Information filtering and reputation will be evolutionary battlefields

By

When I wrote earlier Will our reputation systems be distributed? Probably not for a long time, a number of people noted that reputation systems will be gamed.

Absolutely. The more valuable a system is, the more people will try to game it. If reputation systems influence who people buy from, who they date, who they read, then massive efforts will be made to game the systems.

However the value of a reliable reputation system is such that it is worth doing anything possible to counter-act the gamers, and it is possible. The energy and money devoted to trying to game Google is extraordinary, yet these efforts have limited success.

In the last chapter of my book Living Networks I made a number of predictions for the future of the living networks, including that Information filtering will be an evolutionary battlefield.

While reputation is a slightly different space from information filtering, both are absolutely evolutionary battlefields where each party’s tactics will evolve in response to the others’, all the while creating valuable outcomes for users. Here is the excerpt from Living Networks.

Information filtering will be an evolutionary battlefield

Bats’ use of echolocation to find their prey is one of the marvels of nature. Bats produce high-frequency sounds, and by picking up and distinguishing the immensely quieter echoes off insects in the air, can instantaneously calculate the location of their next meal. The evolution of this extraordinary capability has led to moths evolving in response. The soft outside of their wings and bodies absorbs the bats’ ultrasound. Moths engage in evasive flying stunts when they hear bats squeaking. Some moths have even evolved the ability to produce ultrasound as well, possibly to startle and throw off bats. In turn, bats have developed complex flying behaviors to confuse moths, and occasionally turn off their echolocation to stop the moths jamming their signals.

This is a case of what biologists call a “coevolutionary arms race”. Each participant in a system evolves new capabilities and behaviors in response to others’ development, in turn requiring them to evolve yet further in order to survive. There are many parallels in human society, not least the planet’s very real arms races. One of the best examples in the years and decades ahead will be the coevolution of information dissemation and filtering, involving battles of words, legislation, and more than ever, technology.

AdSubtract is one of a wealth of programs available that remove advertisements from web pages. You can surf at will, and never see an ad. David Mann of the University of Toronto sees this kind of functionality going a lot further. He has designed spectacles that take into account your head and eye position, and replace anything you don’t want to see with the images of your choice. His favorite example is a man standing at a urinal replacing a condom advertisement on the wall with a film of a soothing waterfall. You could just as well replace every billboard on your route home with pictures of your spouse and children.

Jamie Kellner, chairman and CEO of Turner Broadcasting, was quoted in early 2002 saying that skipping commercials when watching television programs is theft. Fortunately, he did allow that there was some scope for taking bathroom breaks. Phew. Several commentators recalled the scene from the book and film A Clockwork Orange, in which Alex has his eyelids forcibly held open as he is shown videos for his reeducation. Hopefully our legislators will stop before we get to that point.

We know as a certainty that we will be swamped as marketers endeavor to reach us with their messages everywhere we look, everywhere we go. Companies are going to ever-greater lengths to get through to their target audiences. In response, individuals are trying to escape, using devices that make call center systems think their telephones are disconnected, setting up e-mail spam filters, and returning junk mail. As communication becomes increasingly digital, the nature of this will change. Agent technology, which we discussed in Chapter 4, will increasingly be charged with selecting and presenting to us only what we want to see from the onslaught of information. Indeed, this is the domain of some of the most promising—and useful—current developments in artificial intelligence. In response, marketers will develop technologies to attempt to fool and bypass those filters.

The battle is engaged. As people find ways of filtering advertisements and messages effectively, marketers will find ways around them, leading to yet further advances. Technologies will be pitted against other technologies in a coevolutionary dance. At some stage, we will become onlookers as our agents engage in information warfare on our behalf.

  • Ross, eBay is a working example of a reputation system/network that combines both the quantitative and qualitative aspects needed for reliable and trustworthy participation.
    * Global – tens of millions of live participants
    * Relied upon – influences hundreds of personal and business transactions per minute
    * Challenging to game – multi-dimensional ratings combined with safety audits and tied to actual monetary transactions
    There are methods to “game” eBay’s feedback and detailed seller rating (DSR) system; however, they require a lot of effort, time, money, and run the risk of getting your account revoked. eBay has made it counter-intuitive for astroturfing feedback seekers to game the system.
    I’m not suggesting the system is perfect; however, you can rely on their rating/reputation system, as a buyer, with confidence. I believe eBay’s reputation system is worth studying, from its inception to its ongoing evolution.
    Take care,
    Neil

  • Hi Neil, yes eBay is one of the first and certainly broadest reputation systems, and there’s much to learn from it.
    However it is flawed in a number of ways, not least that there is no incentive to give accurate feedback if it isn’t perfect. I wrote about some aspects of this a couple of years ago: http://bit.ly/cf2hiC
    The other issue is that it only functions in the eBay ecosystem. We need reputation measures that transcend a particular marketplace.

  • Tessa Ye