-

What are we missing?

31 January 2025

4 minute read

Will Hobbs considers how we can use emerging technology to enhance our expertise.

“Do you solemnly swear that you will tell the truth, the whole truth, and nothing but the truth?”

It is perhaps inevitable that we place too much weight on what we think we can see, touch and measure. The statistical framework we use to quantify the economy was forged amidst the steel and wheat economy of the 1930s when we needed to assess physical production, from tanks to butter. In any case, making stuff feels more economically substantial than providing services, a fact that populists the world over have truffled around in with great success of late.

Similarly, we implicitly prize the knowledge that we can codify and communicate over the ‘tacit’ knowledge that we can’t, which is often dismissed as woolly and of negligible epistemic worth. This week, we take a look at how a world resistant to measurement and codification links President Trump’s trade war, China’s breakthrough in generative AI and the noisy doomsayers assailing the UK’s economic outlook.

Tacit knowledge

When teaching children to learn how to ride a bike, we don’t tend to spend time drilling them on gyroscopic physics beforehand. Most of the learning is done on the fly. A gentle push and off they go, or not. We can think of the same in our professional lives. We know more than we can say, as Hungarian philosopher Michael Polanyi famously observed1. Much of our experience at work, our knowledge and individual expertise cannot be codified or communicated clearly.

This feature has protected many professions from technological disruption this last few decades – many tasks that were amenable have been swallowed up by the machines in the computer age. The question many are asking now is whether large language models are encroaching on areas once safeguarded by tacit knowledge, thereby threatening labour substitution. Large language models grow in sophistication daily and this week added further evidence of sharp cost declines ahead.

Tacit knowledge would no longer appear to be the preserve of humans. ‘Machine’ progress in predicting protein folding2 and other areas of the scientific frontier cannot be easily explained with theory. There is still much we do not understand about how large language models work (and don’t). Whether they represent a step towards AGI (artificial general intelligence) or an expensive off-ramp on the way is still to be determined.

What we can begin to see, however, is that this new technology is likely going to be helpful in both exploiting our existing stock of knowledge, as well as expanding it further. This current information revolution has the internet and social media playing roughly the role of the printing press in the 15th century equivalent.

Telling fact from fiction

The significance of this new revolution is not that we can now read on a screen what we could read in a book. It is that marginal access costs to codified knowledge of every kind have declined in ways that would have been unimaginable only a few decades ago3. However, as with the printing press example, one of the problems we are wrestling with surrounds the wider, non-monetary costs of accessing this knowledge mountain. Telling fact from fiction, the inherently elusive nature of truth and the usual range of bad actors are all hindering us.

One of Britain’s answers to this problem in the centuries following the messy arrival of the printing press came in the form of what some have called Knowledge Access Institutions4. These came in a range of formats, but perhaps are best embodied in the Royal Society’s motto embodied in its first charter of 1662 – Nullius in Verba (on no one’s word).

A flourishing culture of debate, disagreement and experimentation centred in various ‘societies’ cropping up like mushrooms across the country in the 17th and 18th centuries. In turn, this helped to harden important chunks of the knowledge mountain they were then wrestling with, creating more agreed, useful knowledge that could be exploited by the entrepreneurs of the time.

It may be that the contemporary equivalent is messily emerging. Paralysis in front of the torrents of misinformation and fake news is certainly a hindrance. However, that we are now creating a new ‘intelligence’ weaned on all this accumulated information may be forcing us to ask the right questions again. Perhaps again, we should be wary of ignoring our species’ capacity for error correction.

The invisible economy

The problems of telling up from down in amongst mountains of apparently useful data is of course also true of economics and investing too. These are professions that devote a lot of resource to spotting patterns of the past to predict the future. In the right hands, this can be powerful. However, awareness of the limitations of datasets is the first and most important step to avoiding the overconfident quackery often dominating the airwaves.

There is both a short- and long-term problem to face. We’ve spoken before about the struggles of statistical authorities since the pandemic. The survey responses on which much of our perspective on the economy relies have plunged in the UK and elsewhere. Data that already deserved very careful handling has become unplayable5 in many cases. There has not been a concomitant adjustment in how many interpret this information, with the UK employment data a good example.

The wider problem with historic datasets hinges on insufficient history for meaningful inference. A century may sound a long time in an economy’s life, but the variety in context, technological paradigm and more provides too much unobservable, unmeasurable variety under the surface. Meanwhile, the developed world has long since become dominated by the harder to measure services sector.

Conclusion

Don’t waste too much time and emotion minutely following the (normal) ebbs and flows of these economies, as described by data that is getting less, not more, fit for purpose. Do spend time wondering how you can use this emerging technology to supercharge your own expertise. Remember that technology at its most powerful is a complement. Finally, long-term investors can probably afford to tune out much of the dizzying array of White House announcements. The medium-term trajectory of the global economy will likely not notice much.