26 April 2009

Immersed in deception

Filed under: deception, intelligence, spam — David Wood @ 2:43 pm

Over the last few weeks, I’ve received a lot of flattery and what looks like friendly advice.

Here’s an example:

Ah! This is the sort of thing I have been looking for. I’m doing some research for an article. You should add buttons to the bottom of your posts to digg, stumble, etc your content. I think this is great and want to share it, but as it stands, I’m a lazy lazy person. Just kidding!

And here’s another:

I’ve just found your blog and I really like it. This is the first time I’ve written a comment. I’m not sure what to say, but please keep up the good work!

I found these compliments while checking the comments posted in reply to my own postings – either here, on my personal blog, or on the Symbian corporate blog.

At first, I felt pleased. Then I realised I was being deceived. These comments were being placed on my blogs, simply to tempt unwary readers to click on the links in them. These links lead to sites promoting bargain basement laptops, products made from the Acai “super berry”, and numerous other wild and wacky stuff (much of it not suitable for work). Now that I’m aware of these “link bait” comments, I notice them all over the web. They’re presumably being generated automatically.

The Symbian corporate blog is hosted by WordPress and relies on a service from Akismet to sort incoming comments into “pending” and “spam”. On the whole, it does a remarkably good job. But sometimes (not too surprisingly) it gets things wrong:

  • There are false positives – genuine messages that are classified onto the spam list
  • There are false negatives – deceptive messages that are classified onto the pending queue.

The task of sorting comments becomes even harder when “linkbacks” are taken into account. By default, WordPress lists “pingbacks” and “trackbacks”, when other blogs reference one of your articles. I haven’t yet made up my mind how useful this is. But I do know that it’s another avenue for deceptive postings to get their links onto your webpage. Some of these other postings re-use text from the original posting, chopping it up to give the appearance that a human being is providing intelligent analysis of your ideas. But again, it’s now my view that these postings are being generated algorithmically, just in order to receive and harvest incoming clicks.

Companies like Akismet are clearly involved in some kind of escalating arms race. As they learn the tricks employed by one generation of spam-creating program, another generation finds ways to mask the intent more skilfully.

I guess it’s like the way human intelligence is often thought to have emerged. According to widespread opinion, early humans existing in large groups found it beneficial to be able to:

  • Deceive each other about their true intentions;
  • Pretend to be supportive of the ends of the group, but to free-ride on the support of others when they could get away with it;
  • See through the deceptive intentions of others;
  • To keep track of what person A thinks about what person B thinks about person C…

This kind of evolutionary arms race was, according to this theory, one of the causes of mushrooming human brain power.

For example, to quote from Mario Heilmann’s online paper Social evolution and social influence: selfishness, deception, self-deception:

This paper endeavors to point out that the selfish interests of individuals caused deception and countermeasures against deception to become driving forces behind social influence strategies. The expensive and wasteful nature of negotiation and impression management is a necessary and unavoidable consequence of this arms race between deception and detection.

Natural selection created genetic dispositions to deceive, and to constantly and unconsciously suspect deception attempts. In a competitive, selfish, and war-prone world, these techniques, proven in billions of years in evolution, still are optimal. Therefore they are reinforced by cultural selection and learning. Conscious awareness of deception and countermeasures is not required, often even counterproductive. This is so because conscious deception is easier to detect and carries harsher sanctions.

Humans not only deceive, but also deceive themselves and others about the fact that they deceive, into believing that they do not deceive. This double deception makes the system so watertight, that it tends to evade detection even by psychologists.

Deception may be widespread in human society, but the associated increase in brainpower has had lots of more positive side-effects. I wonder if the same will result from the rapid arms race in electronic deception and counter-deception mechanisms – and whether this will be one means for genuine electronic intelligence to emerge.

Blog at WordPress.com.