5 ms·

If we accept Popper's falsification, one of the characteristics that distinguishes real science from psuedo science is the making of falsifiable predictions. Tu

If we accept Popper's falsification, one of the characteristics that distinguishes real science from psuedo science is the making of falsifiable predictions. Turchin's Cliodynamics reminds me of Elliot Waves, and other technical (as opposed to fundamental) analytic techniques for financial market price prediction. Which begs the question: do Turchin's models make tradable predictions?

dredmorbius 6y ago
Popper's falsification is useful, but may not be a truly universal test. It seems most ineffective in the case of systems theories, of which complexity research (closely aligned with Turchin's work), systems theory, cybernetics, and evolutionary theories and models are several examples. Few of these produce reliable forward-looking predictions, most especially for anticipating specific states achieved at specific times. Much of this hinges on simple complexity and measurement error (uncertainty over initial conditions). What these models do provide, though, are useful models for drawing associations, predicting or describing specific patterns, and establishing the likely existence of specific features or elements of a system. Darwin predated concepts of genes, chromosomes, DNA, and mutation mechanism. But his notions of variation and inheritance presupposed these and outlined their general characteristics. Later discoveries conformed with, and further refined, these predictions. Where humanities and social sciences (Turchin's cliodynamics somewhat spans this divide) run into additional complexity is that models and predictions themselves become part of the informational context in which individuals, groups, and societies see, analyse, process, express, and respond to the informational environment. This is often compared to Heisenberg uncertainty, but this is a category error. Heisenberg is addressing fundamental knowability, where the act of ascertaining system state changes the state itself. Epistemic interactions affect not the state itself, but the informational loop associated with it. The model and derived understanding are part of the sensing, processing, learning, memory, communications, and control interactions --- the model operates within the modeled system, endogenously rather than exogenously. How to incorporate this distinction into philosophy of science is an interesting problem, though I don't think it is impossible. Pseudoscience lacks not only falsifiability, but generally, consistency and causality. Sound systems theories should be consistent and denote causality.
- nemo44x 6y ago
  To your point, I believe this why is we need to be careful when describing things like many fields of social science as a “science”. It’s more “social educated guesses” than real science. Social sciences demanding to be taken as concrete as rigid sciences is a problem today I think. We even see them dismissing actual science in some cases not through falsification but through using their own theory to “prove” certain science is “problematic” of heterodox therefore restricting knowledge.
  - zanellato19 6y ago
    They demand to be taken as seriously, not as concrete, in my experience. Which is fair, we can maybe reach something like what is described in this article eventually - but as Turchin himself put it, it would not exist without History. If we are to devalue social sciences, we are giving up on ever expanding that knowledge.
    - nemo44x 6y ago
      Should we take religions seriously then? They demand it anyhow. I do not mean to dismiss the entirety of the humanities or social sciences (or religion for that matter!) I do get concerned however with their motivations and the power they wield over public policy and politics in general. A certain claim to the truth they don’t really have. I also find it curious that most elite universities and scholars come to more or less the same conclusions regarding subjects that aren’t falsifiable. You’d expect this to be the case in the hard sciences (reproducibility is vital) but not in others. For instance, in religions we can see great divergence. Even in the way economic theory (another thing that isn’t science but often claims to be) has differing conclusions based on the same observations. I also see today in the application of some of these social sciences that asking to debate or to ask for more evidence is taken as a challenge to it and in fact evidence that the theory is right. See some national best selling books from this year, for example. At what point do social sciences look more like religion than “science”?
      - dredmorbius 6y ago
arbitrage 6y ago
In the article, it is briefly brought up. Turchin dodges the falsification issue by saying, to paraphrase, "I don't make predictions. That's somebody else's job." To me, right there, is when my suspicion that this was all huckster social-darwinism malthusian doomsday preaching solidified into certainty. Put up, or shut up. Or you are not mathematically rigorous, as Turchin claims to be.
- macintux 6y ago
  I’m curious what you’re looking at to create that paraphrase. He says that about how to fix the problem, but I don’t see him dodging predictions.
- verisimilidude 6y ago
  Back in 2010, he made prescient predictions about our current state of affairs. That’s why he’s getting attention.
blueyes 6y ago
He called 2020's civil unrest. A trader would have to extrapolate to place a bet on that.
js8 6y ago
The hypothesis "if I come too close to this ledge, I will fall into the precipice" is falsifiable. The question is, do we really want to test (falsify) it? It seems to me when it comes to threat of civilization(al) collapse, we will have to accept incomplete evidence.
- dredmorbius 6y ago
  The principle strength of history is in providing a set of extant examples. There's the risk of overfitting, but models can be tested. Present-day models are at least exogenous to the past, if not its present interpretation.
psychoslave 6y ago
Here is your epistemic "eat your own reflexivity" of the day: Is Popper’s falsification falsifiable?

dredmorbius 6y ago

Popper's falsification is useful, but may not be a truly universal test. It seems most ineffective in the case of systems theories, of which complexity research (closely aligned with Turchin's work), systems theory, cybernetics, and evolutionary theories and models are several examples. Few of these produce reliable forward-looking predictions, most especially for anticipating specific states achieved at specific times. Much of this hinges on simple complexity and measurement error (uncertainty over initial conditions). What these models do provide, though, are useful models for drawing associations, predicting or describing specific patterns, and establishing the likely existence of specific features or elements of a system. Darwin predated concepts of genes, chromosomes, DNA, and mutation mechanism. But his notions of variation and inheritance presupposed these and outlined their general characteristics. Later discoveries conformed with, and further refined, these predictions. Where humanities and social sciences (Turchin's cliodynamics somewhat spans this divide) run into additional complexity is that models and predictions themselves become part of the informational context in which individuals, groups, and societies see, analyse, process, express, and respond to the informational environment. This is often compared to Heisenberg uncertainty, but this is a category error. Heisenberg is addressing fundamental knowability, where the act of ascertaining system state changes the state itself. Epistemic interactions affect not the state itself, but the informational loop associated with it. The model and derived understanding are part of the sensing, processing, learning, memory, communications, and control interactions --- the model operates within the modeled system, endogenously rather than exogenously. How to incorporate this distinction into philosophy of science is an interesting problem, though I don't think it is impossible. Pseudoscience lacks not only falsifiability, but generally, consistency and causality. Sound systems theories should be consistent and denote causality.

nemo44x 6y ago
To your point, I believe this why is we need to be careful when describing things like many fields of social science as a “science”. It’s more “social educated guesses” than real science. Social sciences demanding to be taken as concrete as rigid sciences is a problem today I think. We even see them dismissing actual science in some cases not through falsification but through using their own theory to “prove” certain science is “problematic” of heterodox therefore restricting knowledge.
- zanellato19 6y ago
  They demand to be taken as seriously, not as concrete, in my experience. Which is fair, we can maybe reach something like what is described in this article eventually - but as Turchin himself put it, it would not exist without History. If we are to devalue social sciences, we are giving up on ever expanding that knowledge.
  - nemo44x 6y ago
    Should we take religions seriously then? They demand it anyhow. I do not mean to dismiss the entirety of the humanities or social sciences (or religion for that matter!) I do get concerned however with their motivations and the power they wield over public policy and politics in general. A certain claim to the truth they don’t really have. I also find it curious that most elite universities and scholars come to more or less the same conclusions regarding subjects that aren’t falsifiable. You’d expect this to be the case in the hard sciences (reproducibility is vital) but not in others. For instance, in religions we can see great divergence. Even in the way economic theory (another thing that isn’t science but often claims to be) has differing conclusions based on the same observations. I also see today in the application of some of these social sciences that asking to debate or to ask for more evidence is taken as a challenge to it and in fact evidence that the theory is right. See some national best selling books from this year, for example. At what point do social sciences look more like religion than “science”?
    - dredmorbius 6y ago

arbitrage 6y ago

In the article, it is briefly brought up. Turchin dodges the falsification issue by saying, to paraphrase, "I don't make predictions. That's somebody else's job." To me, right there, is when my suspicion that this was all huckster social-darwinism malthusian doomsday preaching solidified into certainty. Put up, or shut up. Or you are not mathematically rigorous, as Turchin claims to be.

macintux 6y ago
I’m curious what you’re looking at to create that paraphrase. He says that about how to fix the problem, but I don’t see him dodging predictions.
verisimilidude 6y ago
Back in 2010, he made prescient predictions about our current state of affairs. That’s why he’s getting attention.

blueyes 6y ago

He called 2020's civil unrest. A trader would have to extrapolate to place a bet on that.

js8 6y ago

The hypothesis "if I come too close to this ledge, I will fall into the precipice" is falsifiable. The question is, do we really want to test (falsify) it? It seems to me when it comes to threat of civilization(al) collapse, we will have to accept incomplete evidence.

dredmorbius 6y ago
The principle strength of history is in providing a set of extant examples. There's the risk of overfitting, but models can be tested. Present-day models are at least exogenous to the past, if not its present interpretation.

psychoslave 6y ago

Here is your epistemic "eat your own reflexivity" of the day: Is Popper’s falsification falsifiable?