[KLUG Advocacy] Linux Outpacing Macintosh On Desktops

[Adam Williams]

>>If you survey a business that has, say ~100 desktops, on the whole those
>>machines will be pretty homogeneous.  The managers will have
>>documentation of the exceptions.  Maybe not across the entire business,
>>but certainly each business unit will possess this information.
>That's probably true (at least in well-run organization, or with well-
>run networks). I don't know how much you get out, so I don't know how
>much diversity you've seen in how well networks are run. My own experience
>is that there is a *lot* of diversity here.

I've seen some really terrible networks.  But at large sites they are
always able to describe the terrible-ness.

>There is also a lot of diversity in business, in terms of size, scale,
>coverage, degree of heterogenaity, security, up-to-datedness, and so
>on (I'm sure I've missed at least FIVE imprtant variables, and I wouldn't
>be surprised if some reader can enumerate a lot more).

There may be heterogenaity,  but the business will be able to describe
this.  They have to be able to do so,  or the SBA will string them up by
their b????s.

>>Home systems are likely to be misreported,  either deliberately or
>>because the person doesn't even understand the question.  Also home
>>systems may be dual-boot machines (pretty rare on a business desktop),
>>etc.. that muddies the meaning of the question.
>This is probably true, and surveyeos are adept at creating questionairres
>that double-check this sort of thing, perhaps asking the same questions a
>couple of different ways.
>>Q: "What OS do you use at home?"
>>A: "My system is a Compaq."
>Or
>Q: "What computer do you have at home?"
>A: "I just upgraded to XP Pro."
>That's right, and some question like this, when answered this way, 
>contribute (negatively in these cases) to the confidence we have in 
>these surveys. 
>On the other hand, good surveyors do not leave it at that, or they 
>supply questions that are more specific. Even a slight rewording,
>like "What Operating System do you use on your home computer?" is
>a clearer way to ask the above question. Telephone survey takers
>are also trained (again, assuming a "godd survey") to filter out
>unacceptable answers, with a bit of interactve prompting of the
>vic^H^H^H folks being surveyed.

Ok, I'll concede that.  It you have people to separate the wheat from
the chaff (or the ballots from the chad, or bearded mean from the
terrorists, etc...) at the point of contact you could make it work. 
Must be that alot of the numbers published in things like Infoworld are
simply not good surveys.