Just my $0.02 here, but I don't think anyone's going to be completely satisfied until/unless a really valid poll is conducted next year (or whenever the next one is).
Therefore, my recommendation would be to start planning _now_ for one that will satisfy all sides. Infoworld wants one that they can take to the advertisers, and that has a significant number of responses. The readers want to be counted. Coming up with a happy medium shouldn't be too terribly hard here.
IMHO, what went wrong with the first vote is that there was general apathy on the part of the readers that don't use OS/2, as witnessed by the horrible turnout and the significant number of OS/2 votes. As an aside, I want to state that I do NOT agree with Sandy's "They stuffed the box" conclusion. I think the NT/95/Linux supporters saw the Editor's Choice awards, and didn't bother to vote.
The second poll, which we have just seen the results of, is not relevant due to A) extremely small sample of only part of their readers, B) those outside the chosen few were not allowed to express their own opinion. In addition, I don't think that there was a significant overlap between Infoworld and Infoworld Electric readers. Many of the people I know who hit the Infoworld Electric web site do so at least once a day, if not more. Citing that the print readers in the poll hit it "Once a month" is a joke.
In my opinion, it would be relatively simple to do a real poll of the users, even though it wouldn't be "scientifically accurate". Simply use email:
Starting immediatly, require an email address from people signing up for both the print subscriptions and for forum access on IWE. Create a system for updating them due to changes. Explain that this address will only be for voting for the Reader's Choice poll.
When it comes time for the Reader's Choice awards (preferably before the Editor's Choice comes out to prevent swaying of votes), email everyone on the list a voting ballot. Include on the ballot a randomly-generated number (different one on each email), and log which number went with each email address. Include instructions to keep that number in the reply, and to make sure you are sending from the same email address it was sent to. When the replies start coming in, you have a program to check the number against the email address it was sent from. Give them a week or two to send it back in, then close the balloting, and post the results. Include a forum for the results.
There. No fraud possible. Since you're not advertising that the poll has started, you can't have a number of new subscriptions just to fill out a ballot. Since the numbers are checked against the email address, you can't have faked responses. Since it's via email, less chance of apathy. No votes being based on what the Editor's think since those awards haven't been announced. The only reason for complaints about the results would be either really serious apathy, or general lazyness.
If you wanted to go even further to make sure the addresses are those of readers, ask your webmaster's to set a logging feature and disqualify any registrations that haven't been used in the 30 days preceding the balloting. They should be able to do this fairly simply.
Questions, Comments, Discussion?
-steve
For questions or comments, email the Forums Editor