Inflation in information

I take a wheelbarrow of facts to the shop and ask for one sensible decision. The cashier looks inside. Three thousand articles. Forty dashboards. A satellite image of my own roof.

“Have you got anything relevant?” she asks.

I explain that relevance is phase two. We are currently scaling the wheelbarrow.

This was supposed to be the deal: give us access to humanity’s knowledge and we will become wiser. Instead, I can inspect a distant galaxy while failing to choose a fucking frying pan. Somewhere between the third comparison video and the metallurgical history of breakfast, I have become less certain that I need food.

The data grow, the dashboards glow. What should we do? We do not know.

Claude Shannon’s information theory never promised to fix this. His framework deals with selecting and transmitting messages; its quantities do not measure how meaningful or useful those messages are to us. The engineering problem deliberately sets meaning aside. A channel can faithfully deliver a message that is no help whatsoever.^[1]

“Your trousers have been appointed Minister of Agriculture” can arrive with perfect fidelity. We have achieved communication. We have made no progress with lunch.

The cashier says my information is abundant, but the department that turns it into understanding has one employee. The employee is me. He is tired and has opened another tab to find out why.

Herbert Simon described the underlying problem decades ago: information consumes the recipient’s attention, so an abundance of information makes attention the scarce resource. The receiving end still has to do something with the delivery.^[2]

We have installed a fire hose above a teacup and are celebrating improvements in water pressure.

The teacup is also biologically awkward. Nelson Cowan’s review of short-term memory argued for a capacity averaging about four chunks under carefully controlled conditions. That is no universal four-fact limit: grouping, prior knowledge and the task all matter. It does, however, make the prospect of uploading civilisation directly into my forehead seem administratively optimistic.^[3]

There is a second problem. I can mistake having directions to the library for having swallowed the librarian.

In nine experiments, Matthew Fisher and colleagues found that searching online for explanations could inflate people’s estimates of their own knowledge, including on unrelated questions. The result concerns self-assessed understanding. It does not establish that the internet makes everyone stupid. Apparently it can help us feel qualified before the qualifications have arrived.^[4]

I skimmed a review. My confidence grew. I now know precisely what you ought to do.

This is the inflation I mean: the pile grows, while extracting something worth believing or acting on still costs time. Repetition adds bulk. Contradiction adds homework. A thousand copies of an assertion can look terribly substantial from the wheelbarrow.

Of course, information has delivered enormous value. An accurate forecast helps. So does the repair manual for whatever is currently leaking. More good evidence can improve a decision. My plumbing would object strongly to a romantic return to ignorance.

The exchange-rate metaphor has limits. A useful fact does not become less true because someone posts another cat. But finding it, checking it and understanding where it applies remain work. Counting the contents of the warehouse tells me very little about whether dinner has arrived.

And sometimes I already have the information. I know the difficult conversation needs having. I have simply commissioned further research into whether it might have itself.

“What does one sensible decision cost?” I ask.

“A clear question. Some relevant evidence. Time to think. Then you have to do something.”

I dislike the final condition. Is there a downloadable alternative?

We gathered the facts and admired the view. Then downloaded a guide to what we should do.