MSNBC Accused of Rigging OS Poll

KlausBreuer writes "According to the German news report Heise MSNBC has produced a poll for the most popular operating system. This time, the poll was rigged rather blatantly: Friday morning, Linux hat 28% (18.500) of the vote, but miraculously dropped to 3% by sunday evening (European time). It appears that 126.500 votes came in on Sunday - all of them for Windows." Now, not knowing people at MSNBC or anything like that, I would offer the possibly that someone ran a script against it. These things have been known to happen before. Thanks to Donald van de Weyer who pointed out that this originally appeared on LinuxToday.

Comments

Re:Online polls are meaningless

[Danse]

Ugh. This got modded up? While I can appreciate Godwin's Law in circumstances where it is warranted, some people can't seem to recognize a valid analogy even if it sits on their face and wiggles. When someone makes a comparison to Nazis or Hitler to discredit or silence the opposition, then Godwin's Law should certainly apply. However, when some Nazi-related idea or symbol is used simply to illustrate a point rather than to attack the opposition, why should it be declared invalid?

Personally, I think there should be a similar law invoked when people try to declare that the opposition supports child pornography if they don't agree with censorship. I lost count of how many times people tried to play the child pornography card in the discussion of the Freenet story earlier. It amounts to the same thing as playing the Nazi card really.

So let me get this straight...

[sheldon]

Kevin Reichard posts a number of links to polls on various ZDNet, MSNBC, CNN websites encouraging readers of linuxtoday to go out and stuff the ballot box.

Then on one of the polls their is a movement by some other group of Windows users to go out and stuff the ballot box in return.

And Linux Today accuses Microsoft of cheating?

I don't get it.

Maybe if it'd been a fair and realistic poll, but the Linuxtoday editors pretty much destroyed that when they decided to encourage stuffing the ballot box.

They sad thing is Kevin Reichard probably doesn't even realize the harm he is doing to the Linux community by encouraging online poll stuffing.

I suspect Kevin used to be a member of Team OS/2. :(

Slashdot goes unbiased?

[rotor]

Much applause to Hemos for taking a neutral stance and giving MSNBC the benefit of the doubt. I seriously doubt that they would have done something so obvious when one half of the MS/NBC partnership so desperately needs to start reconstructing their image as a company that ISN'T cutthroat and unfair.

Here's what happened

[Platinum Dragon]

Here's the deal:

LinuxToday got wind of an e-mail that went out to WinME beta testers that encouraged them to rig a poll at ZDnet regarding whether people would buy WinME. Soon after, LT found out about another poll. This was the MSNBC favourite OS poll. They encouraged LT readers to go vote in it, noting it wasn't as easily rigged as the ZDnet poll. A couple LT readers figured out the poll could be rigged by deleting a cookie MSNBC placed on a voting machine and checking for on future votes. It's likely the Linux number of 28% came from a flood of LinuxToday readers, a few using the cookie-delete trick, though I'd like to hope most didn't stoop that low. It should be noted that LT itself didn't promote poll-rigging in its own posts. At one point, Linux had more votes than WinNT/2K.

Early Sunday morning is when the apparently faked votes started flooding in. One report from a reader claimed that for a while, all votes were going into NT/2K, then switched to adding votes for every OS but Linux - the percentages for Mac, BeOS, and Win9x/ME didn't significantly change like the NT and Linux counts.

It looks like the whole thing is a popularity pissing contest. LT is still encouraging their readers to vote (fairly) in polls that have appeared in the last few days, and LT released an open letter to MSNBC regarding the sudden, suspicious increase in NT/2K votes.

Of course, if MSNBC were really carefully rigging things, they also would have rigged the other poll on the same page as the OS popularity poll - the one that showed only 8% of voters were going to buy WinME, opposed to 92% saying ixnay:)
-------------

Let me get this straight

[donutello]

So some online poll on some website is showing fluctuations. Where one option was leading earlier, another one is now. What am I missing? Why in the world is this news?

When was the last time anyone actually trusted or paid any attention to an online poll, anyway? If you do, I have a bridge in New York I'd like to sell you.

And the insinuation that MSNBC rigged the poll is preposterous at best. Besides the point that it's hardly worth the effort or risk for them to do it, the far more likely possibilities are that 1) someone with a script skewed the numbers or 2)that the initial spike was because of Linux Today asking its readers to vote on that poll or 3) The initial spike was because someone with a script pumped up the Linux numbers and MSNBC took those votes away or 4) Horror of horrors, more people actually Use NT/2000!!

So is this news worthy of posting on Slashdot because it involves Microsoft or because it involves Linux or because it slings some mud at Microsoft based on some pretend charges, hoping some of it will stick?

Re:Entia non sunt multiplicanda praeter necessitat

[digitalhermit]

This was not the only problem. A Linux/Netscape browser could not even get to the page to vote. If you clicked on "Complete Story" nothing would happen. The only was I was able to vote for Linux was to access it from an IE browser. In other words, the poll prevented non-IE users from participating. This fact is a lot worse than ballot stuffing because the latter could be blamed on users.

Slashdot Accused of Rigging Spork/Foon Poll

[vslashg]

Posted by Hemos on Monday September 18, @1:48PM
from the i-call-it-a-sporkle dept.

CharChar writes "According to MSNBC news, Slashdot has produced a poll for the name of that plastic spoon/fork combination you get at cheesy restaurants. This time, the poll was rigged rather blatantly: Friday morning, 'spork' had 28% (2,213) of the vote, but miraculously dropped to 3% by Sunday evening. It appears that 70,102 votes came in on Sunday - all of them for 'foon'." Now, not knowing people at Slashdot or anything like that, I would offer the possibly that someone ran a script against it. Still, occurances like this make you question the validity of Slashdot poll results, no matter how significant or important the question.

Re:Online polls are meaningless

[Wellspring]

Another way to (ab)use polls is to phrase the questions in a manipulative way. There, you don't care what the results are, you're using the 'scientific neutrality' of being a polltaker to lure people into believing what you say.

example:

PONDS: "Sir, I am from the Ponds Institute of Knowledge, and I wanted to ask you a few questions about skin cream. It is, after all, For Science.

JOHN Q. PUBLIC: "Well, since it is For Science, I'll just put my dinner on hold and answer a few questions for the sake of Pure Research."

PONDS: "Yep, that's right. So, sir, do you use skin cream?"

JOHN Q. PUBLIC: "Well, no, not really."

PONDS: "Do you suffer from any skin conditions?"

JOHN Q. PUBLIC: "No."

PONDS: "Are you experiencing symptoms of dermal dehydration? Is that a condition you suffer from?"

JOHN Q. PUBLIC: "I don't know what that is."

PONDS: "The symptoms of Dermal Dehydration are dryness of the skin, occasional irritation, and some sensitivity."

JOHN Q. PUBLIC: "Well, when I shave, it does sometimes get a little sensitive in places. Especially if I haven't shaved in a few days."

PONDS: "OK, then, I'll put you down as a yes for skin conditions in this Completely Academic survey."

JOHN Q. PUBLIC: "Wow, I really do have a skin condition. With a medical sounding name and everything! I wonder what the cure is?"

PONDS: "Well, that isn't our survey, sir, but on a Totally Off Topic Note, this condition is completely treatable with the application of a topical dermal hydrating application, known to Laymen Like You as a skin cream."

JOHN Q. PUBLIC: "Oh! I'll have to buy some!"

PONDS: "Good for you! Now, on to our last question. Are you aware of the fact that only Ponds (tm) skin cream hasn't been reported to cause festering sores, which ooze pus at the dinner table?"

Results: 100% hadn't heard that report, but 100% have now. Not that they know who reported it, or where...

Note, I totally and unfairly single out Ponds, which to my knowledge has never done a poll at all, let alone a 'push poll'. I pick ponds out because of that Ponds Institute they always talk about on TV-- kinda like the Halls of Medicine, or the Center for Bubbliciousness Studies (last my invention). But the point is that you can use a trick like a scientific survey to manipulate people without them ever hearing the results. And since these are usually targetted calls, your competitors never hear about the rumors you start until it is way too late.

Re:Online polls are meaningless

[PD]

The program did indeed run under Windows.

The program was structured like this:

1) autorun vote program
2) vote
3) reboot

Microsoft identified a bug in Windows 98 that caused the step #3 to execute slowly. You will find that the new operating system from Redmond, called "Windows Me Harder", fixes that bug causing step #3 to take much less time than before. Indeed, the ballot stuffing was much enhanced when run on the new operating system.

At least I got a good Sig

[Millard Fillmore]

The babelfish translation is worth it in its own right, and the story is mildly interesting. But the best part of it is the last line, as translated by the fish (which means it could be a real idiom or an artifact of machine translation):

"Trust no statistic which you did not falsify."

New Poll

[mooredav]

How many votes will you cast in this poll?

• 0
• 1
• 10
• 100
• while(1) { vote("CowboyNeal") }

Online polls are meaningless

[FascDot Killed My Pr]

The sample is non-random in several ways.

1) Only online people are included--less than 4% of the world is online. Probably moot in this particular example, though.

2) Only people who read the site see the poll. Imagine if they put a poll about "best skin color" at the end of "Mein Kampf"--do you think "black" would win?

3) Only those people who care enough to answer respond, even if they see the poll. A real poll goes out and asks people (on the street, on the phone, whatever) and requires an answer be marked down (even if it is "no response"). If an online poll marked a "no response" category for every webhit that didn't vote we'd see an alarming amount of apathy...

4) It is always possible to stuff the box--setting a cookie is useless since it can be deleted. Recording IP address is mostly useless since it doesn't work very well for dialup--not to mention privacy and spoofing issues.

In short, never pay any attention from an online poll. Magazine polls have most of the same problems, ignore them as well.
--
Linux MAPI Server!
http://www.openone.com/software/MailOne/