My Bad Predictions

And why I made them …

I am frequently invited to offer my foresight on science and technology, and I accept with characteristic confidence that my instincts and enthusiasms are a reliable indicator of what is to come. I must, however, acknowledge that this confidence is frequently misplaced, and my predictions generally unreliable. I have cause, therefore, to contemplate my bad predictions and to understand what has given rise to them.

On the whole, amongst these bad predictions, I have made many more ‘false positive’ predictions – ascribing success to a technological development that does not then manifest – than ‘false negative’ predictions – anticipating failure for a technology that is subsequently successful. This may reflect my personal inclination towards optimism, something I share with other techies.

My confessions are as follows:

1. Neglecting context. Technological developments fail most frequently not because the science and technology do not pan out, but because the ‘systemic context’, that is the conjunction of social, market, organisational and other factors, is a poor fit with the technology. Thus, the potential remains nascent, awaiting a change in this context, which may not immediately arise.

… I have been wrong about ed-tech.

2. Underestimating the problem. Optimism can also lead to misunderstanding the science, or rather to a blithe disregard for the difficulty of the scientific and technical problems that must be overcome. A good scientific training, conditioned as it is by accounts of breakthrough success, cultivates a “how difficult could that be?” mentality. This encourages the bravery necessary to take on hard problems. Well, sometimes the problem turns out to be very difficult, and sometimes impossible, and sometimes even provably impossible.

… I have been wrong about progress in formal software development.

3. Forgetting the quick fix. It is easy, at least for me, to become fixated on a particular technology or approach. I was taught in my formative engineering training that ‘for every problem there is a solution that is elegant, appealing … and wholly wrong’. Quite often, the ‘right solution’ can be displaced by a seeming compromise – extending existing technologies – that gets things done for, at least, the time being. This can push a promising technology off the priority list. Why invest in solving technical problems when there is a working alternative available, one that might indeed be better in the short to medium, and possibly even the long-term?

… I have been wrong about decentralised data architectures.

4. Miscalculating the timeline. You may not be strictly wrong, but you may be sufficiently wrong that this makes no difference. There are a couple of well-known quips that bear on this cause of bad predictions. ‘X is the technology of the future … and always will be’ (X is often fusion) and Amara’s Law: ‘We tend to overestimate the effect of a technology in the short run and underestimate the effect in the long run.’ In the first case, timelines shift forward and, however many achievements accumulate and milestones are passed, the destination always seems to remain impossibly distant. Perhaps because of ‘underestimating the problem’ (see above). In the second case, the habit of overestimating the rapidity of maturation and adoption of technology can lead to an initial burst of enthusiasm followed by disillusion as a technology is seen as failing to fulfil its potential (a hype-curve effect). This disillusion can lead to underinvestment that stretches the timeline.

… I have been wrong about quantum computing.

5. Discounting technological convergence. Sometimes technologies in adjacent fields, often (but not always) in tools and processes, converge, delivering a step-change in the potential of a technology. Thus, for example, progress in a wide range of areas has been accelerated, and often disrupted, by increasing computational capabilities. What might be thought to be inherent limitations turn out to yield to bigger, faster computers. Similarly, problems that required ever more complex mechanistic models are accessible to machine learning. An overly narrow aperture on what might drive progress can yield bad predictions.

… I have been wrong about systems biology.

6. Identifying the wrong community. Progress on technology can depend a lot on who is driving that technology. Technologies can become ‘stuck’ within narrow groups or become ‘owned’ by particular specialisms that limit the extent to which they can be advanced. Whatever the inherent promise of a technology, it will not advance if the problem-solving community is insufficiently diverse.

… I have been wrong about knowledge representation.

I have it seems been wrong about a lot …