Windows vs. Linux vs. Mac vs. Windows Vista (Computers & Electronics)

An article in PC World reminded me that I had thought about doing a computer operating systems face-off in the past. However, this was made difficult in Google Trends due to the many non-OS uses of "windows" as well as "apple." Google Insights for Search can handle this better because it is able to filter for "Computers & Electronics" only. I decided to also include Windows Vista next to Windows in general.



As expected, Windows dominated worldwide but was declining steadily. Linux was second and also decreasing to the point where an ascendant Mac overtook it in December 2007. Windows Vista on its own was decidely not Google-popular: it ended with half the number of its introduction peak in February 2007. This illustrated its lagging acceptance. The data can be compared with market share evolution (source: Net Applications):

• 4th quarter of 2004
  

1. Windows 96% (of which Windows Vista 0%)
   
2. Mac 3%
   
3. Linux <1%
   

• 2nd quarter of 2008

1. Windows 91% (of which Windows Vista 15%)
   
2. Mac 8%
   
3. Linux 1%
   

Notice how Linux generates a lot more buzz than market share. I expected the results to be a bit different in the US alone:



Mac precedes Linux overall and overtook it more specifically already in August 2005. How about Japan?



Linux, albeit still second, is stronger but still is joined by Mac in September 2007. One more comparison: the UK:



Windows is totally dominant here. The cross-over of Linux and Mac took place early, in May 2005.

Leo Laporte vs. Guy Kawasaki vs. David Pogue vs. Walt Mossberg vs. Cali Lewis

Today, a face-off of five tech journalists:


Scale is based on the average worldwide traffic of "guy kawasaki" in all years
"leo laporte" 3.90
"guy kawasaki" 1.00
"david pogue" 0.60
"walt mossberg" 0.30
"cali lewis" 0.20

There's something odd about this 4-year graph—I ran the Google Trends search several times but it came out the same way each time. Notice how all journalists except Laporte have no data until what looks like exactly January 1, 2006? Hmm... fishy. For what it's worth, Laporte overwhelms the field. But let's look at the 12-month graph for whose time span we have a complete dataset:


Scale is based on the average worldwide traffic of "guy kawasaki" in the last 12 months
"leo laporte" 2.04
"guy kawasaki" 1.00
"david pogue" 0.76
"walt mossberg" 0.50
"cali lewis" 0.38

Laporte still is twice as "Google-popular" than Kawasaki. There's not much of a clear pattern otherwise. However, I thought I'd try the beta Google Insights for Search in this case, to see if it had data for all five journalists from the beginning of the 4-year span. Lo and behold: it did!


Scale is based on the worldwide 2004-present interest over time
"leo laporte" 45
"guy kawasaki" 14
"david pogue" 11
"walt mossberg" 9
"cali lewis" 3

So what's going on here? I would have thought that both Google tools use the same search-history database but maybe I'm wrong? Google Insights for Search is in many ways more powerful than Google Trends: for instance, it allows one to taylor a time span to any length, down to the level of months, e.g., March 2005 to November 2007. The numbers used are different: scaled to the average search traffic for one of the terms (Trends) vs. scaled (on a 100-point scale) to the total number of searches done on Google (Insights). Honestly, I'm not sure about the real distinction between the two approaches. I don't think that it makes much of a difference for my purpose. Also, the trend lines in Insights are too fine to allow for easy differentiation. Remains the strange omission of pre-2006 data for 4 out of 5 search terms in Google Trends...