As far I read the comment on the methodology for this poll, that must be called "Infoworld Subscribed to the printed publication and working for computer related vendors (probably in US) choice (?) awards" or something like that... not exactly "Readers' choice award" at all, and as that, don't need to have relation with previous polls.
Worse than that, was a directed poll, done in a very specific group of subscribers, in a perhaps very specific geographic area. What if I poll about drug legallity to drug providers and say that the result of the poll was for all US population? Maybe a very extreme example, but perhaps show the problem about not really random samples.
There are also some other minor objections, as I don't know how exactly was done the questions (could be some influence here, in example if someone asked "what do you mean with server software" and the interviewer say "like Windows NT or so"), and the previous request for the boss of the reader (as I see in a previous post) that could interfere (?) with the freedom of choice of the subscriber.
Could exist a perfect way to poll? I doubt that, but some approachs could be better than others. Some sort of mail-autenticated poll could reach to most of the internet-subscribed readers, and that will point to choice because the reader could answer if he like to do that. Perhaps the poll also could be done to active readers (not only subscribed, also that in some moment readed forums or some other subscribed-specific feature of the internet version of Infoworld) or even a sample of subscribed users, but in a more random basis, without geographic or work limitations.
Perhaps the poll was valid, well conducted, but the proper name of the results is not Reader's Choice Awards at all.
For questions or comments, email the Forums Editor