o09rdk

In my experience, among those who are employed and paid well, the hours are often brutal and huge sacrifices are being asked.

Then as a society we pretend it's all a meritocracy and if you are just talented enough and hard working enough it will work out. If it doesn't, it's your fault. Says the survivors.

Is it any surprise so many are tuning out?

It's cliche that's nevertheless true, but for many the system is rigged and it's looking more like that all the time. If you feel like it's all a rigged game why deign to play it at all?

I worry that the focus on basic income etc and loss of labor jobs is kind of missing the point: a lot of these individuals could and would work if the right social contracts were in place, and people were really allowed to provide what they are capable of in a competitive market.

This is possible if there were some systematic shift in instrumentation with measurement year, but it would be blunted by the extent to which different ages were being sampled at each year.

I agree though that it would be nice to have some information on instrumentation over time and how it relates to temperature measurement.

It's worth noting that the results have practical implications regardless of the explanation of the observed trends, though, which is it's commonly assumed normal body temperature has certain characteristics, without recognizing that that might have shifted in significant ways over time.

It's interesting to me personally because I've often observed my body temp when I'm feeling fine is usually just above 36 C. That doesn't sound like much but it becomes a bigger deal when deciding whether or not I have a significant fever. Some of that might just be individual variation, but this is suggesting there's cohort trends within that too.

That said, I think he might be wrong about (3) (among other things) and (5) is related in my mind. China is certainly growing, in part because of problems in the Euro-American world, but it is also brewing problems and unrest with increasingly autocratic behavior, and a number of other economic problems are showing.

I do see increased adoption of decentralized technologies, in part because of increased political volatility and mistrust of involved centralized IT corporations. That is, I see (5) happening in part because I think (3) might be a bit of an overprediction.

It won't just be China though, and hasn't been. People will increasingly be dealing with censorship and monitoring, and trying to bypass it, and will also become increasingly distrustful of things like Facebook, which correctly or incorrectly has been implicated in things like Russian-UK-US subterfuge. Adoption of decentralized tech is already happening with Hong Kong and Catalonia protests; I suspect it will grow to become more mainstream.

I don't see the traditional centralized internet going away, but I do see people gravitating toward a more hybrid system. Maybe where more critical, and more intimate communications with close others moves to more decentralized architectures, along with other stuff that doesn't depend on speed so much, whereas other mass communications remains more centralized.

I do think this is a problem everywhere, in academics, in medicine, and probably other fields that I have less knowledge of. I think income inequality problems are really less of a problem per se (although still definitely a problem) than the power inequity problems that income inequality reflects. Some of it is rent-seeking monopoly problems, but I think there's a broader trend societally (as reflected in right-wing authoritarian trends globally) toward a lack of skepticism about power differentials, whether the inequalities we see are really justified in the magnitude we see them, and whether other power structures would be better.

I personally think with academics it's particularly problematic because the endgame (supposedly) is acquisition of knowledge where we don't really know what's going on, and I happen to believe the best environment for that is one where it's everything goes, closer to a democratic-libertarian-deregulated kind of environment.

My impressions of the time I was tenured are very close to what this guy is saying, although I think it's the tip of the iceberg and a lot of problems are being brushed over (appropriately I admit given that it's not really the right piece for it).

To be fair to Alexander, therapy research and literature has the kind of weird duality he points out, and it permeates academics as well, where the problems he mentions are maybe even worse. Everyone does know about the dodo bird effect, but then people either embrace it or they ignore it.

A big problem is the exact same thing that's permeating all of medical science, which is a huge pressure to obtain grants and notoriety for discovering a cure. There's much more rewards for overhyping than being sober in science. You could replace therapy with drugs in his essay and it would read the same. In fact, I sincerely was confused for a bit by his figure because the exact same sorts of figures have been published about psychiatric medications, and it took me a while to be sure what he was talking about.

One thing that's maybe not immediately clear to the general public is that with psychiatry and psychology there's an additional pressure to "be scientific" which then often backfires in the way he's witnessing. Both psychiatrists and psychologists are the targets of witch doctor labels, which then leads to this behavior where researchers and theorists jockey to outdo themselves in being "real scientists". The good part of this is that then you get a ton of randomized controlled studies and so forth, but the downside of this is you end up with this additional reason to promote your drug or therapy as being somehow better than treatment as usual, or to denigrate other interventions as being "unscientific". That is, certain things (e.g., CBT in therapyland, or arguably things like prozac in medicationland) become banners that you're expected to carry if you want to be part of the "real science" club. Criticizing the efficacy of these interventions implicitly doesn't just mean criticizing those particular interventions, or criticizing the specificity of their mechanisms, it means you get labeled as not being on board with the "real science" club. It is just one more reason for hype, in addition to the typical hype that is flooding biomedical research at the moment.

Alexander mentions that he feels like he's never seen "sudden, extraordinary, long-lasting" effects of therapy. I'd suggest it's one of those situations where you get to pick 2 of 3, nothing more.

It is true that these are self-report questionnaires, but as such they are small samples of behaviors of the people in question. Samples of how they perceive themselves, how they think about life, how they think about others, and what they value.

The Big Five, and the measures used in studies such as this, has been validated (in the sense that the ratings have been associated concurrently and predictively) over decades in many ways, with regard to daily reports of behavior, emotion, and life events, diagnoses, work ratings, performance on tests, ratings by peers and colleagues, ratings by strangers, just about everything you can imagine. These self-reports aren't perfect, but they do provide a fuzzy snapshot of someone at a given moment in time. Yes, it would be better to obtain all sorts of other measures of behavior, but they would be too expensive to obtain on large enough samples to be representative.

A major paradox in understanding human behavioral differences is that the more specific and "real world" you get, the less and less they generalize. That is, you can get a very concrete measure of a real-world behavior, but it ceases to be representative of that person across a large number of contexts and situations. Say you want to measure theft, for example. Do you set up a honeypot? Is that representative of that person? Do you use police reports or records? Is that representative? It turns out self-report on online questionnares is a very good measure of things like this because people are less self-conscious, and report things that don't go on the official record.

Faking is also controversial in this area. You're right to bring it up as an issue, but to understand the research on it it's important to think about why someone would fake. That is, what's the motivation for large proportions of people to systematically fake in one direction? And if they do do go to the trouble of doing that, what's "real" and what's "fake"? That is, let's say people make themselves look more dominant than they really are -- what does it mean if one person does that and another does not? It turns out that the person who wants to make themselves look more dominant often is more dominant, all other things equal, because it means they value that.

Also, strangely enough, it turns out that people who are callous and aggressive don't really care about that, especially on online questionnaires, because they are callous and aggressive.

This has all been very thoroughly researched and it turns out to be much more complicated than it seems at first glance. It doesn't mean things can't be better, but it does mean that over very large samples of persons answering questions on a low-stakes questionnaire (in the sense there aren't real consequences to them answering one way or another), a lot of these things average out. It's not the end of the story, but it's not something to be dismissed either.

In the end, questions of sex differences in behavior are about sex differences in behavior. And that's what this research addresses.

Speaking from personal experience, the problem with large public institutions (maybe any institution?) hiring these consultancies is that there's enormous intrinsic pressure to accept their recommendations.

Say, public institution X spends exorbitant fees on consultancy Y. Now they're going to choose to ignore the recommendations?

If the recommendations are bogus, there will be hellfire about why the public institution is transferring huge quantities of funds to a private organization that is incompetent or sketchy. If the recommendations are not, then there will be hellfire about why the public institution is ignoring the recommendations.

As a result, a lot of times there's suspicion that these consultancies are just cover for administration or management making decisions that would be unpopular, controversial, or self-serving. At least in my experience, they're seen as a power grab for those already in positions of power. That is, the consultancy comes in, asks management/admin what they would like to see, and then they put it in the recommendations. Then admin can say "look, we didn't tell ourselves to [do incredibly controversial thing Z, like cut basic staff and give administration big raises], this fully independent third party said to do this."

As a result, I think hard questions about what you're getting from consultancies are actually apropos of the Vox article, and relevant to the perception that shady constituencies are somehow hijacking public services for their own benefit. When you pay large sums of money to a consultancy, whose methods are unclear, and have unclear scientific support, and assumes that, e.g., management of publicly traded for-profit companies and management of non-profit government organizations can be treated similarly, it leads to legitimate questions about whose motives are being served.

The WHO would have been better off imho convening a special strategy or reform committee made of inside and outside members than something like McKinsey.

Of course this will be correlated with wealth, because dressier clothes are more expensive, and older individuals probably will have more wealth than when they were younger. However, the clothing isn't the spectrum I was expecting necessarily, and I'm not sure "rich" versus "poor" are the right labels.

Also, I can understand how most of the clothing came to be categorized the way it was, some of it is confusing to me. There's definitely items that I think would be classified in the other category. It would be nice to tailor this more to specific individuals more.

To be honest I'm a bit surprised this is getting this much attention because the social psychology literature is full of findings like this already.

I kind of think Brent could have gone further with some of the implications of the ED effect size being so small. To me, it's interesting that it is basically zero (I don't think it's even .08, based on Bayesian prior choice effects, which is another interesting angle on this story). Why? Because I think it's a fairly intuitive idea that gets applied by people in everyday life over and over again, and has broader implications.

For example, you are in a situation where you're tempted by something, have to resist temptation, and experience it as exhausting. You go home from work and are tired and need to relax, ostensibly because you're tired by being on your best behavior all day. There's recent very cogently argued public health literature arguing that diets don't work largely because it's too hard (read: too fatiguing, too effortful) to resist our favorite foods, either in the short term or over the long term.

In short, I think there are a lot of people who would be surprised that the ego depletion effect is so close to zero. Maybe not everyone, but a lot of people. It also raises the question of why it's difficult to resist temptations over a sustained period, and how to model that process more accurately (my reading of the ego depletion literature is that a lot of the effect can be explained by sustained negative emotional state or heightened attention, rather than resource depletion, but that's a different issue; another explanation is with reward prediction models, but I think those are predicting a different aspect of behavior).

I guess I'm agreeing with you, but I think the magnitude of the effect size being zero or close to it is important for a lot of reasons aside from meta-scientific issues pertaining to replicability and trustworthiness of results.

My sense is that a grant, like an academic paper, is often basically well-received or not. So prior to going into it, if you're familiar with the details of the particular grant, you can get some sense of interest or not. The grant may have even been solicited in a certain sense by a program officer or something, so it's quasi-invited.

Those grants that are quasi-invited, or well-received, will basically involve polishing on subsequent revisions. Those that are totally unsolicited and not well-received it doesn't matter how much polishing you do often, it will not go anywhere.

I think where this becomes relevant to the paper is that this often has very little to do with the process of the grant revisions. It says little about "how someone responds to failure", and everything about connections with the grant agency, program officers, and luck. All that stuff that happens before the grant is even submitted plays into it, and this paper kind of ignores that.

You could say that it does speak to learning, in that you could ask "why doesn't the other scientist adjust their strategy?" To which I'd say, there are enormous pressures to submit anyway, and for someone who doesn't get good mentoring about what to target, or what agency to target, or whose research doesn't jive with the priorities of the division head at that time, or whatever, they might just not get it, and it might easily be over the 5 year period of the study.

Basically, at least with NIH grants I think this study is really misleading and potentially harmful, because it kind of suggests failed applicants just aren't learning, when I think what's really happening is that you have a mixture of two groups, one of which can learn something, and the other of whom it just doesn't matter if they learn or not, because the outcome has kind of been preordained.

That is, the process they predict from is an indicator of which mixture class the applicant belongs in, not the cause of the outcome.

HR gets a million applications, people are evaluated on what boxes can be checked off rather than their skillset, etc.

With regard to the educational system, the problem is that degrees are seen as signals or credentials (to use the language of the article) rather than as a background. That's admittedly a fuzzy distinction, but it really reduces the degree to the degree per se rather than the host of courses and experiences the person during college. So, we talk about "useless degrees" without recognizing that someone might have had equivalent coursework and experience without majoring in something else per se. In the end it doesn't matter.

I think your impression is generally spot-on although there's nuances and it's more complicated than that.

In general, universities are suffering from many of the same pressures as other fields. Nonprofits are being run as profit-generating centers, and the implicit promotion track goes from faculty to senior faculty to administration, if you're even on tenure track. The administration salaries are increasing, as are the number of administrative positions (deans, associate deans, vice associate deans, assistant deans, etc.), and the structures are becoming increasingly hierarchical. Things are very top-heavy and expensive to run because the salary budget is so disproportionately distributed.

I don't think that's all of it, though. The rest of it is harder to quantify, but is the flip side of the educational bubble coin. Students are rushing to go to college in record numbers, which then puts pressure on colleges to stand out to attract more students and more high-paying students who become high-donating alumni. Institutions that should be getting public funding are not, which then trickles down to students; other private institutions then benefit from the increased tuition norms etc.

It's a complicated problem that involves HR and administrative practices and norms, economic incentives within university funding structures, incentives coming from a broken employment system in the US, problematic incentives coming from federal grant structures, etc. etc. etc. etc.

Also too, I think the software-as-a-service and cloud model has created a certain expectation among society, that everything is handed off to someone else. Decentralized tech requires a bit more involvement, that might be more familiar to older experienced users but not newer ones, or might require native users who gravitate it as their first choice for various reasons.

Finally, there are limitations with decentralized tech; I suspect that any eventual equilibrium will always be in flux and will include some centralized, decentralized, and federated technology in various proportions depending on societal circumstances.

Everything can be cast in terms of economic incentives perhaps?

Some people take drugs.

They tend to onset around the same age (adolescence to early adulthood).

Ergo, by base rates, some will start around the same time completely by chance.

Also, I get the sense that some hallucinations are "amplifications" of various types of pre-existing sensory experiences. I wouldn't be surprised if some people just started noticing certain things they were doing already but hadn't noticed before.

Not really arguing against the idea/phenomenon of [semi-]permanent drug-induced hallucinations. Just suggesting that it's hard to interpret the data in some cases because of various alternative explanations.

There's a well-known phenomenon whereby narcissism and self-evaluation is generally related to positive mood. It's part of an overall positive tint.

So one interpretation of this is that people who tend to be optimistic also tend to just feel better in general, for whatever reason.

It's not clear to me these "benefits" would transfer to all scenarios, nor is it really clear that it's about insight in general in the way people usually understand it.

Not sure if a junk food diet would fly, but they could have at least given both groups a gift card and some gifts (maybe the control could have received non-food items of equal value?) and had the same phone calls, but with some "neutral" questions in the no diet change group.

There are plenty of studies that do behavioral controls. Not sure why they didn't do it in this one. Especially when there's a large literature on waitlisting effects on depression (basically people who are waitlisted as opposed to assigned to a behavioral control tend to have the worst outcomes).

Where my family is from, in the northern US, the night sky is very clear and visible (barring clouds etc which aren't uncommon for various reasons). But where we live you can barely see much other than the moon and planets. I wish more attention were paid to light pollution.

She's too young to appreciate it now when I point it out to her when we have the opportunity, and I worry that when she's old enough to appreciate it she'll be someplace she can't see it. For me it was a common thing up until I was a bit older, and now it's uncommon to be in the right place at the right time to see it. I wonder if the unlit night sky will become something like old-growth forests or other ecological sights that are long gone.

He seems to be in favor of large randomized-controlled comprehensive diet protocols, for example, where many aspects of diets are manipulated simultaneously. He dismisses them too, though, for very good reasons. Given that there were additional reasons to be skeptical of such studies he didn't mention, I was left wondering if his pleas for study designs should be heeded very much (as opposed to his criticism of specific flaws). I love his work, but am wondering if it's best approached as worst-case criticism rather than an inspiration for what to do.

It wasn't that long ago that the role of microbiome was recognized, for example, and even now it's contentious. Conversely, scores of epidemiological findings have had nice physiological explanations that turned out to be much more complicated, like those involving saturated fat, cholesterol, and carbohydrates. DHA, omega fatty acids, etc. are similar. Physiology is complex enough that you can accommodate just about any finding.

To carry this a bit further, it can be problematic going in the other direction too, from physiology to populations. Some of the anti-sugar advocacy, for example, was rooted in lab studies that neglected the effect of fiber in digestion of fruit, so you had people advocating that whole fruit consumption be avoided based on incomplete understanding. It's hard to argue that unprocessed whole fruit is undesirable on the basis of observational nutrition studies. Someone looking at the wet lab data and advocating abandonment of whole fruits might have done well to look at those observational studies to see why different trends were occurring.

I'm not saying that tying one level to the other at some point isn't necessary. I just think one or the other level can't be dismissed out of hand just because things don't map onto each other initially. It's possible that one level provides clues that the other is missing; it's also possible to come up with plausible but wrong bridges between the two levels.

On the one hand the traditional Bayesian response is something like "yes, we're making our prior assumptions explicit and then incorporating that into a formal inferential paradigm."

However, this prior is being used to bias the estimates, rather than to avoid the bias. That is, it would be akin to Dunnington taking the current estimates of e/m and using that to shape any new estimates from data. The argument is then that at least he was being explicit about his biases and how they are used to make an estimate.

This has always seemed backwards to me, though. It seems what is more defensible is to use a formal theory about how prior biases affect estimates, and then to leverage that theory to minimize biases. This is basically the idea of the reference prior, to estimate things such that any role of the prior is minimized in an information-theoretic sense. This seems more analogous to what Dunnington was doing.

I really wish reference priors were more widespread, although they can be computationally pretty hefty. It's one of my hopes that quantum computing might make these types of approaches more feasible in general.

He's famous for lambasting psychology, some of which is deserved, but some of which is just ignorant. Earlier this year he started complaining about psychologists ignoring long-tailed/skewed distributions, which is so absurd and widely written about I don't even know where to start, as it's a basic corollary of a lot of modern models. He also selectively miscites literature; god forbid he should do a systematic meta-analysis.

More of my concern is reinventing the wheel. I think he deserves a lot of credit for drawing attention to wrong models and assumptions in economics but model misspecification has been a topic that's been written about for decades in the statistics literature (e.g., "all models are wrong but some are useful", sandwich error estimators, pseudolikelihood ratio, model averaging, the work of Dawes, etc.). I agree with the general point he's making with these types of things, but his discussions would be a lot more interesting if he gave credit where it's due, expanding on others' past work rather than burying it.

Mostly I wish he'd tone himself down a bit, to get his signal:noise ratio in a better place. But then maybe he wouldn't draw attention to some of these issues.

Other times he's spot on. Usually it's not anything actually new, but he deserves some credit for being such a bulldog about it.

At the same time, there's been acknowledgment that the forms of illness encountered in reality don't really map on well to diagnostic categories. So rather than major depression and generalized anxiety disorder, it's more like there are levels of general distress, and specific problems with low positive emotion, or high negative emotion, or arousal. This has contributed to things like the NIH RDoC initiative.

These two things have sort of combined to lead to a lot of questions about what antidepressants are actually doing, in terms of how it's affecting dimensions of mood and other areas of functioning, thought of in a more detailed way. So in this study rather than targeting a specific diagnostic population, they did a randomized controlled trial of a heterogeneous group of patients, and then assessed antidepressant effects on distress and well-being in a more detailed, comprehensive way than the typical diagnostic assessment.

The results aren't really surprising but are interesting to think about in the context of these issues. For instance, it's not surprising that antidepressant effects might manifest in well-being or anxiety measures, because general anxiety and depression measures can be correlated as high as .8 or .9, and well-being, while less correlated, still tends to tag along. From a replicability standpoint it raises some interesting issues because there's sort of a "fishing expedition" given all the measures, even though things were preregistered, which is worth some discussion, but the effects are all in the anticipated direction even if not significant, raising the question of whether or not p-values really are the most important thing in general.

Overall the study is one piece, but it fits together with a more general body of evidence suggesting that antidepressants are really fuzzy in their effects, and might better be thought of as "antidistress" agents, or as having a general calming effect. Not mind-blowing, but does contradict the way the meds are typically portrayed in public discussion.

This issue has been a central one in psychology for decades, especially in clinical psychology. There are many variants of the problem, with lots of corollary problems.

Once central challenge is that even when you say you are interested in inferences about an individual, you usually actually evaluate your inferential strategy across individuals. This becomes problematic because it's easy to identify a serial killer post hoc; it's harder to avoid the inevitable avalanche of false positives if you apply this to thousands or tends of thousands of individuals regularly. In this way, you're not actually interested in one single individual; it becomes critical to be really clear about what your inferential population is, and what you're actually trying to generalize to.

Like you're pointing out, a lot of this too reduces to the significant challenge of knowing which predictive model to use, when you're effectively faced with thousands, if not an infinite range of models (one for each person/situation) to choose from. You might improve your prediction by using a more tailored model for an individual, but then you increase the risk of model selection error. Even if you have a lot of data on a person, they will probably change, circumstances will change, and so forth. The challenge is in knowing how to decide which model to use and when, when to decide that a particular predictive case is an exception.

This is sometimes known as the "broken leg" problem in clinical psych, so called because of a thought experiment in which you're tasked with predicting commuting behavior. You might have volumes of data on people, even individual people, but if you know that someone suddenly breaks their leg, or there is some other anomalous event, it compels you to logically alter your prediction. The trouble arises in knowing when to shift your prediction, when a scenario is different enough from that under which your default predictive model was developed. In the age of big data, you might be able to capture an increasing variety of scenarios, but there will always be cases that are not sufficiently captured, or where logic impels one to switch predictive strategies. But how do you do that?

I personally think what's undervalued is a focus on predictive uncertainty rather than the point value of the prediction. That is, realistically capturing and recognizing the true uncertainty in prediction for individuals, rather than improving the accuracy in prediction itself, which might actually have some fundamental limit.

I've sort of become convinced that with the right bike fit, crank length, foot position, and good pedals, there's probably not really any meaningful difference between the pedal options overall. I don't mean that to denigrate anything, just that they all probably will work well depending on the person, and I don't know that moving to clipless needs to be a rite of passage for anyone necessarily.

https://www.youtube.com/watch?v=CNedIJBZpgM

https://www.youtube.com/watch?v=FV0Fz2i2a8s

They're universally recommended as an entry to "serious" biking even though there's no scientific evidence they improve power or speed at all (except maybe marginally in accelerating from a dead stop).

For what it's worth, I have used and continue to use on occasion flats, clips, and clipless, and I prefer some kind of securing mechanism just because it's "set it and forget it" (I prefer clips actually).

However, somewhere in that first set of links (or in a link from one of the links) there's a long essay explaining that people feel less secure on flats because they are used optimally with a different foot position. That is, on clips and clipless you push with the front of your foot, and there's an argument that flats you should push with the middle of your foot. I've been trying this lately and it seems true to me. So some complaints about foot security might have to do with that; people trying the same foot position on both might be doing the wrong thing.

Also, a lot of money has gone into improving clipless pedals, and not so much on flats or clips, so there's some unfair comparisons there too.

The problems in the article amounts to fraud. Police that and let people decide what they want about what they're seeking.