This is a classic chicken-and-egg situation. I experienced this with OS/2 in its heyday. Users don't want to use the OS because there are no apps, application developers don't want to write or port apps because there are no users. Same with hardware and drivers.
The problem with convincing vendors and developers that there is a market is that quantification of the market is characterised by very poor estimates. Simply, we cannot count sales and therefore there is no easy way to compare market share figures with Windows and Apple OSs.
There have been attempts to address this issue in the past. With respect to GNOME we have previously debated the issue, but it has now become dormant. Other projects, such as Linux Counter, have also tried, and failed.
The problem seems to devolve into two potential solutions:
- Some kind of "phone home" software, probably opt-in rather than on by default
- Encouraging people to visit a web site to "stand up and be counted"
The second option suffers from the problem of publicity. This is a classic e-commerce problem: it's no use having a great site/service if people don't know about it, and know how to find it.
Up until lately GNOME and the Linux community in general has not had the resources to mount an information campaign asking users to stand up and be counted. But now we do.
Google (and maybe others).
I would venture a guess that every internet-connected individual in the world would hit google.com at least a few times per year. If we could convince them to (based on some detection system that could identify potential GNU/Linux or other Free OSs) offer a text ad pointing people to a counter, it would be problem solved as far as I can see.
The remaining problems would be purely technical (e.g. how to prevent "vote-stuffing", bots and so on).
I have no experience at dealing with global mega-corporations. Would it be best to develop the solution first (i.e. register a domain and set up the counting software) or just float the idea to them as-is?
By the way, www.linuxusers.org seems to be available and www.linuxusers.net is parked.
If you have thoughts on this issue, please leave a comment, or perhaps edit the GNOME "Counting Users" page.
Thanks,
John.
Posts
10 comments:
Hello.
I just wanted to note, as a GNOME user, that I probably would not click on the link to identify myself, even if recommended especially by Google.
I also think I know people who would make an effort to be counted multiple times, to inflate numbers.
Uniquely identifying users also seems nearly impossible. (Perhaps distributions could provide a unique identifier including multiple pieces of data (eg. MAC address) that could be requested.)
I think having installations "call home" would probably be more accurate and reliable, but I also agree regarding the potential for abuse and distrust.
SIGH.
What about package managers? Do the vast majority of installations make use of them? Maybe not conventionally with regard to corporate installations. But perhaps at distribution installation, the package manager could "register" itself somehow with the distributions repositories.
Should re-installations count as new installations? If not, perhaps an ID for an installation needs to be derived from the hardware. But I suppose that fails to consider how many people log into that machine. If we just do it by the number of accounts, then we inflate numbers with temporary accounts for friends and such. Too bad profiling usage for the accounts and trying to distinguish between different users based on behaviour would be far too invasive for anyone to sincerely push :)
In a "utopian" society, everyone would be assigned a universally unique identifier, and they would all be required to enter it to access their accounts, and that would make counting Real Easy :)
Cheerio,
Richard
I don't actually understand your plan. You want a) Google to identify Linux users, and then b) Google to agree to link to a certain website, and then c) users to go to that website and perform some action to be counted. If Google is capable of identifying Linux users, and if we're going to be trying to persuade them to do something with that data anyway, why on earth not just ask them for the number of session keys they generate in the year for Linux users? If you wanted to get fancier than that, you could get firefox (and epiphany and konqueror, but mostly firefox) to include information in the User-Agent about which desktop was being run, even if it was only to Google; most people wouldn't turn it off or fake the data.
The smolt project is trying to do this with Fedora, and is designed to be usable in other distros as well.
http://popcon.debian.org
Due to google's crappy treatment of Konqueror, majority of KDE'ers fake as Safari when using google properties.
So, as a bonus, google-based Gnome-detection may actually make it look like a more popular desktop.
Kudos for an excellent plan.
there is The Linux Counter
http://counter.li.org/
Canonical has a good idea of the number of Ubuntu machines installed since they 'call home' via NTP. The default NTP server being ntp.ubuntu.com. This of course really shows you the number of IP addresses. But by considering the frequency of requests you could guestimate the number of machines behind a NATed IP address.
@daniel: Most KDE users use firefox anyway. Just like most Gnome users do... firefox doesn't tell which desktop it uses, does it?
But, what counts most is that firefox and konqi report they're linux, and we can count linux users...
People around the world are using Google Analytics on their web pages. So probably Google has some statistical data about users around the world, the question is if they want to share this data.
google already knows this, all we need to do is ask
Post a Comment