Economics is for me the prime example, even more so than psychology. They layer mathamatical models upon models to make it look like "hard" science. But even the most basic assumptions are disputed and you can basically find an economist (even from a reputable institution) for pretty much any position about any topic.
No economic theory has ever been found that would be capable to make even modest correct predictions (beyond very basic "rules of thumb" which pretty much have been established by Adam Smith and a couple of following generations). It would be fine with me, if they owned up to the fact that its a social science which can deliver some interesting models to think about the past. But every day you see economists on TV steadfastly claiming that they know whats going on and what's going to happen. Their chance is pretty much always 50/50.
Surprisingly, they can't even agree on the role of entirely artificial constructs with a small number of participants (no "invisible hand" involved) like the TARGET2 mechanism of the ECB [1]. It's all just completely arbitrary.
Economics can attract people making wild claims because it is difficult to truly prove or disprove some claims in the subject. Still, there is good science done in economics, though conducting science is more difficult in fields like economics than others. I think it is important to distinguish between microeconomics and macroeconomics here. It is possible to conduct controlled experiments in microeconomics, but in general it is difficult and often impossible to conduct controlled experiments in macroeconomics. For example, you simply cannot take half the population and apply one policy to them and take the other half and apply a "placebo" and see what the results are. The scale and ethical problems prevent such experiments.
Instead, macroeconomists like to use "natural experiments" (as they call it) that occur at discontinuities in current policies. For example, let's say that some policy has an income cut off at X USD/year. People making less than that can use the program, but people making more than that cannot. You can look at people making X-1 USD/year and compare them to people making X+1 USD/year. The slight difference in income is basically negligible, so in effect you are studying the effect of the policy, but only at a particular income level unfortunately.
Note that the difficulties using the scientific method continue to other fields like weather prediction. In these fields scientists can conduct observational studies but not experimental (controlled) studies, since we do not have the means (or even the desire) to effect the phenomena in such a way to verify or falsify some questions directly. Indirect (and far more clever) methods are still possible, though.
But isnt meteorology a good example of a science capable of producing models with high accuracy (albeit just for a couple of days in advance) in absence of lab conditions. I can't think of a single economic model as accurate as modern weather models (or even climate models).
Yes, you are correct. Weather forecasting is much more accurate (at least in the short term) than economic forecasting. My point wasn't about comparing the accuracy of the predictions between meteorology and economics, though. It was about pointing out the similar difficulties in using the scientific method. For example, someone might come up with some sort of hypothesis about what would happen if a category 5 hurricane appears in an odd location in the Atlantic, but nobody would then attempt to create that kind of hurricane just to see what happens. That would be an ethical and logistical nightmare. Instead, they would rely on some sort of simulation to test the hypothesis. This is the same kind of problem as in economics, but for meteorology the equations you are using to make the prediction are obviously on a much more solid footing.
Unsurprisingly I learned more about economics from historians, anthropologists (Adam Tooze, David Graeber) than practicing economists.
I’m currently enjoying reading “What’s Wrong with Economics”[1] which expands on some of the points you make. Recommended if you haven’t come across it yet.
It's a scam, and the best predictor of success as an economist is being politically useful as a scientific fig leaf for one interest group or another, sadly.
Here’s the thing to understand: the “economists” you’re watching on TV are not representative of the field, and TV economics is hardly a better source of information on economics than Facebook is on immunology. Confusing the cargo-cult for the real thing is the actual mistake here.
I think it’s easy to claim all the economists at Harvard, Stanford, MIT, and UChicago have no clue what they’re doing, and you could design a market better than Vickrey himself by throwing darts at a wall.
But the reality is that while they do disagree about a great deal, this disagreement isn’t different from the disagreement you see on the cutting edge of cancer biology or physics.
And sure, you’ll be able to find “Marxist Economists” for every Nobel Prize winner with a PhD in mathematics, but if you stick with the respectable academics (and especially with those focusing on microeconomics and game theory) they tend to agree about much more than you’d expect.
Economics is key to social engineering. Many concepts and core ideas of different schools of thought are completely loved and sponsored by actors interested in an agenda (say, maintaining their status quo or inflation https://wtfhappenedin1971.com/).
For example, we have the Keynesians saying that government spending is good. Guess who is going to sponsor this point-of-view? People who benefit from making the state spend or who believe the state has utter importance in defining society (as they call it, economic planning).
Others will downfall resort to mathematical formulations as they were absolute truth, and their premisses might contain a lot of prejudice.
I myself love the Austrian School of economics (https://en.wikipedia.org/wiki/Austrian_School) approach of the likes of Carl Menger and Mises, and their approach based on praxeology or individualism, and I never thought it could be abused for interest groups until I started seeing how many people are abusing its idea of subject value to promote Bitcoin as the solution to everything (when I believe Bitcoin with its proof-of-work strategy has more issues than benefits).
Some funny anecdotes / tricks I have heard about to defeat or detect cargo cult copying or cargo cult approaches:
1. Before our age of GPS maps and precision surveying, some map makers used to purposely put in little artificial defects into their roads or neighborhoods (non functional little things). This was so that someone who couldn't be bothered to do their own survey and copied the map exactly would not know the difference and reveal themselves by the defect (and be sued for copyright, etc). Such as little hooks on ends of roads, or even fake small villages in the middle of nowhere.
In fact, I recall one story where a village actually became real after enough companies had copied a certain map issued by someone who had put in one such fake village to defeat copying, and a general store named itself after the place, thinking it must be real.
2. When people plagiarize code without understanding it, leaving in non-functional variables or loops that the original author either didn't delete or put in as a trap.
Land Surveyor here. I have no idea where you got the impression that surveyors introduced artificial defects in the past, but this is flat wrong. Cartographers, who make maps at incredibly large scales, sure, occasionally. But “precision surveying” vastly preceded GPS (which...is still nowhere near our most precise instrument) and nearly all of the worlds infrastructure was laid out by incredibly skilled and competent individuals who would have been risking their livelihood had they done what you’re implying. Survey maps aren’t “copyrighted”.
The Ordnance Survey in the UK won £20 million from another map maker (AA) by showing that AA maps contain "fingerprints" the OS includes in its maps. According to the OS these tend not to be significant mistakes but small changes like the width of roads and the like.
I find it sad that over 40 years after this famous speech, which should have been a wakeup call, psychology was ground zero for https://en.wikipedia.org/wiki/Replication_crisis. Psychology is making small improvements. But most of what Feynman said was wrong with psychology back then is still wrong today.
The lack of this being seriously addressed and remedied hurts the reputation of science in general. It also undermines the argument in areas of public policy around say Covid, where "listen to the science" has to overcome the perception that a lot of what passes for science... isn't.
In Covid it was far more visibly clear this was and continues to happen. Everyone knows in their hearts that temp check is bullshit but that’s something everyone does even today. The about face on mask policy initially probably still plays a role in people’s perception and inability to follow rules today.
> Everyone knows in their hearts that temp check is bullshit but that’s something everyone does even today.
All our local visitor attractions (and I suspect it's the same UK wide) do temp checks on condition of entry. Pointing out the flaws with it is futile as it's the covid equivalent of security theatre. It ticks a box and makes it look like they are taking precautions.
It reminds me of those snake oil bomb detectors that are still used in some parts of the world despite the fact that the 'inventors' have been jailed for selling the fraudulent devices: https://www.bbc.co.uk/news/uk-29459896
if it's being sold as a silver bullet that stops all covid (or something else) , well sure, obviously that's bullshit.
or is it the idea that it's OK to open everything if you just do temp checks, cause yeah, I can see how that would be bullshit too.
But set a high enough threshold (to escape measurement error) and with accurate enough tools, it's one of the few public health things that seems somewhat evidence based: if you've got a verifiable fever, you probably shouldn't be out and mixing with others in confined spaces.
hell, it probably made community health sense before COVID.
obviously this doesn't stop non-fever, non symptomatic transmission, or people who dope themselves or their kids up to avoid detection, and it doesn't magically get rid of covid, but it picks out the low hanging fruit, and assuming properly calibrated equipment, seems like an obvious public health win to me (as I said, if you've got a fever, you should bloody well be going home/ denied entry to a workplace even in non-covid times).
By the time the 75% of patients get fever from Covid they are often fully aware that something’s not right and are probably not leaving home or have already gotten tested, only the crazy ones would still venture out in that stage.
Importantly, the reason Covid spread this fast is because the disease seems to attain peak infectiousness before symptom onset and definitely before fever onset. This is why SARS was nipped in the bud with IR scanners while Covid wasn’t.
You say that temperature checks don’t hurt but they most definitely do, most places where they do a check, especially in places like India, they think they’ve done enough and don’t enforce strict mask policies or distancing measures any further. If people are forced to stop using temp screening then these places will have to come up with alternative methods of comforting their audience they are taking measures so they would be forced to do actually effective screening methods like taking o2 measurements. So yeah it’s stupid to allow temp checks to pass for any theatre.
>only the crazy ones would still venture out in that stage.
Depending on location, that could still be a useful segment to exclude. People takkng their kids on a long-planned and long-promised visit to an attraction, who wake up feeling 'a bit under the weather' may be tempted to go anyway.
Internal temperature is indeed a pretty decent indicator. Yes, everyone's normal body temperature varies a bit (and thus their fever threshold), and, yes, different activities will swing your temperature around a bit, so it is sometimes tricky, but internal temperature works.
The problem is that no one measures internal temperature outside a hospital (partly because of sanitation issues). Everyone measures external -- skin -- temperature, usually using an infrared (IR) thermometer.
Those things are terrible to begin with. But they're not even used properly!
To get good results from an IR thermometer, you have to perform an emissivity correction. Which is easy enough for one measurement on one person in a lab, but it will and does vary between people by more than enough to cross the fever threshold. We've got a Flir E60 at work and have tried to use it to make this measurement reliably. We can't do it. It just isn't accurate enough. (You can get relative differences shockingly accurately, but unfortunately temperature screenings need an absolute temperature.) And that's a $10,000 IR camera, set up by skilled R&D personnel. Most places have a cheap Amazon-junk IR gun "thermometer" operated by the hapless.
Maybe we could fix that somehow, so we always get an accurate measurement of skin temperature. But, as you've probably noticed by now, skin temperature isn't internal temperature. If you, for example, walk a mile outside to reach the "temp check" station, on a chilly day... your skin will be colder than the rest of your body. Well below any fever threshold, in fact, unless your fever is so strong that you're already well aware of it (and therefore actively lying, or you'd be in some kind of quarantine by now).
So it's just not a measurement that you can make accurately enough to mean anything, at any kind of scale.
Thus, the cries of "security theater". Because it is.
I wish temperature screening worked, but it simply doesn't. Not the way it's used in practice, anyway. (And don't ask what thermal camera someone is using, because it kinda-sorta looks like the high-end one you tried out that doesn't work at all... you'll just get that good old deer-in-the-headlights look. Or worse. I was probably lucky.)
Case in point - I was stopped from entering the supermarket here in Singapore because I measured over 37 degrees Celsius....but 5 seconds later I re-measured at 32 degrees, and was waved in.
So this comes back to my statement about proper equipment and proper thresholds.
obviously 37 practically isn't a fever, and 32 is ridiculous.
But let's accept it on face value, so it gave you a false positive, which caused you to pause for a second reading, and then a true negative (well, unless you count the 32 as a false positive for hypothermia) and you went in.
That temp checks create false readings isn't really news: for me the question is so they create statistically significantly better outcomes. As a statistics guy, my head says they do. The objections I hear to them sound to me like the objections to bmi: that it's not universally perfect is not reason to dismiss it. This seems to me like the perfect being the enemy of the good enough.
You called yourself a stats guy so I’m going to push, can you show any rigorous analysis by anyone suggesting that theres any benefit to doing temp checks? Is there any number anywhere that gives false positive rate estimates for typically used temperature checking methods? Or false negatives?
First of all, the important thing is the R number - how many people does the average infected person infect. If that is above 1, you get spread. Below 1, it dies out. Without mitigations, R is estimated at about 3. With mitigations, the UK is posting various numbers in the 0.7-1.0 range. (See https://www.bbc.com/news/health-52473523 for a source.) Let's assume that the USA is similar.
The question to ask is this. Do the temperature checks make the difference between being above or below 1.0?
Per https://www.webmd.com/lung/news/20210110/59-percent-of-covid... about 60% of spread comes from people who are asymptomatic. So about 40% happens from people who are symptomatic. Per https://www.healthline.com/health-news/what-is-the-risk-of-g... about 55% of spread happens at stores. If those factors were independent, about 1/4 of the spread would happen from symptomatic people at stores. If the temperature checks avoid a significant fraction of those, then in the real world it may indeed be the difference between pandemic spread and dying off.
Well in case of public policy it's rather lack of trust in public policy. And in cases like Covid public policy should be ahead of science anyway, because there is no time to wait.
Science (especially social sciences) is many, many times wrong, by definition (that's what science does, it gets less and less wrong until it reaches a certain epsilon of "less wrong" which it considers as the "truth").
This is not the fault of science because, again, errors/being wrong are included in its definition, so to speak, the problem is that those errors/wrongs, when applied to the life of real people, with real human lives, will most probably kill them or affect their lives negatively in very big ways, especially so if it's a concerted effort coming from the top.
There are countless examples for that, the latest that comes to my mind is the pro-austerity policy imposed by the IMF on Greece, because that's what the science of economics was advising at the time, i.e. austerity. It turned out that that was a mistake (and IMF were quick to acknowledge it), but by that moment the damage had already been done.
There's also the example of failed policies imposed by the likes of Robert McNamara and his men (they were mostly men) back in the 1960s, but I didn't catch those times in person, so I only know about those failures via links found on this website (and which I'm too lazy to search for right now).
The way I read your argument is "science is not 100% correct, only 85% correct. That means they are 15% wrong. Since science is 15% wrong and those 15% wrong will have negative consequences, we should ignore science and do what we feel is the right".
I acknowledge that science doesn't have all the answers and probably never will. However, I believe it is better to take its results into account when making public policy instead of relying on gut feeling or ideology.
I certainly hope science is taken into account when deciding things like building regulations, fisheries quotas and food additive regulations.
If we disregard "science" how are we going to make any decisions? It's not like we're using magic here. Science is simply a way of attaining and verifying knowledge. To abandonment means to abandon those thought process for evidence, etc. Rejecting bad science is important, but science is the thing most equipped to do that. The alternative is doing things based on personal experience or a belief system, which historically have had worse results.
Every way of thinking is wrong, but science very quickly becomes least wrong.
Yes of course there will be cases where doing what the science says produces a bad result, but that is no reason to trust the tea leaves instead. The question is not who is right, it's who is closest to being right. On average, science will consistently outperform all other approaches.
there's lots of public policy that is antagonistic to more grounded science than the social sciences, science that while it can be proven wrong at some point is unlikely to be and if it is proven wrong will probably be in small particulars rather than in the large. Evolutionary theory would be one obvious example.
Some social sciences are not even trying to be scientific anymore. They have denounced even the mere concepts of truth and of the scientific method as a tool to approximate it. These people need to be removed from academic life and from all teaching positions as a matter of self defence.
(Ed. And they and their sympathisers obviously know who they are.)
The difference with CS is that there's no one to trust. Everything you need is right there.
Other disciplines have people fake results. You can't fake a result in CS since people only really attack decidable problems, and you need no empirical results for any. It's like how even though mathematicians ultimately do make errors, Mathematics has no replication crisis.
However, most so-called 'hard' science are also full of knowledge that poorly models the world. Sadly, since most people who say this are also climate kooks, everyone will label you as a climate kook if you say this, though.
People constantly downvote me on HN for claiming that there are photoshopped papers out there. I'm not complaining about the downvotes. We make the culture of this place together and I can accept if I say things that you believe aren't right. But I am right.
Edit since rate limited: oh, I didn't consider ML. Okay, consider my position reversed. Yeah, the empirical sciences do have reproducibility issues.
The ML field alone is full of unreproducable papers.
Example: "“Probably 50%-75% of all papers are unreproducible. It’s sad, but it’s true,” another user wrote. “Think about it, most papers are ‘optimized’ to get into a conference. More often than not the authors know that a paper they’re trying to get into a conference isn’t very good! So they don’t have to worry about reproducibility because nobody will try to reproduce them.” https://bdtechtalks.com/2021/03/01/papers-without-code-machi...
This does not seem very good data to empirically underlay the claim of 50%-75% papers being irreproducible. I don't doubt that there are (too) many of them, but having some real data before making strong claims about numbers would be more credible.
When I said that psychology was ground zero, I did not intend to imply that there wasn't a significant blast radius.
That said, the blast radius is significantly less than you're describing. Most research in the hard sciences was not impacted at all. But that is in large part because they absorbed lessons about the difficulty and importance of replication many decades ago. And some of the lessons that they absorbed are being passed along in Feynman's speech.
Experiments involving human behavior are especially hard to design. Finding ways to figure out what people really do in real world conditions is quite a lot more challenging than most people want to believe.
This is why I think all UBI experiments are bunk: The subjects know it is a limited-time experiment, not a new government policy, so they don't behave like they would if they believed this income was permanent for the rest of their life.
I have this idea of a MOOP - Maasively Open Online Psychology.
Basically it's - we have these constantly on monitoring devices, phones, watches, alexa, video - we can collect constant human behavioural data - and given the right
regulatory framework (ie this is medical data treat it like that) we should see enormous commonalities (correlations) between millions of people.
Psychology should then become a field of epidemiology- as opposed to limited behavioural studies in lab conditions.
Maybe psychology just has not had enough in the field data?
It's easy and tempting to say things such as 'X is not science' or 'X is cargo cult science'. However, if you take your X to be large enough, like a whole discipline such as psychology, or computer science, or archaeology etc., the statement becomes a truism - not useful at best, misleading at worst.
Let's take psychology as an example (use computer science in your mind if that's more familiar), it's a vast field spanning the whole spectrum from theoretical to experimental work. You start with something completely unscientific at one end and end up with the essence of the scientific method at the other.
Well here I wish to offer a modern update of cargo cult science.
Today we suffer more from cargo cult enthusiasm -- the collective delusion that all we need is someone very passionate about something to lead us, and we will get there. Force of personality and personal passion must mean correct thinking.
Maybe just building coconut headphones would do less harm.
I would suggest the entire science funding apparatus is also a type of cargo cult. Money is allocated to high profile academics who are good at writing compelling proposals in the hope that they will continue to produce the ‘cargo’.
Younger academics, who may have fresh ideas or be closer to the technical details struggle to get funded because they are less skilled in the buzzword incantations needed for an attractive proposal.
The response is for younger academics to write their own cargo cult proposals that align with what they believe the funding body wants rather than what they think is good science.
This is 100% because NSF funders are not allowed to make "mistakes". They need every dollar of grant money to lead to published research, so they almost always bet on the established horses. Simply changing the rules to allow funders to bet on longer shots would fix much of this.
I wouldn’t consider them empty headed, just focused on the cargo rather than funding good science. There is too much of a focus on ‘strategic plans’ and theme alignment because the circular reasoning tells them that pumping more money into an economically relevant area will make more jobs in that area. It just leads to recycling of ideas and instead of producing new science we become tax payer funded finishing schools for industry.
My understanding is that there is frustration in the European Commission that H2020 has not produced more economic impact. Of course every proposal is going to promise transformative ideas because that’s what was listed as a requirement but when you fund low risk research spread across geographically and culturally diverse consortia you get the lowest common denominator output.
My humble opinion is that you should fund more small scale projects and fewer big consortia. Let the risky ideas have a chance to publish and stop investing in rock stars.
"When you have put a lot of ideas together to make an elaborate theory, you want to make sure, when explaining what it fits, that those things it fits are not just the things that gave you the idea for the theory; but that the finished theory makes something else come out right, in addition."
Intelligent people are good at pattern recognition and finding logical explanations for events: why BTC went up, why an election swung a certain way, what's happening in the economy.
For me, the main benefit of science is not that it can "explain" but that it can "predict". Next time you try to offer an explanation for something, ask yourself if your explanation can predict some events in the future.
In a very fundamental level I think that is exactly what the idea of science is: to predict future. That is the whole point. We want useful models that given an input (observation of current state) produce output that says how/what things are going to be/happen.
Any scientific theory has two well defined goals: to tie multiple part of the known together, and also tie those parts to some part of the unknown.
Prediction is an important point, but it's just as important as meaningfully stitching together what's already known to produce new knowledge.
I'm academically ignorant about economics, I'm honestly asking is there really a body of knowledge about economics, like different theories conversing between each other? Are theories built on top of each other?
I think that's not just your opinion, but rather the actual definition of scientific theory. Lots of popular science or introductory textbooks begin with this very definition.
> A great deal of their difficulty is, of course, the difficulty of the subject and the inapplicability of the scientific method to the subject.
"Inapplicability of the scientific method to the subject" is exactly what I wish was more widely appreciated. Perhaps I would just rephrase it to say that "the scientific method won't get the same results it does in other fields". It's not that psychologists/etc are stupid or malicious. If you talk to any big player in that field they will gladly acknowledge all the problems. A lot of research will try various things to minimize some of the limitations. Nobody will hate you for pointing it out. But the problem is none of that can ever work. And if you take that position there's just no place for you in these fields other than being a hater.
Take something simple like non/fake respondent bias. Everybody knows about it. Nobody flat out "denies" it. But most sketchy fields think they can work around it. You can do better than the other study, but you cannot cancel it out. It's fundamental. I don't care how stable the result is. If what you're studying can only ever be ascertained by asking people you're basically a pollster. You're studying people's responses, not the "nature" of some psychological concept you came up with.
This is not to say all the work is useless. Polls, stock market models and these kinds of things can be useful in a very bounded context. But pretending it works like some other fields where you can just build on top of previous results is a futile endeavor. I'm not going to argue the science/not science part, I'm fine with people calling it "science" if applying the same methods is the only criterion. I think useful/not useful is a better debate. For example, biology, medicine in particular, is IMO rife with sketchy stuff, but once in a while the field can indeed knock it out of the park and produce something useful.
A lot of what this talks about can be summed up as trying to root out confirmation bias and personal blind spots.
Since humans make science, dealing constructively with human tendencies is a necessary part of science, though it's one we tend to not really address directly. We act like the process matters and equipment matters but we give short shrift to the people, their character, how aware they are of their own bad habits, what best practices counter such human tendencies, etc.
But this quote is about a kind of systemic issue, about scarcity of resources and about broken bureaucratic processes:
In order to compare his heavy hydrogen results to what might happen to light hydrogen he had to use data from someone else’s experiment on light hydrogen, which was done on different apparatus. When asked he said it was because he couldn’t get time on the program (because there’s so little time and it’s such expensive apparatus) to do the experiment with light hydrogen on this apparatus because there wouldn’t be any new result. And so the men in charge of programs at NAL are so anxious for new results, in order to get more money to keep the thing going for public relations purposes, they are destroying—possibly—the value of the experiments themselves, which is the whole purpose of the thing. It is often hard for the experimenters there to complete their work as their scientific integrity demands.
And I am reminded of how in the early days, scientist often seemed to be wealthy or eccentric individuals who found some way to buy or create what they needed on their own rather than cogs in a bigger science machine. And we should try to recreate more of that, which is part of why I liked that "I baked ancient bread!" story so much:
Honestly it reminds me of the Enlightenment, when most research scientists and mathematicians were bored rich guys who were looking to impress other bored rich guys.
The practice of cargo cults in Papua is actually far more interesting than Feynman gives it credit for. It's worth going and reading about them. There was a whole lot of other social stuff tied up in them beyond "oh, we want cargo."
Tldr:
Wanting without understanding leading to superficial mimicking in hope of gaining exacerbated by (insert any or all compounding factors that lead to defending ignorance)...
This is one of the classics, vindicated again and again, even though the author is long dead.
I also wonder whether Feynman would be accused of racism for writing this today, given that he talks about a non-Western cultural phenomenon in a way that someone surely finds offensive.
No economic theory has ever been found that would be capable to make even modest correct predictions (beyond very basic "rules of thumb" which pretty much have been established by Adam Smith and a couple of following generations). It would be fine with me, if they owned up to the fact that its a social science which can deliver some interesting models to think about the past. But every day you see economists on TV steadfastly claiming that they know whats going on and what's going to happen. Their chance is pretty much always 50/50.
Surprisingly, they can't even agree on the role of entirely artificial constructs with a small number of participants (no "invisible hand" involved) like the TARGET2 mechanism of the ECB [1]. It's all just completely arbitrary.
[1] https://en.wikipedia.org/wiki/TARGET2#Intra-system_credit_fa...