Statistics of use of a language

[Michael Bauer]

Before we get too excited, I'd like to query the ... likelihood of these 
stats.

I went and looked at those I'm most familiar with
Irish 5418
Gaelic 4570
Manx 1244

Now, I'm not sure what the starting point of these is but they look very 
wrong. At best, there are 100 Manx speakers in the world. Also, I get 
weekly blocklist reports from Mozilla which is essentially a very 
reliable way of measuring active users of Firefox and according to which 
there are some 100 regular users. Even if this is a cumulative figure 
over many years, both the Manx and the Gaelic figures are waaaay over 
what one might expect.

Or looking at it another way, Linux OS has an approximate share of 1-2%. 
So since there are less than 58,000 Gaelic speakers in the world, at 
best we would expect 1160 users or so at 2% penetration. Which is 
essentially impossible.

I have noted a similarly bizarre bloating in the Joomla langpack 
download figures and I have the sneaking suspicion that the stat counter 
there makes no difference between some bot downloading the pack for 
whatever reason and actualy users.

I don't know how the technical details of this work but perhaps we would 
get more reliable figures if we checked to see how many langpack 
downloads there were in conjunction with downloads of system software 
upgrades to weed out downloads of langpacks which don't actually 
represent systems in use.

Michael

04/07/2013 13:00, sgrìobh ubuntu-translators-request at lists.ubuntu.com:
> Original David's message can be found (with the references to the code)
> here:
>
> http://ubuntu.5.x6.nabble.com/LanguagePack-statistics-td1747839.html