Momentum carries you up to 3,000 feet, and then you start to fall. Your co-pilot is going through his checklist to re-start the engines as you plummet downwards. Soon you’re racing through 1,500 feet. You need to act quickly. You’ve got three options. Try to get back to where you took off, try another airport close by, or ditch in the river that splits the metropolis below.

There can be fewer places richer in data than the cockpit of a modern aircraft. And fewer professions better trained in real-time decision-making specific to their situation than airline pilots. And in this case, the captain was a 57-year-old ex-fighter pilot with vast commercial airline experience.  So, with lots of data, finely honed decision-making skills and vast experience, how did he do? Well, he didn’t fallback on a well-rehearsed procedure, didn’t use his sophisticated instruments, or the latest in avionics computing power. He looked out the window. He used a heuristic known as the gaze heuristic.

As he wrestled with the plane,“Sully” could see the control tower of another nearby airport to his right and decided that if that control tower was moving up the windowpane, he wouldn’t make it, but if the control tower was moving down the windowpane, he could divert. Try as he might, he couldn’t get the control tower to move up the window pane. And so, he told air traffic control “…we’re going in the Hudson.”[i] Find the link to the audio is here.

A heuristic is a rule of thumb or decision-making approach that purposely excludes much of the available information, focussing on a few select cues. The heuristics and bias (H&B) school advocates that deliberately thinking through the facts is better than using a rule of thumb, provided you've got the time. Heuristics, while economical, can lead to bias or systematic errors of judgement[ii]. If there isn’t enough time for instance, then a heuristic will do, but at a cost to quality[iii]. In other words, a quick decision is worse than a slow one but sometimes you don’t have time. Or so the H&B school would have us think. But what if heuristics were better at decision-making regardless of time pressure?

The problem with complex decision-models according to the H&B school is that those models are “over-fitted” to past data and lack the flexibility to deal with new information[iv]. Heuristics are said to be at their best when dealing with uncertain, noisy environments, rather than with clearly known quantities. And noise and uncertainty are what we deal with most of the time, particularly when we are flooded with data as in the cockpit of an airplane. In the real world then, it looks like “satisficing” or making a good decision is better than optimising. Because in the real world, not everything is knowable. At least not in time to make a decision. Which is why real-world life and death medical decisions for instance are made with simple heuristics.[v]

And so, you’d have to wonder about the high priests of big data, with their cavernous servers jammed with information, feeding ever complex decision-models. Are they making things worse rather than better?