Comments by "Mark Pawelek" (@mark4asp) on "19 Common Fallacies, Explained." video.

  1.  @bohanxu6125  I have no disagreement regarding those 3 methods of reasoning. But, everyone should also be taught basic scientific reasoning (how to experiment, how to hold a variable constant, how to infer a scientific law), as well as reasoning from empirical evidence (how to apply actually existing evidence). Add those as 4., and 5. Another idea I have is to take actual policy debates, and redo them - to take arguments which politicos actually used, in reality, to decide policy and reapply them to look for flaws and or improvements to their arguments. Nearly all university degree students should also study statistics too. Note: "simple proof based mathematic" reasoning (such as proof by induction?), is harder than you think. If mathematicians find it hard, imagine how hard it'll be for the rest of us?
    5
  2. A good list, but one important point not made is that fallacies are often meta-fallacies: - such as projection and deflection. Projection and deflection are often unconscious, or spontaneous. * Projection: one projects when one "reasons" by imagining what other person thinks, recounting a precis of it to them, then condemning the ideas in the precis. So projection can include many fallacies (ad hominem, strawman, whataboutism(s), ..., because we find it almost impossible to precis what another person actually thinks (as opposed to what we imagine they think) * Deflection: talk about something else, it avoids facing the actual topic under debate. Can often be done by recounting an example, or story, or evidence which is either tangentially relevant or irrelevant. - and multiple-fallacies - which may even be overdetermined. In such a case one identifies the prime fallacy, but, one may even fall for a hidden fallacy entwined with it! Also - bad evidence. Common examples of bad evidence are: * bad statistics. For example weak statistics which may have been compiled using one or many of: cherry-picked data, bad sampling, too few data points, weak randomization, weak correlation, obscure of errror-prone maths such as fourier analysis, principle component analysis, or machine learning applied to 'dirty', or 'noisy' data. * bad modelling. Bad models can have unrealistic assumptions, simplistic, irrelevant, logic (such as game theory algorithms), unrealistic causal chains, inappropriate science, hidden maths: embedded within - such that the argument being presented, or supported, actually obscures itself AND is wrong! Models are never evidence. They are tools for speculation. If you take a lesson from this talk, I think it should be to practice steelmannning, and to argue empirically (from the evidence), not from logic. Learn to walk before you run. An empiricist, such as myself, probably thinks every argument made purely from logic is either a fallacy fallacy, or castle made of sand, or some other self-befuddlement!
    1