Radical Uncertainty

What Hacker News says

1. Barrin92Feb 2025
   >As such, empirical laws that were trustworthy in the past, may become significantly less accurate, and can in fact become dangerous to rely upon too blindly.

   Read a really good book on this a few years ago by John Kay and Mervyn King, Radical Uncertainty (https://wwnorton.com/books/9781324004776), where they talk about the issue of applying the logic of "known uncertain" systems (like games of poker, or volcanic eruptions) to completely dynamic "radically uncertain" systems where any attempt to quantify outcomes is effectively completely made up. Election forecasts being one example, where there's no guarantee that you know the rules of the game, if they stay constant between elections and so on.

   I think this is very timely because these days you see this kind of cargo cult empiricism where people will say "I am 30% confident that..." as if just talking in a quantitative language means you've somehow understood something about the future or have reason to be more confident.

   Show morethread
2. Barrin92Mar 2022
   John Kay wrote a good book about this called Radical Uncertainty. It's annoying that there is an increasing tendency to frame what is a purely narrative statement in quantitative terms.

   As the author points out war is not historically predetermined, at least in any way we could convincingly figure out, and pulling numbers out of thin air only serves the purpose of trying to talk about the future with less uncertainty than is appropriate.

   Talking about events in terms of probabilities should be reserved for events that are not dominated by 'unknown unknowns'. You can attach a probability to a volcano outbreak, a transaction in an ordinary functioning market, a game of blackjack. That is to say 'small worlds' where outcomes are known, rules well defined and inputs quantified. You can't predict the overthrow a government in 5 years in numerical terms. That doesn't mean you can't talk about it. It's okay to say something is unlikely or very likely, but it makes more sense to explicitly talk about these things in qualitative terms rather than trying to be 'mathy' to make a prediction appear more objective than it is.

   Show morethread
3. Barrin92Jan 2022
   John Kay wrote a good book about this (Radical Uncertainty). There is a trend to falsely quantify questions that take place either under conditions of uncertainty or where numbers are just axiomatic assumptions rather than based on regular observations.

   It's very much everywhere now, not just in abstract philosophy like this but even in a lot of real world scenarios like the pandemic. Instead of giving a qualitative narrative and leaving space for human intuition people try to present arguments as more certain than they are by slapping some numbers / models onto them.

   Show morethread
4. asplakeMar 2026
   I had the privilege last night of attending a lecture given by Prof Sir John Kay (Obliquity, Radical Uncertainty, etc). He was scathing on two points: 1) the way the world changed in the 1970s from management as responsibility to leadership as prize, and 2) the abject failure of business schools to develop a serious body of knowledge. Taken together, business schools have become cash cows for universities while still being held in disdain by academia. This from the first dean of Oxford's Said Business School.
   Show morethread
5. k310May 2024
   The author (Brian Klaas) references "Radical Uncertainty" by Kay and King.

     In their book, Radical Uncertainty, Mervyn King and John Kay draw a distinction between two kinds of problems: puzzles and mysteries.
   

   Here is the description of their book:

       Much economic advice is bogus quantification, warn two leading experts in this essential book. Invented numbers offer false security; we need instead robust narratives that yield the confidence to manage uncertainty.
   
      Some uncertainties are resolvable. The insurance industry’s actuarial tables and the gambler’s roulette wheel both yield to the tools of probability theory. Most situations in life, however, involve a deeper kind of uncertainty, a radical uncertainty for which historical data provide no useful guidance to future outcomes. Radical uncertainty concerns events whose determinants are insufficiently understood for probabilities to be known or forecasting possible. Before President Barack Obama made the fateful decision to send in the Navy Seals, his advisers offered him wildly divergent estimates of the odds that Osama bin Laden would be in the Abbottabad compound. In 2000, no one―not least Steve Jobs―knew what a smartphone was; how could anyone have predicted how many would be sold in 2020? And financial advisers who confidently provide the information required in the standard retirement planning package―what will interest rates, the cost of living, and your state of health be in 2050?―demonstrate only that their advice is worthless.
   
       The limits of certainty demonstrate the power of human judgment over artificial intelligence. In most critical decisions there can be no forecasts or probability distributions on which we might sensibly rely. Instead of inventing numbers to fill the gaps in our knowledge, we should adopt business, political, and personal strategies that will be robust to alternative futures and resilient to unpredictable events. Within the security of such a robust and resilient reference narrative, uncertainty can be embraced, because it is the source of creativity, excitement, and profit.
   

   The author poses several "what if's" including the Jan 6 attack. Rather than the grim scenario they propose, I recall that a year or two later, the Capitol was blanketed by a giant snowstorm. Bad timing!
   Show morethread
6. Barrin92Jul 2021
   >100% it can be modelled somehow in theory.

   The entire premise of modelling something rests on the fact that the behaviour of what you're trying to model in the future depends on its behaviour in the past.

   This is not true for all systems, it's not even true for most systems. John Kay differentiates between 'resolvable uncertainty' and 'radical uncertainty'. Modelling a volcano eruption is resolvable because it's a reasonably predictable system and you can make confident stochastic claims. Complex human systems are chaotic, have 'unknown unknowns', are simply subject to chance or non-linear behaviour that completely throws any prediction off.

   You cannot model something that is subject to forces you cannot even know at the point of making the model, and this is true for any complex human system pretty much. Modelling Covid as if we have access to Hari Seldon's psycho-history was always a terrible idea.

   Show morethread