The full question is:

  • Is human civilization doomed to repeat cycles of cataclysms because, on average, humans cannot make rational decisions regarding complex subjects for which they have incomplete understanding?

What I'm getting at is there is a persistent trend in human politics to legislate based on feeling and belief, as opposed to the least biased interpretation of the the most robust data sets available.

From a strictly game theoretic standpoint, this is a guaranteed losing strategy on aggregate.

Yet even with the mathematics of optimal decision making well defined, there still seems to be widespread abuse of statistics to support pre-existing biases, beliefs and initiatives. (I won't get specific on issues, but many present problems are the results of previous abuse of statistics, such as projections of growth to fill budget gaps based on wishful thinking.)

  • 1
    i think i get the question. btw i've seen "dialectics" talked about in this way, as thinking about feeling (only)
    – user25714
    Jun 27, 2017 at 19:18
  • @idiotan haha, I'm a Socratic, so I value dialectic as a very effective learning method, but I take your meaning. I'm not sure we've really progressed past Socrates in terms of the average person, or even most people, because I rarely encounter people who question their own assumptions.
    – DukeZhou
    Jun 27, 2017 at 20:07
  • 1
    apologies for the gloss, i saw it said in the sacred wood, a quite wonderful book on modern poetry. i'm an upvote cos i find the question intriguing
    – user25714
    Jun 27, 2017 at 20:11
  • This seems to be about history, politics, game theory, futurology, anything but philosophy. The philosophical question boils down to "humans cannot make rational decisions regarding complex subjects, am I right?", and even that is rhetorical.
    – Conifold
    Jun 27, 2017 at 23:34
  • @Conifold Now that you put it that way, it is a potentially suspect question. In some sense I'm meaning to ask about is the philosophy of governance, the "technocratic" vs. "populist" approaches, but there's also the issue of core values. As usual, I'm trying to get at the connection between ethics and mathematics, specifically in regard to decision-making. Thus, I wouldn't say "all humans". If decision makers are trained, approach the problem with the best known methods, and remove bias to the greatest extent possible, the result will be more optimal decisions.
    – DukeZhou
    Jun 28, 2017 at 0:01

1 Answer 1


Your presumption that legislating on the basis of the "least biased interpretation of the the most robust data sets available" will avoid, or even minimise, the cataclysmic results of such decisions is itself flawed.

Firstly, all the information we have, even if interpreted without any bias at all, is only a subset of all the information there is to have. We do not know how large a proportion. It is reasonable to believe that all the information we do not have outweighs all the information we have and so making decisions based on the information we have is no guarantee that those decisions will be any more right than random, in the long term.

Secondly, whilst you could theoretically (if not practically) eliminate bias in the interpretation of this data, you could not undo the biases in the investigation of the data in the first place. The things scientists have chosen to investigate are the result of their biases, both personal and societal. As a consequence this sub-set of all the data there is that we currently know will not only be definitely incomplete, but will also be a biased selection of all the information there is to know.

Thirdly, no rationality could provide us with the values we assign to competing objectives and this valuing activity is essential to the making of any decision. An unbiased interpretation of data might provide us with with an accurate assessment of the efficacy of a range of strategies where they compete for the same objective, but it would never provide us with any information about which objective to attempt to achieve where two compete. A simple and classic example is the degree of hyperbolic discounting one should apply to the value of happiness now against happiness in the future where the two compete.

Fourthly, legislation is inevitably the result of the human population it legislates, in democratic societies, this is a direct and peaceful relationship at the polls, but even in autocratic societies this relationship still persists albeit a staccato of violent uprisings. No amount of legislation can make a mass of people behave in a way they do not wish to behave for very long (on an historical timescale). There will eventually be a rebellion against any attempt to impose such decisions.

Finally, the human world is incredibly complex because we have so many potential interactions, complex systems tend towards chaotic patterns where small errors can have staggeringly large consequences. Attempting too much social or economic engineering on the basis of the science we currently have would be akin to attempting to improve the entire technology of the mission to Mars on the basis that we know what a diode does.

  • Nice perspective. I do think there are a specific subset of problems that can be quantified and areas where we know algorithms perform better than humans. But it's wiggly, because until recently, in the game of Go, human intuition trumped computational power. That problem has a very narrow, completely defined set of parameters, and even then the threshold is subjective. That said, more optimal decisions are always better than less optimal decisions, and the worst approach to decision making is to manipulate data to confirm biases.
    – DukeZhou
    Jun 28, 2017 at 13:32

Not the answer you're looking for? Browse other questions tagged .