliably?
Which of the two will the mob scream louder about? Code like this ‘hidden’ in the update system? Or as a standalone process, which is clearly visible, and easily removed?
I can’t speak for Canonical, but the latter is the approach I’d use if I wanted to track how many users I have – the former would surely receive a *much* more hostile reception.
I always figured Canonical could keep pretty good estimates by looking at ntp traffic to ntp.ubuntu.com, or the repository update traffic for that matter.
I wish they would publish what they DO know, it would be good to have some Ubuntu usage hard facts.
Neither of them are in anywhere close to being accurate.
Why?
1) Many *untu users change the NTP to use the ‘pool’ ntp servers.
2) Some rsync the repos so that they update their machines locally. apt-get also works totally in an intranet. In these days of capped broadband there is often a ‘free’ period in the middle of the night. Cron the rsync and you can update over your cornflakes in the morning.
This may be the straw that breaks my camels back. I’ve been unhappy with the last few releases so I might jump ship. Quite a few of my friends have already done it with Debian being the biggest winner.
Edited 2010-08-10 21:08 UTC
I don’t know about anyone else, but the main attraction of getting a new Ubuntu machine is the low overheads and the attractions of tinkering. New tinkerers also have a habit of messing up their Linux installs; I’ve certainly never kept a laptop or netbook OS running for more than a couple of months. This doesn’t so much measure brand loyalty or customer satisfaction; it just measures how many people are incurious. But Linux users, by and large, are always off down the rabbit hole.
“I don’t know about anyone else, but the main attraction of getting a new Ubuntu machine is the low overheads and the attractions of tinkering. New tinkerers also have a habit of messing up their Linux installs; I’ve certainly never kept a laptop or netbook OS running for more than a couple of months. This doesn’t so much measure brand loyalty or customer satisfaction; it just measures how many people are incurious. But Linux users, by and large, are always off down the rabbit hole.”
People do that with Windows servers too.
Much ado over nothing. With MS, Apple, Google, Facebook, pages with Flash elements that can do anything they want, etc, etc, you people are getting fired up over something a LINUX company is doing as a move to help show where Linux needs to be worked on, as far as getting it out there.
Remove the tin foil from your heads, please. It looks ridiculous.