2011 : WHAT SCIENTIFIC CONCEPT WOULD IMPROVE EVERYBODY'S COGNITIVE TOOLKIT? [1]

jaron_lanier's picture [5]
Computer Scientist; Musician; Author, Who Owns The Future?
Cumulative Error

It is the stuff of children's games. In the game of "telephone," a secret message is whispered from child to child until it is announced out loud by the final recipient. To the delight of all, the message is typically transformed into something new and bizarre, no matter the sincerity and care given to each retelling.

Humor seems to be the brain's way of motivating itself — through pleasure — to notice disparities and cleavages in its sense of the world. In the telephone game we find glee in the violation of expectation; what we think should be fixed turns out to be fluid.

When brains get something wrong commonly enough that noticing the failure becomes the fulcrum of a simple child's game, then you know there's a hitch in human cognition worth worrying about. Somehow, we expect information to be Platonic and faithful to its origin, no matter what history might have corrupted it.

The illusion of Platonic information is confounding because it can easily defeat our natural skeptical impulses. If a child in the sequence sniffs that the message seems too weird to be authentic, she can compare notes most easily with the children closest to her, who received the message just before she did. She might discover some small variation, but mostly the information will appear to be confirmed, and she will find an apparent verification of a falsity.

Another delightful pastime is over-transforming an information artifact through digital algorithms that are useful if used sparingly, until it turns into something quite strange. For instance, you can use one of the online machine translation services to translate a phrase through a ring of languages back to the original and see what you get.

The phrase, "The edge of knowledge motivates intriguing online discussions" transforms into "Online discussions in order to stimulate an attractive national knowledge" in four steps on Google's current translator. (English->German->Hebrew->Simplified Chinese->English)

We find this sort of thing funny, just like children playing "telephone," as well we should, because it sparks our recollection that our brains have unrealistic expectations of information transformation.

While information technology can reveal truths, it can also create stronger illusions than we are used to. For instance, sensors all over the world, connected through cloud computing, can reveal urgent patterns of change in climate data. But endless chains of online retelling also create an illusion for masses of people that the original data is a hoax.

The illusion of Platonic information plagues finance. Financial instruments are becoming multilevel derivatives of the real actions on the ground that finance is ultimately supposed to motivate and optimize. The reason to finance the purchasing of a house ought to be at least in part to get the house purchased. But an empire of specialists and giant growths of cloud computers showed, in the run up to the Great Recession, that it is possible for sufficiently complex financial instruments to become completely disconnected from their ultimate purpose.

In the case of complex financial instruments, the role of each child in the telephone game does not correspond to a horizontal series of stations that relay a message, but a vertical series of transformations that are no more reliable. Transactions are stacked on top of each other. Each transaction is based on a formula that transforms the data of the transactions beneath it on the stack. A transaction might be based on the possibility that a prediction of a prediction will have been improperly predicted.

The illusion of Platonic information reappears as a belief that a higher-level representation must always be better. Each time a transaction is gauged to an assessment of the risk of another transaction, however, even if it is in a vertical structure, at least a little bit of error and artifact is injected. By the time a few layers have been compounded, the information becomes bizarre.

Unfortunately, the feedback loop that determines whether a transaction is a success or not is based only on its interactions with its immediate neighbors in the phantasmagorical abstract playground of finance. So a transaction can make money based on how it interacted with the other transactions it referenced directly, while having no relationship to the real events on the ground that all the transactions are ultimately rooted in. This is just like the child trying to figure out if a message has been corrupted only by talking to her neighbors.

In principle, the Internet can make it possible to connect people directly to information sources, to avoid the illusions of the game of telephone. Indeed this happens. Millions of people had a remarkable direct experience of the Mars rovers.

The economy of the Internet as it has evolved incentivizes aggregators, however. Thus we all take seats in a new game of telephone, in which you tell the blogger who tells the aggregator of blogs, who tells the social network, who tells the advertiser, who tells the political action committee, and so on. Each station along the way finds that it is making sense, because it has the limited scope of the skeptical girl in the circle, and yet the whole systems becomes infused with a degree of nonsense.

A joke isn't funny anymore if it's repeated too much. It is urgent for the cognitive fallacy of Platonic information to be universally acknowledged, and for information systems to be designed to reduce cumulative error.