I don’t think this is the first time, actually, but I realized something crazy had happened today.
I got spam sent to my e-mail.
Yeah, okay, except that the subject line of one particular spam contained content that was interesting to me. Or, rather, it suggested there was content of interest to me in a place I may not have run into myself. Here’s the subject line:
“and because it’s a class in critical thinking as much as writing, I encourage open dialogue.”
I’ve gotten text from sites and books before, usually technical snips of one kind or another, but this time I decided to use google to find the source: PEDABLOGUE – A personal inquiry into the scholarship of teaching by Michael Arnzen. And, the source was interesting to me. Further, from there I found several other resources of interest to me.
So, it’s all because of spam. I found several things of interest because of, I’m assuming, random text captured from the web. The process would have been better, from my viewpoint, if the source link had been included, but that’s merely a usability issue. As spam tries more and more to socially engineer the recipient into reading, the text will appear more and more like lucid and meaningful content.
And, wasn’t the autonomous collection of content for the reader one of the primary promises of “agent” technology? So, what if spam filters also become interest filters. Right now filters like spamassassin mark things as spam. But, they could just as easily mark things as interesting also. Why not add a layer to mark content that is of interest and offer a technique that promotes it? Instead of a spam score added to the e-mail headers perhaps there would be an interest score, on which one could filter as well. Is anyone doing that, filtering e-mail as more interesting using a bayesian filter? I think I remember that being discussed, but did it get forgotten? After all, “Recognizing nonspam features may be more important than recognizing spam features.” [via, also]
(I can imagine a time when I might be able to sync my black- and white-list to a service that would use it to find new content for me and aggregate my personal filters with others … like Amazon or Netflix, but for all content types … but, such that I can export and import my filter data in a portable format. The catch is, of course, privacy and trust.)
Spammers, it turns out, are creating technology that trips toward an answer to the turing test. And, that’s an emergent quality because it’s not the outcome they really want, which is to get the recipient to signify the paid-for content.
And, that really is part of the trick:
“To the recipient, spam is easily recognizable. If you hired someone to read your mail and discard the spam, they would have little trouble doing it. How much do we have to do, short of AI, to automate this process?” [via]
It’s about creating better and better AI, on both sides. One pill to make you larger and one pill to make you small, so to speak. And, you know if that’s too hard to code, maybe we could have the gold farmers turk it for us? And, then when both sides want that labor, let the bidding war begin!