Saturday, June 11, 2011
Information Overload: The Anatomy of a Delusion
Sometimes the proper use of a product is ignored, owing not to a lack of awareness of its effectiveness, but to a lack of an explanation of how it works. For example, in 1839, draping your bed with finely woven gauze curtains was known to ward off malaria. Malaria, which means bad air, was generally considered to live up to its name, and be the literal result of well, bad air. Thus, hanging curtains around your bed presumably filtered the air, and thus helped prevent malaria. Nonetheless, going beyond mere curtain hanging to properly covering your bed with curtains didn’t catch on until a proper explanation of malaria was at hand that suggested as a matter of course the preventative measure of mosquito nets. But this didn’t stop another invention that stopped malaria just as well by treating it was thought, all that bad air. The invention was the inspiration of early 19th century Florida physician John Gorrie, who in an experiment sealed off a room and conditioned the air with a special device of his own invention. Of course, conditioning the air wasn’t the cause of malaria, but it had the incidental benefit of cooling it, and air conditioning lived on because it made you comfortable, and not because the system incidentally didn’t allow mosquitoes in. The explanation of mosquitoes as the carrier of malarial parasites resulted in the better use of available procedures that stopped mosquitoes, namely mosquito netting, and it provided the source of new procedures that stopped malaria (e.g. fumigation, swamp draining) as well as explaining why other procedures worked (e.g. air conditioning systems) and why other procedures of the day (e.g. handkerchiefs soaked in vinegar; garlic worn in shoes) that seemed to work, didn’t.
John Gorrie’s ‘Bad-Air’ Conditioner
Now consider a modern malady that doesn’t kill you, but manages quite well to kill your time. This is the modern bane of ‘information overload’. “Information overload refers to the difficulty a person can have understanding an issue and making decisions that can be caused by the presence of too much information” (Wikipedia). A logical problem with this definition is that we have always been in the presence of too much information, as a simple walk through any library can demonstrate. Before the internet, navigating this wealth of information was rudimentary and difficult. You used a card file to determine what you needed and walked around different book stacks to find it. Invariably what you found was often not exactly what you needed, but it had to do because the transaction cost for information, namely rifling through card files and roaming book stacks simply was too high.
In the internet age, our filters are immeasurably better, and we can get information tailored to fit our request, or employ intelligent agents that use a mere history of our internet behavior to find the information we need. Moreover, we can get this information for a negligible transaction cost, or for free. The problem is as any web search demonstrates is that we are handed with many variations of the same information, or information that is nearly redundant. In other words, different variations of the same information in different sites essentially restate the same information. For example, perform an internet search for ‘high gas prices’ and you will get a score of links to different articles that discuss high oil prices. If we were rational agents, we would read one or two articles and then cease, knowing that the marginal usefulness or utility or reading a third or fourth article decline markedly as they would restate the same information and generally arrive at the same conclusions. But the fact is we don’t. Indeed, we may read many more articles on the topic and even far removed from the topic, and then come back to the web after a few hours to read more. The same goes for any matter of internet searching, whether it is social media, email communications, or just looking up a sports score.
Ultimately, the problem of information overload derives from the implicit assumption as to what type of information is actually overloading us, and what overloading means. To gain our attention, information must not just represent random correlations between events, but rather correlations between events that have predicted value or utility. But information also has a degree of novelty, and it is the integration of novelty and utility and not utility alone that determine the importance or ‘incentive salience’ of moment to moment behavior. Utility and novelty are integral aspects of information, and even if information has marginal utility it still becomes an object of desire if that information is novel. Thus an individual may initially access his email because its utility far outweighs the novelty of discovery, but by the fortieth time in the day that he checks his email, the novelty of the act far outweighs its utility. Moreover even if we know the behavior has low utility, we often keep on searching. In addition, we are affectively primed to search for new information by in effect (or perhaps in affect) looking forward to accessing our email, social network, news feed, etc.
The concept of information overload implies that we are deciding between an abundance of information that is of uniform utility, but this is not true. Because of well honed filtering systems provided by the web, we can generally find what we want quickly. The problem is that the internet generates nearly redundant information that is distinguishable not by its usefulness, but by its novelty. In other words, we are not overloaded with information that is primarily useful, but are overloaded with information that is primarily novel. By generating infinite variations of the same information, the internet primarily generates not useful but novel information, and the overload is not in the information we want and need, but in information we want but don’t need.
So the problem, as the example of Gorrie and his air-conditioner demonstrates, is that ‘information overload’ like ‘bad air’ is false cause for a very real problem, mainly the web as a source of distraction rather than value. So the real solution is not to create better filters, since they ultimately don’t matter, but rather to limit our access to the web to those times when it logically means something, not when it doesn’t. In other words, the solution is not found in better filters, but in merely draining the swamp.