What does Back to the Future teach us about the nature of predictions?
In the movie, there were many predictions about 2015, some spot on and some way off the mark. Now that we have reached the predicted date by more conventional means than a nuclear-powered DeLorean, what can we learn about the accuracy of 30 year predictions?
Does today’s date, October 21st 2015, seem familiar? That is because it is the date set by Marty McFly in back to the Future II. It’s always good to look at how futurists look at the present day, and what it can teach us about the effectiveness of our own predictions.
We overestimate the ability to overcome physical constraints
Many of the “hardware” improvements seen in the film seem unlikely even now. Flying cars, self-tying shoes, hoverboards, self-drying jacket all seem as though they are trying to cram too much power into not enough space. Most of these gadgets would be feasible if battery technology had improved appreciably in the intervening 27 years. That seemed like a reasonable bet for the writers, but battery improvements have remained frustratingly slow.
We underestimate information technology
Computer power, according to Moore’s Law (which was understood in 1987), increases one million-fold every 27 years, but it is still hard to guess how to use something that is a million times better. The writers generally did pretty well.
For example, Biff pays for a taxi with a thumb-print which turned out to be be pretty close to today’s reality: thumbprints only appeared five years ago and already seem old hat, and we are rapidly moving away from cash payments, with mobile payments and contactless technology such as London’s Oyster or Hong Kong’s Octopus. However, marks are lost for featuring a regular taxi: Uber, Hailo, Lyft and the rest seem to be showing that when Marty sets the dial for 2025 there won’t be too many taxis left.
The movie shows someone talking to his boss over what looks very like Skype, on a wall-mounted flat screen. I had the exact experience with a remote clerk at the San Francisco Airport Hertz desk yesterday. On the same day I also walked past the Beam shop in Palo Alto and was greeted by a motorized video screen featuring the face of a guy in India.
Fax machines, which were quite cutting edge in 1987, still make an appearance: it’s hard to see if the writer put them in there as tongue-in-cheek or not. But it reminds us that even extensive network effects can be overturned quickly. You would have shocked a British Industrialist in the 1840s who had just seen the emergence of canals all over England if you told him that soon, all of the infrastructure of canals would be overtaken by the next technology: railways.
Social changes are the hardest to predict
First thing I noticed, which is evident from just the picture above, is that the US will remain the last hold-out against the 24 Hour clock. (In the 24 Hour Clock was also predicted in the first sentence of another book set three decades in the future: 1984) Another is that in the shots of the small town, no-one is staring down at a device as they walk along. Compare that to the quick survey I did in Covent Garden last week, when I estimated that one quarter of the people I saw standing or walking were looking at their phone. In Marty’s 2015 there are no studs, tattoos or beards either.
On the nature of predictions
We always overestimate the change that will occur in the next two years and underestimate the change that will occur in the next ten. — Bill Gates
All businesses are in the predictions business. What will my customers want to buy? What will my cost basis be? Where can I get my raw materials from? We can either assume a regular progression from the current state or we can look for the germ of a disruption in the current market. This is essentially a signal-to-noise problem. Do I have the correct data, in an accessible up to date format to be able to recognize what may be a faint signal which could end up being a major innovation.
I just finished reading The Half Life of Facts by Samuel Arbesman. One of the themes is that we are very poor at assimilating new data into existing mental models. The Predictive Business needs to be able to access oceans of data, develop algorithms to look for insight and then be able to action the business change to take advantage of the opportunity. At SAP we call this the Digital Enterprise.