Internet On-Line Polls

Resist the temptation to put on your own polls, or to believe the results of anyone else's. The results are never valid and are more likely than not highly misleading.

• First off: an on-line poll cannot be valid because it is a self selected poll. Even if the objective is to find the opinion of those who select themselves, participation rates will vary according to the answers.

• Secondly: only a very select set of people will even know about any particular on-line poll. It will never be a valid demographic cross section.

• Third: the questions, as with most targeted polls, such as those put forth by political "researchers" are likely to be tailored to get particular answers.

• Most Important, if the poll is on any meaningful subject, it will be stuffed and distorted. The internet enables very fast communications, and the word to stuff the poll will be quickly broadcast within particular interest groups.

• On September the 14th, 2000, the day Windows ME was released, all beta testers of the product received an email from within Microsoft urging them to go to a ZDNet poll and stuff the ballot box. "Vote early and vote often". Details at The Register

• On September the 14th, 2000, an item appeared in Linux Today about a poll attached to an MSNBC article on Windows ME, urging Linux enthusiasts to vote for Linux. The author noted that "unfortunately" this site was a little more sophisticated than many, and would allow only one vote per person.

  Linux enthusiasts did go there, and the votes soon showed Linux is nearly as popular as Windows NT/2000 (the leading vote getter).

  This was obviously intolerable - the MS in MSNBC stands for Microsoft. Soon a "bot" was dispatched to do automatic voting. The bot started stuffing the ballot box at the reate of 200 votes per minute. At first all it's votes were for Windows NT/2000, but that made things look odd, so it was modified to proportion its votes to the various operating systems, except no votes for Linux.

  Somehow the "one vote per person" rule was suspended for the bot. Details and discussion at Linux Today.

  [Update: MSNBC took the abused poll down. The owners of the 'bot then turned it against another poll. That poll showd that only 9% of responding readers intended to upgrade from Windows95/98 to Windows ME. Within hours that poll showed more than 60% intended to upgrade. ]

• Early in 2000 Sm@rt Reseller magazine held a poll for its's readers to give their opinion between Novell's NDS (Novell Directory Services) and Microsoft's AD (Active Directory). Sm@rt Reseller polls normaly get a few hundred responses, but NDS votes were soon in the thousands. Then AD votes kicked in and exceeded NDS votes by many thousands.

  According to the magazine, Novell's people at least covered their tracks reasonably well, but Microsoft staffers either didn't know how, or didn't bother.

And Vice Versa

InfoWorld, a prominent computer technology weekley, had, for several years, put on an on-line poll to let the readers chose their favorite products for the year. As it happened, OS/2 was a winning reader's choice 4 years running. Stewart Alsop, editor in chief, swore each year it could never happen again - then he turned the editor in chief job over to Sandy Reed.

Sandy made the statement "Windows NT is clearly the future", and set out to make it true, come hell or high water. By time she left, this once highly respected magazine was (and still is) little more than an echo of Microsoft's marketing machine.

Anyway, a very high number of readers of InfoWorld's on-line Web site, IWE (for InfoWorld Electric), and participants in their on-line forum were technical professionals, and a great number of them were OS/2 users (they still are, but InfoWorld's forums are defunct and hardly anyone goes to their Web site any more). Naturally, OS/2 won the reader's choice poll a fifth year running.

Sandy Reed wrote in an editorial that the "balot box was stuffed by OS/2 zealots", that the results were invalid, and were withdrawn. Sandy absoutely refused to produce any figures or any evidence of ballot box stuffing. It is presumed, then, that she had no evidence and the figures were probably a fairly legitimate reflection of the site's on-line readers.

Sandy then organized a "scientific" pole that was by invitation only with the demographic carefully selected to assure a Microsoft win. Anyone with real technical skill was eliminated in favor of bosses and desk jockys. Neither the questions nor the rational were ever released for public viewing. Some of the published results were, however, quite amusing, showing a rather low level of product knowledge by the participants. Perfect for Sandy, who despised technical proficiency wherever she found it.

As you might expect, the on-line readers had plenty to say about Sandy's integrity, and exactly what so obviously unqualified a person was doing as editor in chief of a (formerly) technical publication (no, she wasn't accused of sleeping her way to the job - that wouldn't have been credible in view of the viewable evidence).

Incidentally, upon the total collapse and demise of InfoWorld's on-line IWE forums (due to an entirely incompetent "upgrade"), the participants opened their own forums at IWEThey where technical (and other) subjects are still discussed.