The idea “I don’t have to believe if it’s not absolutely certain” wasn’t invented to protect beliefs. It’s the natural default for humans. The idea that propositions /have/ probabilities was, AFAIK, invented by Plato, who immediately rejected it, saying that philosophers must deal only in claims that are 100%, absolutely, positively known to be true (by introspection or a memory from a past life, or by dialectic based on such truths). The skeptics, who took over Plato’s academy in the 3rd century BC, re-introduced the idea of probabilistic knowledge. They were then silenced and slandered for the next 2000 years as having said “you can’t know anything”—which they did literally say, but they made it clear that was because the word for “know” literally meant “believe with 100% confidence”, and still does to philosophers. Other philosophers /could not grasp the notion of uncertain belief/, and so thought the skeptics were just saying you can’t have beliefs.
The belief that knowledge must be 100% certain is at the core of rationalism, the Greek philosophical tradition which rejects the evidence of the senses as unreliable, and which dominated the West from the time of Constantine until the 19th century. That’s why David Hume upset people when he said you couldn’t be sure the sun would rise tomorrow. In fact you can’t, not 100% sure. But we can now pin approximate numbers to the the probabilities of sun-won’t-rise scenarios, and they all look to be so small that any “rational” thinker will ignore them. That idea—that a thing can be possible, but with such a small probability that it can be ignored—was not available in the 18th century.
The “rationalist” movement has been burdened from the start with the legacy of rationalism, which is actually the source of much Western “dark” epistemology. The apogees of rationalism were Plato, the medieval scholastics, and continental philosophy (which is the philosophy forced on Europe in the 1930s by the Nazis, because a core doctrine of rationalism is that empirical observation is trumped by introspection and dialectic, and the Nazis needed to delegitimize empiricism; see for instance The Murder of Professor Schlick: The Rise and Fall of the Vienna Circle). Rationalism really is subject to the Christian criticism that it needs foundational assumptions. Empiricism provides, among other things, a way to ground concepts in sense data, thus doing away with the need for foundational assumptions. How this might work for the English language has only very recently been demonstrated, not by linguists or philosophers, but by neural networks.
Many bad rationalist beliefs have slipped into the rationalist movement, and aren’t called out because rational thought is a necessary beginning point for science, quite useful in its place; and also because rationalists have claimed much unearned credit for the accomplishments of empirical science. Greek rationalism is the blessing and curse of the West. It accelerated the West’s progress in understanding the world, but also made the West uniquely murderous due to everybody in the West being 100% certain they were right.
The West pushed rationalism about as far as it could go, past the point where rationalism by itself fails. The German philosopher Heidegger provided a rational basis for Nazism by taking advantage of the fact that rationalism is ultimately based on unprovable assumptions, locating all those assumptions in places that could not be visually inspected, like the “authentic self” and the essential nature of the German Volk. The post-modernists, themselves all starting from a position of strict rationalism, then blamed Auschwitz on the Enlightenment, because they thought “Enlightenment” and “rational” were synonyms, and declared thought itself had failed when really it was just rationalism that had failed them.
Failure to be aware of the ways rationalism needs to be supplemented with statistics and heuristics is probably responsible for the breakdown of the rationalist movement into the post-rationalist movement.
One bad rationalist practice in the rationalist movement was the initial focus on symbolic AI, and the work on formal proofs in logics where terms and predicates bottom out in atomic symbols which are not grounded to sense data. Symbolic AI assumes that a concept (a predicate or name) can be an atomic “symbol” (which actually means what is now called a “sign” in linguistics)--a static, unchanging, context-free descriptor—that the word “dog” denotes an Eternal Form such as Plato believed in. You don’t get to look inside the atom (which IIRC means “indivisible”); just the word “dog” is sufficient; it always means exactly the same thing, so you can plug into propositions and do deduction in an algebraic manner. (The word for this is “logocentrism”.) This is necessary if you want to convert English sentences to logical form, reason with the resulting logical propositions, and then conclude that the results are True with a capital T. And it is horribly wrong, and is the main reason symbolic AI never worked and never will. Assigning probabilities to propositions is a start; but you need to track the probability distributions as well, and also the distribution, in some continuous space (such as the weights of one layer of a deep network), of the interpretation of each term and predicate.
The Aumann agreement theorem is another relic of rationalism. Aumann formalized and proved a belief which has been popular at least since Socrates, reiterated by Leibniz, and popularized by Rousseau: that all good, intelligent people must ultimately agree. (And Rousseau added the unspoken lemma: those who don’t agree, must be “forced to be free”.) It’s necessary for all grand Platonic schemes, like Marxism, to believe that honest and correctly-reasoning people can’t disagree, and therefore anyone disagreeing with you is dishonest (and should be eliminated). In real life, the agreement theory never applies, because no 2 people agree 100% on the meanings of their words, and their meanings in specific contexts, and their moral evaluations—and the theorem requires such exact, rationalist agreement. I’ve too often seen it referenced in rationalist circles to blame disagreements on bad logic rather than on diverse experiences and values.
The idea “I don’t have to believe if it’s not absolutely certain” wasn’t invented to protect beliefs. It’s the natural default for humans. The idea that propositions /have/ probabilities was, AFAIK, invented by Plato, who immediately rejected it, saying that philosophers must deal only in claims that are 100%, absolutely, positively known to be true (by introspection or a memory from a past life, or by dialectic based on such truths). The skeptics, who took over Plato’s academy in the 3rd century BC, re-introduced the idea of probabilistic knowledge. They were then silenced and slandered for the next 2000 years as having said “you can’t know anything”—which they did literally say, but they made it clear that was because the word for “know” literally meant “believe with 100% confidence”, and still does to philosophers. Other philosophers /could not grasp the notion of uncertain belief/, and so thought the skeptics were just saying you can’t have beliefs.
The belief that knowledge must be 100% certain is at the core of rationalism, the Greek philosophical tradition which rejects the evidence of the senses as unreliable, and which dominated the West from the time of Constantine until the 19th century. That’s why David Hume upset people when he said you couldn’t be sure the sun would rise tomorrow. In fact you can’t, not 100% sure. But we can now pin approximate numbers to the the probabilities of sun-won’t-rise scenarios, and they all look to be so small that any “rational” thinker will ignore them. That idea—that a thing can be possible, but with such a small probability that it can be ignored—was not available in the 18th century.
The “rationalist” movement has been burdened from the start with the legacy of rationalism, which is actually the source of much Western “dark” epistemology. The apogees of rationalism were Plato, the medieval scholastics, and continental philosophy (which is the philosophy forced on Europe in the 1930s by the Nazis, because a core doctrine of rationalism is that empirical observation is trumped by introspection and dialectic, and the Nazis needed to delegitimize empiricism; see for instance The Murder of Professor Schlick: The Rise and Fall of the Vienna Circle). Rationalism really is subject to the Christian criticism that it needs foundational assumptions. Empiricism provides, among other things, a way to ground concepts in sense data, thus doing away with the need for foundational assumptions. How this might work for the English language has only very recently been demonstrated, not by linguists or philosophers, but by neural networks.
Many bad rationalist beliefs have slipped into the rationalist movement, and aren’t called out because rational thought is a necessary beginning point for science, quite useful in its place; and also because rationalists have claimed much unearned credit for the accomplishments of empirical science. Greek rationalism is the blessing and curse of the West. It accelerated the West’s progress in understanding the world, but also made the West uniquely murderous due to everybody in the West being 100% certain they were right.
The West pushed rationalism about as far as it could go, past the point where rationalism by itself fails. The German philosopher Heidegger provided a rational basis for Nazism by taking advantage of the fact that rationalism is ultimately based on unprovable assumptions, locating all those assumptions in places that could not be visually inspected, like the “authentic self” and the essential nature of the German Volk. The post-modernists, themselves all starting from a position of strict rationalism, then blamed Auschwitz on the Enlightenment, because they thought “Enlightenment” and “rational” were synonyms, and declared thought itself had failed when really it was just rationalism that had failed them.
Failure to be aware of the ways rationalism needs to be supplemented with statistics and heuristics is probably responsible for the breakdown of the rationalist movement into the post-rationalist movement.
One bad rationalist practice in the rationalist movement was the initial focus on symbolic AI, and the work on formal proofs in logics where terms and predicates bottom out in atomic symbols which are not grounded to sense data. Symbolic AI assumes that a concept (a predicate or name) can be an atomic “symbol” (which actually means what is now called a “sign” in linguistics)--a static, unchanging, context-free descriptor—that the word “dog” denotes an Eternal Form such as Plato believed in. You don’t get to look inside the atom (which IIRC means “indivisible”); just the word “dog” is sufficient; it always means exactly the same thing, so you can plug into propositions and do deduction in an algebraic manner. (The word for this is “logocentrism”.) This is necessary if you want to convert English sentences to logical form, reason with the resulting logical propositions, and then conclude that the results are True with a capital T. And it is horribly wrong, and is the main reason symbolic AI never worked and never will. Assigning probabilities to propositions is a start; but you need to track the probability distributions as well, and also the distribution, in some continuous space (such as the weights of one layer of a deep network), of the interpretation of each term and predicate.
The Aumann agreement theorem is another relic of rationalism. Aumann formalized and proved a belief which has been popular at least since Socrates, reiterated by Leibniz, and popularized by Rousseau: that all good, intelligent people must ultimately agree. (And Rousseau added the unspoken lemma: those who don’t agree, must be “forced to be free”.) It’s necessary for all grand Platonic schemes, like Marxism, to believe that honest and correctly-reasoning people can’t disagree, and therefore anyone disagreeing with you is dishonest (and should be eliminated). In real life, the agreement theory never applies, because no 2 people agree 100% on the meanings of their words, and their meanings in specific contexts, and their moral evaluations—and the theorem requires such exact, rationalist agreement. I’ve too often seen it referenced in rationalist circles to blame disagreements on bad logic rather than on diverse experiences and values.