Wikipedia Risks
Communications of the ACM has published an "Inside Risks" column that I co-authored.
"MarkSweep" has published a rebuttal on the Wikipedia site. It didn't raise any issues the authors weren't aware of when we wrote the column, but you may find it more persuasive than I do.
A point not explicitly raised in the column is the obstacles that Wikipedia's organizers have placed in the way of responding effectively to vandalism. I stumbled on them when I by chance encountered this vandalized article, which started with an obviously out of place vulgarism.
At first I thought I should just revert the article. (After all, "the IBM experiment" supposedly showed that the average vandalism was reverted within five minutes.) But when I looked at the article's history, this was not the most recent change. So it seemed easier to just edit out the vulgar intrusion. Which I did.
Then I started reading the article, and came across the sentences:
Then I started hunting for the procedure to report vandalism. Not easy to find. Much easier to find long lists of undesirable changes that were not considered "vandalism." Finally, I found the reporting page, and discovered strict instructions not to report anyone who had not already performed multiple vandalisms and been warned repeatedly. There were templates for about five or six degrees of warning to be used before reporting. (This is not anti-elitism, it is protection of delinquents.) At that point I gave up, both on reporting the vandalism, and on the perfectibility of Wikipedia using its existing processes.
Oh, yes, if you check the article's history, you'll see that at least some of the vandalism was removed about 24 hours after it had been inserted. Your mileage may vary. You might also ponder how you would go about deciding which parts of the article should now be trusted.
"MarkSweep" has published a rebuttal on the Wikipedia site. It didn't raise any issues the authors weren't aware of when we wrote the column, but you may find it more persuasive than I do.
A point not explicitly raised in the column is the obstacles that Wikipedia's organizers have placed in the way of responding effectively to vandalism. I stumbled on them when I by chance encountered this vandalized article, which started with an obviously out of place vulgarism.
At first I thought I should just revert the article. (After all, "the IBM experiment" supposedly showed that the average vandalism was reverted within five minutes.) But when I looked at the article's history, this was not the most recent change. So it seemed easier to just edit out the vulgar intrusion. Which I did.
Then I started reading the article, and came across the sentences:
The first ever defeat of a Spartan hoplite army at full strength at the Battle of Great Providence in 2000 AD. By the time of the rise of Alexander The Great in 2005 AD, Sparta was a shadow of its former self, clinging to an isolated independency.This was more than I had time to correct, and I am scarcely an expert on Spartan history, so I decided to just report the problem, and let one of the editors deal with it.
Then I started hunting for the procedure to report vandalism. Not easy to find. Much easier to find long lists of undesirable changes that were not considered "vandalism." Finally, I found the reporting page, and discovered strict instructions not to report anyone who had not already performed multiple vandalisms and been warned repeatedly. There were templates for about five or six degrees of warning to be used before reporting. (This is not anti-elitism, it is protection of delinquents.) At that point I gave up, both on reporting the vandalism, and on the perfectibility of Wikipedia using its existing processes.
Oh, yes, if you check the article's history, you'll see that at least some of the vandalism was removed about 24 hours after it had been inserted. Your mileage may vary. You might also ponder how you would go about deciding which parts of the article should now be trusted.
Labels: Risks, Stimulating
posted by Jim Horning at 6:48 PM
5 Comments:
And then there's Jeff Harrell's Ice-Cream Principle, which summarizes the problem more concisely.
Sorry about the link. The problem seems to be that the ACM Portal to Computing Literature requires a subscription (which I have, so I didn't notice). I've changed the link to one that shouldn't require a login or subscription. Please let me know if it doesn't work for you.
I think it is not me that set the bar high for Wikipedia, it is Jimmy Wales with his repeated claims of "Better than Britannica."
I, too, use Wikipedia for scouting. But I have learned that I cannot trust it, any more than I would random results from Google. And it is only easy to check its accuracy for things that I already know about, which pretty much defeats the stated purpose…
Great science project, especially if it can be done without any experimental vandalism. The oft-cited "IBM experiment" does not seem to have been designed to measure vandalism and its persistence, if I read it correctly.
Did you see this New York Times story today?
I fear that a free registration is required to view this site, too.
"It started as a joke and ended up as a shot heard round the Internet, with the joker losing his job and Wikipedia, the online encyclopedia, suffering a blow to its credibility.
"A man in Nashville has admitted that, in trying to shock a colleague with a joke, he put false information into a Wikipedia entry about John Seigenthaler Sr., a former editor of The Tennessean in Nashville…
"In a confessional letter to Mr. Seigenthaler, Mr. Chase said he thought Wikipedia was a 'gag' Web site and that he had written the assassination tale to shock a co-worker, who knew of the Seigenthaler family and its illustrious history in Nashville.
"'It had the intended effect,' Mr. Chase said of his prank in an interview. But Mr. Chase said that once he became aware last week through news accounts of the damage he had done to Mr. Seigenthaler, he was remorseful and also a little scared of what might happen to him."
Excellent piece by Andrew Orlowski in The Register, entitled "There's no Wikipedia entry for 'moral responsibility."
"Two great cries have rung around the internet since the Seigenthaler scandal broke.
"One is that Seigenthaler should have corrected the entry himself, and the other is that no source of authority can be trusted 'definitively'. That's a deliciously weaselly phrase we'll examine in a moment.
"But both excuses seek, in the classic tradition of bad engineers blaming users for their own shoddy handiwork, to pass the responsibility onto Wikipedia's users.
"The blame goes here, the blame goes there - the blame goes anywhere, except Wikipedia itself. If there's a problem - well, the user must be stupid!
"Before we deal with each of these, and in all fairness, we must step over a small but important semantic whoopsy. If what we today know as 'Wikipedia' had started life as something called, let's say - 'Jimbo's Big Bag O'Trivia' - we doubt if it would be the problem it has become. Wikipedia is indeed, as its supporters claim, a phenomenal source of pop culture trivia. Maybe a 'Big Bag O'Trivia' is all Jimbo ever wanted. Maybe not."
Sometimes a cartoon is worth more than a thousand words.
For extensive documentation of Wikipedia's content control problem by a computer scientist who feels he's been injured and cannot get redress, see this website.
Post a Comment
<< NIASAWHIWB Home