skat20phys

It's difficult to go through everything with specific examples, because that would be an essay onto itself.

But, for example, in Book 3 (via the Rackham translation; http://www.perseus.tufts.edu/hopper/text?doc=Perseus%3Atext%...):

"If then whereas we wish for our end, the means to our end are matters of deliberation and choice, it follows that actions dealing with these means are done by choice, and voluntary. But the activities in which the virtues are exercised deal with means.

Therefore virtue also depends on ourselves. And so also does vice. For where we are free to act we are also free to refrain from acting, and where we are able to say No we are also able to say Yes; if therefore we are responsible for doing a thing when to do it is right, we are also responsible for not doing it when not to do it is wrong, and if we are responsible for rightly not doing a thing, we are also responsible for wrongly doing it.

But if it is in our power to do and to refrain from doing right and wrong, and if, as we saw,1 being good or bad is doing right or wrong, it consequently depends on us whether we are virtuous or vicious."

These types of passages assume or assert (1) free will, (2) the notion of responsibility for behavior as a consequence of this, and (3) a schema that is framed in terms of broad evaluative personal characteristics rather than specific decisions. If you believe that the notion of free will is specious, even partially in significant cases of vice, the entire Ethics starts to become questionable. To be even more specific: what if you believe that society (if not now due to scientific-technical limitations, then some day) has a moral obligation to address criminal wrongdoing by means of neurobehavioral intervention, to treat criminal predisposition as a disease? Is relying on an Aristotlian view of ethics as personal responsibility then unethical because it elevates an erroneous assumption of free will above the societal consequences of reducing suffering from criminal behavior?

This is taken out of context to some extent, as Aristotle also discusses free will, voluntary vs. nonvoluntary actions etc. But I'd argue that those discussions are kind of beside the point, and amount to logical loopholes of sorts in that they amount to something like "I'm not talking about cases where there isn't free will." But what if that is the main issue at some level?

The discussion of virtue ethics in general, to be contrasted with deontological or consequentialist ethical reasoning, for example, is a whole other topic, about which books have been written.

It's been awhile since I've read this. Rereading about it now, it seems like a conceptual mess.

I also have this sense, hard to put into words, that it is written from a perspective of socioeconomic privilege and implicitly assumes a certain freedom, or has as its focus certain concerns that reflect that. It comes across to me now as narrow-minded and lacking in understanding of the broader human condition and diversity of challenges affecting persons. Put another way, it seems to be written as if advising royalty or the wealthy is the implied goal, ignoring the broader experiences of mankind, which seems to me in turn ironically a moral failure.

Maybe related to this, it seems to presume a certain set of things that I see as actually being fundamental societal and philosophical questions, such as free will and personal agency and all that encompasses. Failure to attain well-being in a broad sense is a failure of the individual to practice free will, under The Ethics, and not of society to foster or intervene in a way so as to facilitate individual improvements in well-being. The sort of paradigm being asserted in The Ethics provides no way out to address the question of "how do I improve the lot of my fellow persons?" or to develop virtues in others. If someone rejects the notion of free will, either in itself or as a meta-phenomenon (that is, as something that can be manipulated itself), The Ethics seems misguided at best and pernicious at worst.

Virtue ethics also seem naive to me often, in that there's often little self-awareness of its limitations. For example, what if two virtues conflict? How do you resolve this? How do you interpret a behavior vis-a-vis its outcome in the presence of fundamental uncertainty?

My sense of the literature is that it's accurate in the limited sense that masks often don't have statistically significant effects in studies. It's maybe also accurate in the sense that the effects of masks are really being overrated, even many n95 masks.

However, in many studies there's a consistent trend toward masks doing something, in the sense that measures of dispersion are attenuated with masks vs not.

This paper is pretty poor in that it commits the classic statistical blunder of equating multiple nonsignificant results with overall effect size. It's very hastily put together.

One of the meta-analyses cited is an example of some of these issues (Offeddu, V. et al. 2017). You can see there's few studies and the ones included result in a nonsignificant effect, but it's almost significant and in the intuitively expected direction (PPE having an effect). But then there's issue of publication biases etc.

My general impression is that the truth is somewhere in between, maybe closer to what this essay is arguing in the sense that masks are much less effective than people think, but probably not actually technically correct in that masks probably have some effect.

In what direction? I think part of the point of this paper is that most masks aren't taking things seriously enough.

Sometime not too long ago there was a blog post or article covered on HN, about data showing student scores on some academic achievement measure over time. If I remember, all students generally improved on this measure throughout undergrad, and there was some average tendency for universities with higher average entering standardized test scores to have higher scores, but the effect seemed to be mostly driven by the students. That is, it wasn't that some schools accelerated scores more, it was that they just tended to select for higher-performing students. Also, the signal provided by school, if I remember, was relatively weak. The author was basically arguing that universities in general provide some educational value (or students just learn over time; there was no control of students who didn't go to college) but that the ranking was by and large just a signaling thing.

I can't find it now though because all the search terms I can think of just return a bunch of other stuff related to universities and students.

I also think the random deviation from "best" strategy isn't realistic.

In the real world, true randomness is kind of unusual. There's usually some kind of spatial or temporal autocorrelation.

So something might change in a quasi-random way, but it often doesn't just return to "normal" immediately. There's often something continuing (like an antagonist, or some environmental variable that decays over time, but slowly).

So these random "blips" do happen in reality, but they're not normal. What's more normal is for there to be changes, and for those changes to be somewhat, but not completely, stable with some decay. They might not decay at all.

At one time there was intrinsic cost to the lender in the form of risk. It seems not so anymore, or at least less than it should.

I'd argue that's in part because of the protection racket that's partially at the heart of the moral hazard: the poor regulation of healthcare services, at least in the US.

What this should be laying bare is the failure of federal regulation, how this got in the way at a critical moment, and when it was too late, things only got better when they got out of the way. Or what about a healthcare system oriented toward protecting the hospitals, those at the top of a hierarchy, the rent seeking, and so forth, rather than oriented toward public health? What about an artificial reduction in the number of providers to increase their salaries, a lack of competition in provider models and delivery of services?

There was a Slate article about how one side-effect of the pandemic is how much it has exposed bullshit everywhere, how easily so many rules are set aside when the cards are on the table. Some of this is happening now with healthcare, but few are asking what else could have been different, other than single payer care.

To me this all feels like some kind of abuser-victim relationship, where people are subject to some broken system, and then when things go wrong and are failed, they're made to be dependent on their broken system and start cheering it on.

I realize this all sounds bleak but to me there are so many things that could have been avoided, from the highest levels of the government all the way down, and there seems to be little introspection about this. Instead we just blame the virus, rather than what caused our systems to be so vulnerable in the first place.

One thing about case reporting is that they tend to ignore the denominator, the number of tests.

I've been tracking DPH reports in our state since they started, and one thing I've noticed is that although the case counts have gone up, the probability of testing positive has remained roughly constant. It's increased slightly here in the last week or so but not a lot.

One thing this suggests to me is that at least here, the virus was present in significant numbers before they started testing, or at least in similar numbers.

If you combine that with consistent reports of substantial fractions of asymptomatic carriers, it seems very plausible to me that the virus was circulating before authorities in the US started tracking it.

A colleague of mine pointed out it's important to look at the conflicts of interest and funding sources. One of the patients seems to be from a family funding the study, and some of the authors have filed patents on the protocol.

... without randomization or controls to randomize to.

The protocol is actually fairly intensive and expensive, involving tailored brain scans and all-day treatments over multiple days. It's the sort of thing that I could see in inpatients or partial hospitalization patients, or as an intensive outpatient protocol. You'd basically have to take off a week and go in every day all day.

This is promising but it's not randomized or controlled so hard to make anything of at some level (you can see how this kind of protocol would induce massive placebo effects).

If there is an effect, it may be that some kind of mild version of it, say, for a day or half-day or something might be useful in more mild forms of depression. But it's an intensive expensive treatment in its own way so might be hard to implement as an initial intervention.

But I agree that if it works this well it might be worth it to take a week off or a couple of days or something.

In the context of the press release it was a little confusing to me because the published paper doesn't really report on anything that would be considered long-term. I assume she meant something like "the less treatment resistant, the more pronounced the effect"?

Treatment resistance probably refers to prior treatments, which is a sort of indicator of severity or, well, resistance to treatment. So it makes sense to me that people who were less resistant to prior treatments would show more immediate longer-lasting effects.

But it's confusing.

This is impressive but it's not a randomized controlled trial at all, and it's a fairly select group of patients, so it's hard to know what to make of it. TMS treatment of depression has kind of been plagued with publication bias effects -- not to say there's no effect but more rigorous meta-analyses have suggested that publication bias is significant in the area.

Sometimes with very severe patients you see more regression to the mean, in that they have bigger apparent improvements in control conditions just because they can only stay the same or get better.

I don't mean to sound snarky, but aren't those two problems related?

I'm at the point where it shouldn't matter who did it, what should matter is its integrity. I'd prefer everything was anonymously posted at some level so author background didn't go into consideration regarding how it was received.

I agree with you completely, but the flip side of the coin is that experts can be blinded by assumptions that the field has. Sometimes outsiders aren't aware of these basic assumptions and so are less biased.

There's also the simple issue that sometimes expertise comes from places you least expect for reasons you might not anticipate.

For me there's as many problems in this pandemic related to appeals to authority (at least in the US) -- problems with testing related to FDA regulation and the CDC, problems with lack of healthcare providers due to long-term rent-seeking monopolies in licensing and practice scope, problems related to academic fraudulence and incentives (see: Didier Raoult) -- that I think it's dangerous to raise appeal to authority as anything but a bias.

For me there's multiple levels of problems to this, the first of which is the conspiracy and anti-science culture surrounding the pandemic. Above that is an appeal to medical and scientific expertise and authority that has sometimes been helpful but sometimes harmful. Above that still is an appeal to rigorous thinking and risk management, which transcends expertise boundaries.

Most of that is oriented toward yeast breads, but as a minor point that was confusing to me for awhile:

At least in the US, there is no technical legal definition for baking powder. So companies vary widely in what is included in baking powder. Ideally, baking powder would include a dry acid and base that react only in typical baking temperatures, but this is not the case. Some are very similar to baking soda in formulation, and others use different ingredients that require heat to react to any substantial extent.

There's an article somewhere I found that has a really great discussion of the issues, with history and reviews, but I can't find it. This is a sort of abridged version that smooths over some of the differences you'll run into:

http://www.thefreshloaf.com/node/12997/baking-powder

I personally don't believe them for a variety of reasons: I don't know that any country's numbers are entirely accurate, the substantial reports of active suppression of reporting about it (leaked communication, disappeared providers etc), and features of the reported data are suspicious.

On the other hand, there's probably an opportunity if not now than soon to start modeling the plausibility of Chinese reports. That is, a lot of them use China as a sort of baseline, but at some point things are going to get flipped on their head and the aggregate of other country's trends are going to start affording the ability to ask questions like "given what we know about the distribution of trajectories in other places, how plausible is it that data from country X is valid?"

There's complications due to heterogeneity in population networks, healthcare, response, etc. but I think quickly we're going to start seeing things being sorted out empirically. You can only hide things for so long.

There have been studies of this from previous pandemics. Together with studies from the Great Recession / financial collapse the gist is there's definitely going to be an increase in mental health problems of all those sorts -- substance use, suicide, etc.

The interesting thing is that in previous situations deaths attributable to these things was offset by a decrease in deaths due to automobile accidents, workplace stress-related incidents, etc. That doesn't make it ok at all, and it's possible this time around is worse socioeconomically so things aren't offset. But it might not show up in overall trends if you just looked at general disability or death.

This is nice but I'd also like some kind of detached evaluation of how well they function in practice, in terms of pros and cons. My sense is that in the videochat/conferencing area there's a big difference between feature offering and how well those features actually work at any given time.

This is great but we really need antibody tests almost as much, if not as much.

Vulnerable world, meet modern academics.

Is there actual good data on aspirin/NSAIDs? Because what I've read was a theory communicated in a letter to the Lancet, and the actual data might suggest the opposite.

A discussion of how regulatory structures contributed to problems in the US (lack of tests, number and variety of providers and services, and so forth).

I have a lot of respect for Ioannidis's work, and kind of lean toward his perspective in some respects, but I think he's misrepresenting certain data and missing certain points.

He's right to point to the Diamond Princess Cruise ship for example, but his statistics are already out of date. As of yesterday, 8/712 infected had died; if you expand this to the Grand Princesss, it/s 10/800, which is at 1.1-1.25% CFR. This is a lot bigger than the statistics Ioannadis cites, and greater than the flu. There are some issues about how to extrapolate the cruise ship population to other populations, but that isn't straightforward (cruise ship passengers are probably older on average, but also might be healthier and/or wealthier). In any event, that number can only increase.

The other issue is that regardless of CFR in the population, the experiences of various locations points to the threat. That is, we can argue with Ioannadis all he wants about CFR and so forth, but in the end, if places are crumbling under the weight of hospital visits, it doesn't matter.

As someone else pointed out, fatality rates are only part of the picture, and in this case it's important. At some level, it's not the the fatalities per se that are the sole problem, it's the much greater percent of cases that require hospital care. It's not like CFR is a fixed statistic that applies everywhere all the time anyway; it will depend on resources and other things.

Finally, there's been a lot of discussion by Ioannadis and others about costs of restrictions and so forth but little data about that that's pointed to (even though it's available). For example, if you look into data about the Great Depression and Great Recession, it appears that overall there was a decrease in mortality that followed. Data on health effects of recessions is actually pretty clear: although there tends to be an increase in suicide and mental-health related deaths (e.g., drug overdoses), there's a much bigger decrease in deaths due to things like motor vehicle accidents, cardiovascular events, work-related deaths, and so forth. So this benefit-cost mortality analysis, as morbid as it is, doesn't necessarily play out in the way Ioannadis assumes.

I've been there and have mixed feelings about this.

On the one hand, the boulder has always seemed fishy to me. It's a remote location and I doubt many people would track something that small, or care enough about it to know about the seasonal changes at that boulder. It also doesn't make a lot of sense to me geologically, although it's possible. I also years ago remember people familiar with the site jokingly talking about how "if there were a boulder there..." which lent the whole thing a sense of fiction when I started reading about it. It was easy to see how a humorous fantasy could change into folk legend and then into documentation.

On the other hand, where are the photos? Where's the meticulous documentation? To trek all the way there to address this and then not take copious photos to share with the media?

They also got lost, which adds to my skepticism they even made it there, or were sure they did. Can you get lost there? I suppose -- there are a ton of islands and things there so I can see that -- but the area is blanketed with topo map coverage, and you've got GPS now. If someone were to get lost there, I'd question whether or not they even got to where they wanted to begin with. Siskiwit Lake is a fairly large feature on Isle Royale, and the nesting doll nature of the goal makes your target hard to miss.

In short, I wouldn't be surprised if the boulder isn't there (or isn't surrounded by water). However, I also wouldn't be surprised if this guy and his mother never actually went to where they claimed to go.

I've always really liked Rhatigan's (that is, the author of the linked piece) Gina typeface:

http://www.ultrasparky.org/work/typefaces/gina/gina_01.html

I read something about it evolving into an internally developed project somewhere but never saw it again. I really wish it were released publicly.

Those estimates are generally in line with what other independent investigators find, maybe a bit smaller.

The elephant in the room in these kinds of discussions is the validity of the tests to begin with. MIT makes a statement that the general tests are predictive in combination with other criteria, but how predictive?

Generally studies of these sorts of things find that they're moderately predictive of first-year GPA (like .4 correlation), and then trail off to zero as the interval between testing and outcomes increases. The effects are even smaller in studies in relatively unselected samples (due to court orders or legal decisions, for example).

So one argument always goes that you should use whatever empirically-supported stuff you can to make fair decisions, but we treat them as if they're more than they are. Sure, you could have a lot of things that are significantly predictive but with small effect size, or where there's lots of noise, but why as a society to we pretend these are huge effects?

The other thing about that paper is the hint that the practice effects are larger in higher-performing examinees, which also makes sense and is consistent with other studies. Why is this a problem? Because those are exactly the types of students for whom these issues are more applicable. A 12 point average gain with practice doesn't matter if you're including people who never had a chance at MIT anyway, but a 20 point gain might in a highly competitive group where small differences are being magnified tenfold.

The issue in all of this isn't the people in the 99th percentile versus the 50th percentile, which is the bulk of what's going into these predictive models and effect sizes, it's the 99th versus the 95th. There's a ton of real-world noise, but society acts as if the noise is nonexistent. It's like we're idolizing the outcome of some kind of survivorship bias process.

I agree with you regarding client-facing staff, and think the parent post has misplaced lack of concern. However, I also think the post makes an important underlying point about hospital administration generally and "where their heads were at."

Increasingly a lot of hospital administration is very profit-focused. Not all but maybe most; where it's not, it's increasing, and where it's not supposed to be it often is, but hidden.

What this means is that the focus of hospital administration is often not on epidemiology and public health, even though it should be. It should include that more. It's focused more on squeezing out the maximum from each provider in terms of billable hours and patient contacts. When you're focused on stuff like that, you're not thinking about public health.

And so we have the consequence of that now. I don't think this is the only problem -- there's also government overregulation, rent-seeking provider monopolies that artificially restrict supply of services, and lack of coverage for patients -- but it's part of it. In general, at an abstract level, healthcare in the US has become very focused on protecting the interests of the established system, and less so on the interests of citizens and patients. It's a house of cards with several problems at once, getting hit by the perfect storm.

There are causal accounts that use mutual information and its variants though. A lot of them introduce notions of directed mutual information and other similar concepts.

This an aside, but this seems like a textbook example of government regulation of healthcare not only failing but getting in the way and actively causing harm. There are literally 1000s of generally certified labs that they could have said very early on "go to it if you have this certification" and instead they plodded through this top-down heavy handed mess and made this pandemic many times worse.

In the US you have this combination of heavy over-regulation and lack of public provisioning. It's like the worst of both combined. Between this, cannabis, ketamine, and the opioid crisis, I absolutely do not understand why there isn't an uproar over overregulation in healthcare getting in the way.

Not exactly what you mean but operacast lists opera streaming broadcasts from around the world:

operacast.com

It's pretty old-school but I've found the schedule pretty up-to-date. I've also noticed that some of the programs aren't opera per se but "opera adjacent", like choral music or instrumental music by opera composers. My impression is that there are some programs that regularly stream operas but at any given week might offer something else.

My sense is that there's a lot of opportunity for a well-done classical streaming service. A lot of the traditional channels don't quite work because their catalogue isn't deep enough or their metadata structures are weird and unhelpful (for example, it's impossible to sort by composer alone).