I am constantly puzzled that NetApplications constantly publishes global GNU/Linux numbers that are way out of sync with reality. I have proof of one of the mechanisms:
“The Net Market Share data is weighted by country. We compare our traffic to the CIA Internet Traffic by Country table, and weight our data accordingly. For example, if our global data shows that Brazil represents 2% of our traffic, and the CIA table shows Brazil to represent 4% of global Internet traffic, we will count each unique visitor from Brazil twice. This is done to balance out our global data. All regions have differing markets, and if our traffic were concentrated in one or more regions, our global data would be inappropriately affected by those regions. Country level weighting removes any bias by region.”
CIA World Factbook reports for Indian Internet users, “61.338 million (2009)” Wait a minute! How many PCs have been shipped in India since 2009? They are shipping/selling about 10 million per year so there could be 30 million additional PCs in India compared to 2009 and far more users of the Internet. NetApplications reports Indian GNU/Linux share at 2.3%. Instead of that share lifting NetApplications’ global share, it may well be depressing it because they are giving India only half the weight it should have in the sums.
M$ complains about weighting here, but they don’t worry that emerging markets are under counted because of the stale CIA numbers…
Who’s the biggest share of Internet users? China with 389 million users of the Internet in 2009. China has 1/3 of its 1.3billion people on the web, 460 million while CIA reports 389 million. Some error.
Internet users:
| Country | CIA | |
|---|---|---|
| Brazil |
76m |
79m |
| Russia |
41m |
61m |
| India |
61m |
92m |
| China |
389m |
460m |
| Malaysia |
15m |
16m |
| Cuba |
1.6m |
1.8m |
Google uses World Bank numbers from 2010! Imagine how different the numbers are for emerging markets in 2012! There’s a reason they are called “emerging”. There’s a reason NetApplications uses “country weighting”. It’s to average out the uneven sample of the world’s PCs by country. A major flaw in their strategy is to under-count emerging economies which are not small.
We have already seen that NetApplications undercounts GNU/Linux used in government and education because of the use of LTSP or NAT at routers (NetApplications uses “unique visitors” and cannot tell PC A from PC B if both have the same IP address and User_agent string.) The same effect undercounts that other OS but it is in a sea of conumers’ PCs bought from retail shelves stocked mostly with that other OS.
17481
12766
206
3
2
23928 
11870 
11722 
4635 
4268 
1641 
198 
14 
2 
0 
0 
0 
Yada yada yada… all stats are biased, Linux is everywhere on the desktop, but it’s invisible… It’s a conspiracy… Pogson angry.
Put on a new record, will you?
Chris Weig there is a basic error idea.
Take a number with a percentage error x it by another number with a percentage error get a bigger percentage error. Do this enough times and you have a random generated number.
One of those random numbers just might match the version number of Opera when it used KHTML source code, as Oiaohm claims.
Love that phrase ‘Internet Users’, just what does that mean? In most of the world people are lucky to have dialup. Except in the downtowns of a few major cities that is all Australians have – dial up. Oh, and lots of government talk about putting in a National Broadband Network, but not much action Billions and Billions (as Carl Sagan would say) dollars later.