June 8, 2006, 6:46 p.m. (Message 45469, in reply to message 45459)
Anselm wrote: | David Yee wrote: | > In defence of Canada, the number of murders in Canada in 2003 was 548, or | > 1.73 per 100,000 of population. The numbers for 1998 would be similar. | | The original number does seem rather high, but that's not really the point | (even with 548 murders, Canada is not 182 times as dangerous as the Vatican). | | The context of Rick Moen's quote comes from the practice of counting the | various security advisories regarding Microsoft Windows on the one hand and | Linux on the other hand. This is essentially comparing apples to oranges; the | details are too non-SCD to go into here, but the comparison remains as | ridiculous as it was even with the lower numbers. Actually, the statistical reporting for computer problems is worse than this. It's common to count bug reports and security advisories for every distribution of linux. Since there are more than a hundred distribution (some very specialized), this radically inflates the figures. To compare with the murder statistics, it's as if one were to count the murder reports from all publications in the Vatican and in Canada. So a murder reported by 30 Canadian newspapers and radio/tv stations would be counted 30 times. This way, 548 murders could easily become 150,000 "murder reports". Such a number would not necessarily be wrong. If you were studying the media's reporting of incidents, you would want to count the reports. But it's all too easy for politicians and marketers to describe such numbers in a way that confuses "incidents" with "incident reports". The linux crowd is very familiar with this sort of misleading PR. In particular, consider the phrase "reported incidents". Does this count the incidents or the reports? Most readers will assume the former, but it often means the latter. I wonder if we could find a similar way to inflate the amount of SCD in an area? Can we find a way to pphrase the numbers so that people think we're counting the dance events, but we're actually counting the individual dances done at each event? Even better, can we count the dancers on the floor for each individual dance, and report the total such a way that readers think we're counting the SCD events? I'd bet that a good PR person could manage this ... ;-) -- _, O John Chambers <:#/> <xx@xxxxxxxx.xxx.xxx> + <xxxxxx@xxxxx.xxx> /#\ in Waltham, Massachusetts, USA, Earth | | ' `