October 03, 2006

Martin Luther King and The Unpleasant Search Algorithm Result

I'm a bit late in punditing about the CNET article noting:

Using the keywords "Martin Luther King," the first result on Google and AOL--whose search is powered by Google--and the second result on Microsoft Windows Live search is a Web site created by a white supremacists group that purports to provide "a true historical examination" of the civil rights leader.

Microsoft says:

The results on Microsoft's search engine are "not an endorsement, in any way, of the viewpoints held by the owners of that content," said Justin Osmer, senior product manager for Windows Live Search. "The ranking of our results is done in an automated manner through our algorithm which can sometimes lead to unexpected results," he said. "We always work to maintain the integrity of our results to ensure that they are not editorialized."

And Nick Carr:

By "editorialized" he seems to mean "subjected to the exercise of human judgment." And human judgment, it seems, is an unfit substitute for the mindless, automated calculations of an algorithm. We are not worthy to question the machine we have made. It is so pure that even its corruption is a sign of its integrity.

This is a good jumping-off point to note why I don't have a "home", for what I think of as technology-positive social criticism. Because my instinct here is not to bemoan the corruption of the machine, but to say, "That's what you asked it [the machine] to do". That is, if you ask the machine to tell you, very roughly, "What's the most popular site for this phrase", and it tells you something you don't like, well, that's the exact opposite of corruption. Sure, it's possible to ignore it - but that opens up a whole range of problems. Such as, which results are now going to be deemed so offensive to public sensibility that they'll be suppressed?

More importantly, if you start playing favorites, people are going to wonder if every oddity is the result of pressure groups - or should be subjected to manual adjustment. And search engines have enough problems with people falsely believing their personal sites been censored.

Think of it as "Government of laws, not of men".

By Seth Finkelstein | posted in google | on October 03, 2006 09:30 AM (Infothought permalink)
Seth Finkelstein's Infothought blog (Wikipedia, Google, censorware, and an inside view of net-politics) - Syndicate site (subscribe, RSS)

Subscribe with Bloglines      Subscribe in NewsGator Online  Google Reader or Homepage


Perhaps prime number generators should exclude the number 13 given it has occult significance?

We must rid our supposedly pure algorithms of their inherent corruption.

If you hold up a mirror to mankind and don't like what you see, distorting the reflection does not actually change the gross into the aesthetic.

TV may have been massaged to present the US as 99% pure, enlightened and devout 'for the sake of the nation's children', but to massage Google to do the same seems...

...just as reasonable.

Posted by: Crosbie Fitch at October 3, 2006 10:26 AM

to omit offensive search results is censorship, plain and simple. the search engines automagically troll web pages and build indices to that content. if you go around purging the pointers to speech we don't like you become just as evil as those who commit heinous acts.

there is NO SUBSTITUTE for an intelligent human in this loop. if you search for something, it is up to you to decide whether any search result you get is valid for the purposes of your search, period. you cannot blame the tool; that is as absurd as saying that "guns kill people".

the solution is to allow everyone to have their say, period. whether that content in on the web, the radio, the TV, a billboard, written, scrawled on a cocktail napkin and let the individual choose to listen or not, to respond or not.

i personally do not like white supremacists but i will defend their right to say what they like. that is their god-given right as freeborn persons, the same right we all have. to summarily eject their websites from a public index of websites is censorship.

Posted by: dennis parrott at October 3, 2006 11:30 AM

You said:

Because my instinct here is not to bemoan the corruption of the machine, but to say, "That's what you asked it [the machine] to do".

My response:

Why did you respond this way? My reaction is, "Holy cow, we told the machine to do the wrong thing!" I guess that's why I was the earliest critic of PageRank, and even though I do all my own programming and sysadmin, my three years in grad school in philosophy is something I wouldn't trade for geeky credentials.

Posted by: Daniel Brandt at October 3, 2006 05:08 PM

Daniel - Probably because I don't see Google as any sort of Oracle Of Truth, but basically a popularity-mining algorithm. So the reaction strikes me as people saying oh-my-god racism is popular. Well, yes, it is. Now, I don't believe results are sacred - but I do think that when one starts saying that the results have offensive content, it should be edited out, that's taking on a huge social issue of who is regarded as so offensive to be removed. It's sort of the banning hate speech topic, in a diluted form.

Posted by: Seth Finkelstein at October 3, 2006 07:25 PM

I confess, Seth, I don't follow you. Correct me if I'm wrong, but didn't you, a couple years back, expose Google's ad hoc filtering system with the implication that it was some sort of scandalous distortion of a pure algorithm? Why didn't you congratulate Google on recognizing the shortcomings of PageRank by installing a corrective, in an effort by Google to make their results more socially responsible? I think that philosophically, you have the cart before the horse. Or, to put it in geek terms, you have the hard disk plugged in where the keyboard is supposed to go.

Posted by: Daniel Brandt at October 3, 2006 09:48 PM

A little exaggerated - the initial news scoop was by some Harvard folks. I wrote up an examination of what was going on. But, exactly, it's a similar problem. That was sites being put on a suppression blacklist because of legal orders from the associated country's goverment. It wasn't a PageRank issue, it was the very fact the sites were visible at all. Now the proposal is, what, exactly? That blacklist should contain sites which are legal, but offensive (hateful?)? It's certain possible - people can even argue it's desirable - but do you see what a problem it's creating?

Posted by: Seth Finkelstein at October 3, 2006 10:23 PM

Don't forget the bias accruing from controversy.

There's more incentive to create a web site to promote something of controversy and of consequent extreme interest to a minority, than there is a web site that most are likely to agree with and find nothing remarkable to say about.

How many "Martin Luther King was a jolly nice chap" websites are people going to create? Will they outnumber those that promote the contrary?

If the web becomes sanitised such that the greatest controversies permitted concern which end of an egg to open first, then we might as well go back to working in the paddy fields.

Posted by: Crosbie Fitch at October 4, 2006 05:30 AM