The Paradox of Karl Popper
blogs.scientificamerican.com
blogs.scientificamerican.com
Intellectual humility is not the same quality as personal meekness and geniuses have always had difficult personalities. Philosophers doubly so. How could it be otherwise? How can one create new stuff if one is too agreeable and too susceptible to groupthink?
Btw, the reason falsifiability doesn't have to be falsifiable (by experiment) is that it isn't a scientific idea; it's a philosophical one. It is potentially 'falsifiable' by criticism, however.
The best modern Popper proponent is David Deutsch. See his books The Fabric of Reality and The Beginning of Infinity. They're both packed with interesting ideas and arguments. He says that Popper did indeed solve the problem of induction. And here's a recent paper of his on the logic of experimental testing: https://arxiv.org/abs/1508.02048
Agreed. I think the main issue is not with the idea itself of the scientific method, rather than the practical implications and its limitations.
- The main idea is a qualitative one. It doesn't consider all the statistical can of worms about reproducibility.
- Real world experiments rarely reproduce perfectly/exactly. The more the experiment involves more variables (or unknown variables) you can detect an effect but its magnitude might vary.
- Not everything can be reproduced, not everything can be measured perfectly, hence the "study of knowledge" needs to be able to handle those cases, and not just bury their head in the sand and go "la la la this is not scientific let's ignore it"
They were also doing that before Popper was a thing.
> Whatever "statistical can of worms" there is regarding it sounds like a problem with the statistics being used rather than science.
Not really. It's easier to prove "Drug X cures dying from disease Y 99% of the time" than "Drug X solves issue Y 30% of the time only in individuals with a certain combination of characteristics".
They were also doing that before Popper was a thing.
Yes, but not sure why you are pointing that out.
> Whatever "statistical can of worms" there is regarding it sounds like a problem with the statistics being used rather than science.
Not really. It's easier to prove "Drug X cures dying from disease Y 99% of the time" than "Drug X solves issue Y 30% of the time only in individuals with a certain combination of characteristics".
If you had a complete understanding you would only apply the drug when it would work and have 100% success rate.
There is no way to have 100% understanding of a specific person to have a designer drug for every condition.
Not even Aspirin is 100% effective nor we have a complete understanding of how it works (though there were some discoveries on that regard around 20yrs ago, I think)
It's hard for physics and it's impossible for biology (and other fields) to come up with completely accurate predictions.
Popper never said ideas needed to be based on reproducible evidence. He also emphasised explanations over just predictability.
A bit of a stretch, but I'm also skeptical of the ability of the scientific method to be able to understand/prove anything with (hidden) state/memory.
For a lot of 'skeptics' unproven means wrong and unmeasurable means wrong.
https://www.nybooks.com/articles/1983/04/28/the-karl-popper-...
> The reasonable man adapts himself to the world: the unreasonable man persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man.
-- George Bernard Shaw
Not sure that's true.
But we nearly all of us underestimate how much creativity and energy is expended in trying to be normal (or to appear normal, which is much the same thing).
The author seemed to not understand Popper's response, which was dead on of course. The type of thing that can be falsified is a scientific theory that makes empirical predictions. Falsification is a prescription for how to do science. It can be argued against from a philosophical standpoint, but it makes no predictions and cannot be proven wrong by observation. It's like asking if the 4th Commandment can be falsified -- the compiler gives a type error.
Again, enjoyable read, but this aspect was disappointing.
Then the author seems pathetic here: "Peering searchingly into my eyes, he asked if one of his critics had persuaded me to pose the question. Yes, I lied." Why lie? Have some guts, like Popper evidently does, and believe in what you want to ask.
Popper at this time was 90 years old, surely not putting his best foot forward. It was a poor essay overall, IMO, but I love to read anything about Popper.
If you haven't read the article yet, my advice is not to read it.
The author seems to have pulled the bulk of the article from 25 year old notes. He doesn't hide his own youthful naïveté at assuming that a taxi driver would know directions to the home of a philosopher. He also preserves what I'd imagine are embarrassing moments like Popper's reaction to his "big question". What you consider silly, might just be an issue of narrative style.
I personally think it captures the wit and home life of an energetic 90 year old philosopher. What well adjusted human being doesn't expect to have a slightly disorganization existence near the end of a very long life?
Huh? The author specifically asked for directions and was surprised when he was (unjustifiably) assured he wouldn't need them.
> I personally think it captures the wit and home life of an energetic 90 year old philosopher...
13415 isn't saying the author succeeds at making Popper look silly. Rather, the author is attempting and failing to do so.
> 13415 isn't saying the author succeeds at making Popper look silly. Rather, the author is attempting and failing to do so.
How would you define failure in this scenario?
You sure do like the word naive!
Do you think he lied about being instructed by the staff member that the taxi driver would know where the guy lived? If not, then you're arguing that the author, having been instructed by the staff that he would not need directions, should have summoned his worldly wisdom and realized this couldn't be true. But the mere bias of someone's staff, which is of course predictable, does not explain why that staff would exaggerate in a way that will predictably be refuted, i.e., when the author gets into the cab, he'll find out they don't know! So you're arguing that he's naive for not assuming that the staff is both biased and stupid. But cab drivers knowing where a local minor celebrity lives is not that crazy of a proposition.
> How would you define failure in this scenario?
Rather than reporting unambiguous (i.e., un-manipulatable) facts that are clear evidence of silliness, he mostly uses rhetorical techniques which are completely up to the author's discretion. When he does report facts, they are completely consistent with Popper being a retired intellectual giant who is just old, forgetful, and busy. Trying to make Popper look bad because he's a forgetful 90-year-old comes off as very distasteful.
a) his main mission seemed to be to show people they didn't know as much as they thought
b) he was basically a huge troll, even if he was usually correct
c) he was such a huge troll that the Athenians actually put him to death for it via a public trial
If our culture didn't have certain moral qualms that the Athenians didn't have too much trouble brushing aside, I imagine Karl's career would have been cut a lot shorter, along with quite a few other philosophers.
Yep, pretty much invented the falsification principle. Imagine a time when scientists often tried to prove things true. That was science before Popper. This makes him the greatest contributor to science that I know of since Francis Bacon.
The philosophers of science that Popper was criticizing didn't think that scientists should try to prove that theories were true. They thought that scientists should try to show that theories were likely to be true given the available evidence. Popper was a radical skeptic regarding epistemic probability, and thought that it was impossible to probabilistically confirm scientific theories.
Apologies to anyone I might have offended by accidentally bundling them with relativists.
I see your point, and I agree, but lest we address the wrong concept, the term you're looking for is 'dictatorial'. Capitalism, just like Communism, produced bloody repressive regimes that were no ground to 'open societies' (e.g. almost every South American country from the 60's to 80's). History teaches us that we should keep our eyes open, for living in capitalist and oh-so-free countries today is simply no guarantee of 'open societies' tomorrow.
"The Most Intolerant Wins: The Dictatorship of the Small Minority" [0]
[0] https://medium.com/incerto/the-most-intolerant-wins-the-dict...
"...Clearly can democracy –by definition the majority — tolerate enemies? The question is as follows: “ Would you agree to deny the freedom of speech to every political party that has in its charter the banning the freedom of speech?” Let’s go one step further, “Should a society that has elected to be tolerant be intolerant about intolerance?”
His theories of science (and in particular, of falsification) are explained in his book "Conjectures and Refutations". I read it years ago when I was in grad school.
Popper's writing is clear and jargon-free, so anyone who is interested in his ideas should take a look at some of the original sources.
Finally, "Wittgenstein's Poker" by David Edmonds and John Eidinow takes an argument between Wittgenstein and Popper as a starting point for exploring their lives and work.
It seems the most detailed and thorough paragraph is the one devoted to the spelling of 'Popperazzi'.
Popper glared at me. Then his expression softened, and he placed his hand on mine. “I don't want to hurt you,” he said gently, “but it is a silly question."
Very sly form of argument, that.
Say I have coin that I predict will always flip heads. The first flip is tails. Hypothesis falsified! But, that still tells me nothing about the future if the coin is fair.
For falsification to work, we have to assume a priori there is not a uniform distribution over events. In which case, we can do something like Solmonoff induction anyways.
So, it looks like falsification is only useful if induction is also useful. In which case, falsification is not a workaround for induction's problems.
Induction, BTW, is completely invalid as a mode of reasoning (speaking here of logical induction, not mathematical induction). You can flip a coin 10,000 times and have it come up heads every time and still not be able to validly conclude that it will come up heads every time (there are a lot of ancillary assumptions that go into the deductive inference, none of which are justified). On the other hand, you can examine the coin, see that there are heads on both sides, and immediately conclude (soundly) that it will come up heads every time because you can provide an explanation: one of the two sides must come up, and they are both heads!
Adding an explanation is proposing a model for the coin's behavior. Regardless of the model proposed, and whether it is confirmed or refuted, if the coin is fair you learn nothing.
On the other hand, if the coin is not fair, Solomonoff induction works.
But that is simply because the theory that a coin is fair cannot be falsified with one data point. (Well, that's actually not quite true. It can't be falsified with a single data point which is "heads" or "tails". It could be falsified if the coin lands and breaks into two pieces revealing some internal mechanism that makes it unfair.)
> if the coin is not fair, Solomonoff induction works.
No, it doesn't. It is possible (given sufficient technology) to produce a coin that generates a sequence of flips that is indistinguishable from fair for any finite number N of flips, but which then produces HEADS forever after the Nth flip. Moreover, it is possible to know that this will happen without even flipping the coin!
The basic problem of distinguishing between patterns and luck undermines both falsification and induction.
As far as I know, there is no resolution to the problem. We just irrationally assume there is no problem out of habit, as Hume states in his original work. A leap of faith, if you will.
(So for example, conspiracy theories fail as explanatory theories not because they are inconsistent with the data -- they generally aren't -- but rather because they are not concise and brittle.)
Because they require less effort to work with. Note that concise is only better all else being equal. Being consistent with the observed data is necessary, but not sufficient. We can only ever have a finite amount of data, and any finite dataset is consistent with an infinite number of theories. Conciseness is just a heuristic we use to select among the infinite possibilities. It just turns out that in the universe we live in, concise theories consistent with the data have a lot of predictive power. No one know why our universe is this way. That's just how it turns out to be.
How is this not inductivism?
OK, on that view, yes, they are the same. But inductivism is generally understood to be more specific than that. It is generally understood to be a particular way of deriving general rules from specific observations, namely, by simply generalizing the data into patterns without any regard for causal mechanisms. So, for example, if you see (say) 100 crows and they are all black, induction tells you to conclude that all crows are black without ever asking why crows are black.
The basic problem remains, though we've consolidated many problems of induction into one.
No, that's not true. The operative word is "explanation", not "short". Explanation changes everything. It is not enough to say, "The sun has risen every day in the past, therefore it will continue to rise in the future." You have to explain why the sun has risen every day in the past (i.e. the earth is round, it rotates about its axis, yada yada yada). Conciseness is not as much a requirement as it is a consequence of the more primary requirement of being a good explanation and consistent with all available data. In actual practice, these requirements generally eliminate all theories but one. It just turns out that those theories tend to be concise.
No. That's inverting the order of the interaction. On the first case you are observing a phenomenon, and using it to construct a model; on the second case you are constructing a model, and then using your observation to check it.
Induction is a great way to create useful models, but only failures on falsification can give you confidence that the model actually works.
No, it isn't. Induction is a great way to create WRONG models. In fact, all of the interesting and useful stuff happens exactly where induction gets it wrong.
Example: the sun has risen on earth every day for the last 10^11 days or so. But we know that in another 10^11 days the sun will stop rising (because it will become a red giant and engulf the earth). We know this despite the fact that we have never actually observed this happen even once.
Probability is nearly always heuristic technically due to lacking perfect knowledge. Quantum effects appear to involve some true randomness.
"Dynamical bias in the coin toss". https://statweb.stanford.edu/~susan/papers/headswithJ.pdf
* Or at least every time the experiment is performed.
Your cycle of scientific inquiry is complete in this case. The test demonstrated that the coin will not always flip heads, and therefore the hypothesis was disproved.
But, that still tells me nothing about the future if the coin is fair.
This is now a new, albeit related, inquiry that can build upon the results of the first test.
So your proposition is constructed in error unfortunately and falsification doesn't claim to address the problem of induction.
Karl Popper, a philosopher of science, sought to solve the problem of induction. He argued that science does not use induction, and induction is in fact a myth. Instead, knowledge is created by conjecture and criticism. The main role of observations and experiments in science, he argued, is in attempts to criticize and refute existing theories.
Kind of a bold take from SciAm.
Falsification is critical to 'real' science, as otherwise the value of what is said rests solely on how it is said rather than what is said. For instance I could say that there's an invisible massless gremlin resting on your shoulder, but you can't tell it's there as it only interacts with other invisible massless gremlins and gravity. Obviously nobody would believe this, because it sounds absurd. But that's not why you shouldn't believe it. The reason you shouldn't believe it is believe it is because it's not falsifiable. You're left trying to prove that there is no invisible massless gremlin on your shoulder, which is something you cannot do.
Predictiveness is of course also critical, but predictions that cannot be falsified are irrelevant. For instance 100 wrong astrology predictions in a row does not falsify astrology. It just means 100 predictions ended up being not true for some individual or another. But because we can happily ignore the times when predictions of astrology are wrong means you ought also be happy to ignore all of the times that it's right - as it's saying nothing where accuracy matters.