Machine Readable
[Caveat lector: It's another Google post. Does it seem like I'm picking on Google a lot? The company has so many fascinating services and products and gee-whiz stuff going on ...]
I've been considering some of Google's services, with the thought of machine indexing and relevance. Mind you, I'm pretty lost when it comes to search engine technology or algorithms of relevance ranking beyond the general notions of such things. But there's been a confluence of observations regarding relevance as assigned by Google, so it's at the forefront of my musings.
It actually started a while ago, when Gmail became a very popular beta but some people were concerned about privacy issues regarding the machine indexing of messages received and sent from Gmail. It's not an issue I hear much about nowadays, but I haven't been keeping up with Search Engine Watch.
Yesterday, I saw Jessamyn's post about GoogleAds and relevance. (Regarding the idea of persons finding them for sale on eBay, I'm sure that many people consider Jessamyn a treasure, but she is by no means a commodity ...) As it happens, it was just a half-hour after reading a column by Robin Peek of Information Today about Google News. She writes:
Has Google created the killer news app? Google News only provides its sources' previous 30 days of content and updates the content every 15 minutes. And the service prides itself on the fact that the results are "compiled solely by computer algorithms, without human intervention." This is a feature that actually forces a person to view a more extensive array of news sources than someone would normally encounter. ...But there is a certain randomness in which sources are presented. For example, items seem to be arranged in reverse chronological order. But on a customized section I created on the semantic Web, a March 9 item came before a March 14 item. The lack of human touch was also evident in another customized section on open access journals. My results included an article from the U.K. on how accessible the great outdoors was.
But this came to a head for me a few days ago. Recently, the New York Metro ran a story about a court case over allegations of 30-year-old accounts of child molestation at a prestigious boarding school -- the story has been somewhat prominent in the blogosphere given that the lead attorney is 1) quite famous within and outside of legal circles and 2) has admitted that he too was a victim of the abuse.
It's a very sober article, but it's displayed within what appears to be the standard template for articles on the site. Each of the 8 pages had a couple of color ads, and 5 had GoogleAds (the remaining 3 had Google Public Service Ads). Most of the Google Ads were easy to ignore but seemed semi-relevant to the topics discussed the pages (I saw, and continue to see, ads for Internet filtering software, self-defense courses, Vienna Boys Choir merchandise [the school specializes in boy choirs]). But then I got to page 7, and I really wish I had taken a screenshot of it.
The first 3 of the 5 GoogleAds on the site related to the Michael Jackson trial where he's accused of child molestation. I was stunned and appalled. So I went back and looked at all of the advertising. I didn't see those types of ads on any other pages. The thing is, the ads were relevant, as the overall subjects are identical. But it felt crass and wrong (and yes, I know there's much worse to be found on the Web, by design and intent). The fact that this was due to no human intervention, that it was decided by machine processes, just made it downright creepy in a way that's hard for me to articulate.
Google's feedback FAQ about GoogleAds/Ads by Google explains:
'Ads by Google' are ads that connect people to information about products and services that are relevant to the content they're reading online. Google technology understands the nuances of language, and closely matches (or "targets") ads and links to the specific content of web pages. For example, if you're reading an article about favorite pasta recipes, you might see ads for related items—like different kinds of pasta, cookbooks, olive oil and so on—on the web page.
Who knew that technology could be tacky? Okay, that's probably anthropomorphizing too much. But I wonder if I was the only person to feel this way. When I checked the article again yesterday, after reading Robin's and Jessamyn's comment, there were radically different ads in the same spot. And there is a way to provide feedback on specific ads to Google. Or maybe the ads had expired and the template was refreshed with new, relevant-but-different content.
I have no good answers to this and I'm not even sure I'm asking the right questions. There's no moral to this story, and it's not much of a story. I'm just trying to figure out if I can discern a pattern ...
