Jon Zelner, Julien Riou, Ruth Etzioni, and I write:
Just as war makes every citizen into an amateur geographer and tactician, a pandemic makes epidemiologists of us all. Instead of maps with colored pins, we have charts of exposure and death counts; people on the street argue about infection fatality rates and herd immunity the way they might have debated wartime strategies and alliances in the past. The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic has brought statistics and uncertainty assessment into public discourse to an extent rarely seen except in election season and the occasional billion-dollar lottery jackpot. In this paper, we reflect on our role as statisticians and epidemiologists and lay out some of the challenges that arise in measuring and communicating our uncertainty about the behavior of a never-before-seen infectious disease. We look at the problem from multiple directions, including the challenges of estimating the case fatality rate (i.e., proportion of individuals who will die from the disease), the rate of transmission from person to person, and even the number of cases circulating in the population at any time. We advocate for an approach that is more transparent about the limitations of statistical and mathematical models as representations of reality and suggest some ways to ensure better representation and communication of uncertainty in future public health emergencies.
We discuss several issues of statistical design, data collection, analysis, communication, and decision-making that have arisen in recent and ongoing coronavirus studies, focusing on tools for assessment and propagation of uncertainty. This paper does not purport to be a comprehensive survey of the research literature; rather, we use examples to illustrate statistical points that we think are important.
Here are the sections of the paper:
Statistics and uncertainty
Data and measurement quality
Design of clinical trials for treatments and vaccines
Disease transmission models
Multilevel statistical modeling
Communication
Information aggregation and decision-making
New issues keep coming up and they didn’t all make it into the article; for example we didn’t get into the confusions arising from aggregation bias. There’s always more to be said.
Another place we’re seeing so much discussion of statistics and uncertainty, at least at the ethernet water cooler if not necessarily the corner store, is with the discussion of the efficacy of NPIs.
My take as an official armchair epidemiologist (and statistician) is that in general, many people waaaaaay, way, way, way over-estimate their ability to assess the efficacy of interventions in a context with so many vast and highly complex confounding variables – in particular when they think they can reverse engineer and wrangle efficacy by comparing the outcomes across different countries.
Looking at the issue of climate change I was quite convinced that motivated reasoning dominates how most people reason about complex, highly uncertain questions when they overlap with ideological orientation. But imo, the confidence displayed by so many people regarding risk assessment in the face of uncertainty vis a vis the impact of NPIs (not to mention the efficacy of masks) actually makes that dynamic even more apparent. The number and intensity of armchair epidemiologists seems considerably larger than what I’ve see with respect to armchair climate scientists.
It is pretty obvious to people that stuff like wearing a (poorly fit) mask when you enter a restaurant then taking it off when you sit down isn’t going to stop the spread of aerosols.
It is sad we actually needed to do studies to show this to the EBM folks.
Anoneuoid –
> It is pretty obvious to people that stuff like wearing a (poorly fit) mask when you enter a restaurant then taking it off when you sit down isn’t going to stop the spread of aerosols.
Precisely my point!
People think they’re in a position to assess the efficacy of masks whwn they don’t even know the basic predictive variables such as the size of infectious particles, the affect of the size gradient on infectiousness, the dose needed to create an “infection,” the dose-dependent dynamics, the interaction effect with air filtration, spacing, length of time, and so many other critical variables such as rate of community spread in the particular context.
And despite such a wall of ignorance, they think that they can draw “obvious” conclusions from reading Twitter or headlines or watching TV.
No need to study the relevant literature (which makes the uncertainties apparent), or learn from conducting field experiments, or performing careful statistical analyses.
And of course, just we can predict what they will find to be “obvious” in most cases merely by looking at their political orientation.
But that’s just a coincidence, of course.
Just like people think they know what’s “obvious” about the impact of CO2 on our climate.
You have to deny 100 years of science to think wearing a (not even good) mask as you enter a room then taking it off has any meaningful effect on aerosols.
Yet that is what people are being told to do.
Anoneuoid –
What I was referring to is the question of whether there’s any marginal utility to wearing masks as a general default behavior when a lot of people interact in confined spaces with sub-optimal ventilation. An individual even marginal utility that would well then compound at a population level.
Of course you could select a constructed scenario to make a marginal utility less – such as if someone has a poorly-fitting cloth mask that they take off as they’re eating in a restaurant. But again, this reinforces the point that I’m making. People cull through the relevant parameters and questions and select those which serve some kind of an agenda as a rallying cry.
I see that a lot, whether from people who confidently say that “masks don’t work,” or those who are likely over-evaluating their efficacy across varied contexts.
I’ve looked at the relevant “100 years” of literature – both epidemiological and experimental analyses – to a modest degree, and I think enough to conclude that there’s a lot of uncertainty. Certainly enough to leave open the door for marginal benefit to mask-wearing as a general practice.
I’ve also seen a rather constant refrain that “100 years of scientific evidence shows that masks don’t work,” and from my modest review I’ve certainly seen enough to know that anyone who says that hasn’t actually looked at the literature. That’s the kind of unfounded certainty I”m talking about. I do realize you didn’t exactly say that – and instead you skewed the question to be one that’s relatively meaningless.
Yes, if people act in a way that any potential marginal utility would be minimized, then the potential marginal utility would be minimized. I agree with you there and would imagine that “100 years of science” would back that up.
Lord forbid it, but if “masks worked”, the little goateed rabble wouldn’t be able to bare their rabble canine teeth and what’s life worth if you cannot cannot strut about wearing the *complete* regalia of aggressive display?
So time of exposure to a certain density of viral particles doesn’t impact your risk of becoming infected?
Here in CA, BTW, the mask mandate in restaurants required keeping them on not simply while entering, but while waiting for service, ordering, and waiting for your meal. Combined with tables being separated and recommendations regarding ventilation. And of course staff masked at all times.
Literally everything affects/impacts everything else. There is no reason to ever ask such a question, although sometimes that is used as shorthand for an actual question.
So I assume what you mean is “does wearing a mask (that isn’t blocking aerosols anyway, it just concentrates them around your head rather than spraying them forward) for a small % of your time in the room meaningfully impact the exposure and spread of aerosols?”
The answer is quite obviously “no”.
Anoneuoid –
> “does wearing a mask (that isn’t blocking aerosols anyway, it just concentrates them around your head rather than spraying them forward) for a small % of your time in the room meaningfully impact the exposure and spread of aerosols?”
>> The answer is quite obviously “no”.
Again making my point, where reaching conclusions of what’s “obvious” is made despite that the very distinction between “aerosols” and “droplets” and what sizes of each are infections is unknown/effectively arbitrary, or how that uncertainty interacts with the differences in mask attributes, or other context specific variables such as humidity or ventilation or time of exposure or air currents.
Also note that even ignoring all those important factors, “100 years of science” including experimental study, shows a range of results.
For infectious disease, an aerosol is just a particle that diffuses around the room and can float there for hours. If the HVAC system is recycling the air without filtering/purifying it then it is even spraying them at everyone. This is why we are seeing the seasonal pattern in covid infections where it is worse in hotter areas during the summer and colder areas during the winter.*
Wearing a mask when you enter a room then taking it off is not going to have a meaningful effect. Asking people to perform this ritual while you wait to figure this out is not helping the credibility of medical experts.
* This cycle also interacts with the approximately yearly population immunity cycle (driven by waning mucosal immunity after infection). I would guess eventually these cycles synchronize and covid starts to be primarily a winter phenomenon since that was the biggest peak. But we will see on that point.
Anoneuoid –
Last comment.
I’ve looked a bit at what’s out there and talked to a very small sample of people modeling, and measuring this experimentally. Here’s my tentative conclusion.
Even the range of our very ability to measure “aerosols,” let alone over a variety of conditions (such as humidity), volume, time spans, activity levels affecting airflow, (both behind and outside of masks) etc., doesn’t support a conclusion of what’s “obvious.” On top of that rests uncertainty as to our ability to then know what are the “obvious” implications of those measurement.
And you’re saying the question of the efficacy of masks to have a marginal benefit should be “obvious” to someone observing some restaurant patrons?
There’s evidence, much of it conflicting, that helps to inform us of probabilities.
Make of that what you will.
Nope.
I am saying that if you tell people they have to do stuff like “wear a mask as you enter a room then take it off when you sit down to stop the spread of an aerosol”, they will eventually stop believing anything you say.
Saying you are uncertain about whether that works or not, but telling people they have to do it anyway, looks even worse!
Even an ill-fitting cloth mask filters out some aerosols. And if the servers at a restaurant are wearing good masks, they are reducing their chance of getting COVID by a meaningful amount, and thus reducing their chance of passing it along to others (including customers).
I haven’t previously seen the claim that Anoneuoid attributes to others, that if you wear your mask until you’re seated but then remove it and leave it off then you’re reducing the chance of getting or transmitting COVID. I’m not saying nobody is making that claim, just that I haven’t seen it myself. As Dhogaza points out, here in CA we are told that you should only remove your mask while in the act of eating or drinking, and should wear it while waiting for your food, waiting to pay, etc. If people actually followed this advice it would result in people wearing masks in a restaurant about half the time, which would surely reduce risks somewhat.
The goal is to get R0 below 1 and keep it there. If there are a bunch of actions that are individually barely effective, but that collectively take you from R0 = 1.2 to R0 = 0.98, each of these actions can be worthwhile.
I didn’t realize people still think a surgical/cloth mask protects you from breathing in the virus (since the air you breath comes around the edges this makes no sense), instead the claim was that it protects others when you breath out. This study shows the effect of a mask on breathing out aerosols is negligible (as expected):
https://aip.scitation.org/doi/10.1063/5.0057100
A simple air purifier was superior to even a well-fit N95 mask in that experiment.
I traveled across the US from lake superior to the gulf of mexico last year and this was the case everywhere they cared about masks. It is now the case again where I live.
No one ever really claimed this works afaik, that is just the rule being enforced.
Also, R0 changes seasonally, thinking about it that way can (and has) lead to much mischief.
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1634916/
Actually, another related point is that the current batch of vaccines are not going to have much (if any) effect on transmission since they are IM injections that don’t trigger mucosal immunity. They only induce humoral immunity vs the fastest mutating region of 1 out of 29 viral proteins.
Indeed, the first evidence for this just got published from Israel. Vaccination appears to be 13x less protective against infection and 27x less protective against symptoms than prior infection (keep in mind this is using the same “real world” methods used to claim vaccine effectiveness):
https://www.medrxiv.org/content/10.1101/2021.08.24.21262415v1
But even immunity due to prior infection is not going to last very long, since that has never been observed for viruses that primarily replicate in the mucosa.
https://pubmed.ncbi.nlm.nih.gov/33901246/
Anon,
You’re very fond of setting up straw men and then knocking them down.
The mask-effectiveness study you cited says cloth masks and surgical masks filter out about 10% of particles near 1 micron diameter. You say this is ‘negligible’ but in the context of the pandemic I do not think that is negligible.
But the bigger issue is that I said that if restaurant servers are wearing “good masks” they are reducing their chance of getting and transmitting COVID. You responded by saying no, cloth masks are of negligible effectiveness. OK, I disagree (see above) but what about good masks?
In this thread as in many others I feel like your first instinct is to disagree with everyone about everything, whether it’s true or not and maybe even whether you agree with it or not. Sometimes you’re right and sometimes you’re wrong, but the long-term impact is to make many of us discount everything you say. Obviously that’s not ideal, since a lot of what you say is right. But it’s just too exhausting to try to separate the wheat from the chaff all the time.
What do you mean by “good mask” then? I just didn’t think it possible that restaurant staff is properly wearing N95 or similar masks. I really find that quite implausible. Maybe they put them on for a moment to bring the food then go take them off, etc.
In that study a poorly fitted KN95 mask was only 3.4% effective at reducing aerosols, even less than the surgical/cloth masks. Probably because it takes more pressure to push through the mask so even more of the breath goes out the sides.
KN95 and N95 are both supposed to filter on inhale not exhale. On inhale these masks should get at least 80% of particles bigger than 1 micron when worn even approximately correctly (like with clean shaven face at least)
But yeah I would expect them to just squeeze the exhaled air around the sides. This too is beneficial because it reduces exposure to a person standing in front of you talking at least short term.
However the basic point that cloth masking is unlikely to be nearly as effective as people seem to act like it is is also correct. Sending kids to school in crowded hallways and repeating the mantra “wear a mask” is imho irresponsible for example. Particularly dumb if you’re letting them go to band class and play trumpet in an indoor auditorium or stuff like that.
Anoneuoid and Phil –
I’m going to cautiously weigh back in here:
> I do not think that is negligible.
I agree. A marginal benefit from mask-wearing as identified in that study, as Phil alludes to, likely compounds at the population level. As to whether that is “negligible” is a matter of perspective, not fact.
But further, it’s just one study and other studies involving lab-based experimentation have found more significant benefit from wearing cloth masks. That’s not to weigh-in on the relative quality of the studies with different findings as I’m not qualified to do so. Just to say that from where I sit, there’s still uncertainty despite that study.
Further, I look at the issue in general and think that Anoneoid isn’t practicing good science by going from that one study to concluding that no meaningful benefit is “obvious.”
Measuring aerosols as reflective of the real world seems to me to be extremely complex. As I assume you both know, aerosols change size and in other ways (over time (such as drying out), and aerosols of different size behave differently, and so I’d imagine measuring or modeling their behavior is extremely complex. Then you have the complicating factors such as the ambient air movements, variation in filtration systems, variations in the background rate of spread, variations in the does-dependant effects of exposure to droplets, variations in how individuals breathe, variations in how susceptible different individuals are to exposure, variations in how virulent or infectious different strains of COVID are, variations in how much ambient humidity there is, etc., etc.,
And I want to limit personalizing the discussion, thus, again, I think that Anoneuoid is providing an example of exactly the phenomenon of “armchair” expertise where people look at partial information and then exaggerate conclusions to state with total certainty what’s “obvious.”
I”m NOT saying that Anoneuoid is a typical armchair expert. He’s clearly quite knowledgable and skilled in a variety of ways that are important to assessing these issues. I’d also say he’s more careful in some respects than most armchair experts I’ve run across.
But that doesn’t inoculate him from the phenomenon I’m describing. Indeed, that phenomenon, a variant of motivated reasoning is a basic feature of human cognition and psychology.
Again considering the importance of caveating single studies such as the one linked above.
A paper that reports different findings than those above, on the value of vaccines vs. “natural” infection. I have no idea about the relative quality of the competing findings (Berenson criticizes the generalizability of this study because it was conducted with younger, healthy people, and in that he’d have a point but typically he runs with a take that pushes his agenda).
Anyway, I’d be willing to bet there’s still a lot of uncertainty
In this study, we longitudinally profiled both antibody and cellular immune responses
in SARS-CoV-2 naïve and recovered individuals from pre-vaccine baseline to 6 months postmRNA vaccination.
https://t.co/xYlrV3M7fs?amp=1
A twitter thread discussing the findings.
https://twitter.com/celinegounder/status/1430661421839302674?s=20
Oy. “thus, again…” should be “but, again.”
It looks like a Freudian slip the way it is, but actually a product of bad editing on my phone…(which I’ll conveniently argue could never be explained by a Freudian slip).
More on that preprint “published” about Israel and “natural” vs. vaccine-induced immunity that was linked above:
https://twitter.com/walidgellad/status/1423567055463276550?s=19
The one study rule should always apply,PARTICULARLY with preprints
Yea, the peanut gallery comes out with this when they don’t like the results but stays silent when they like them. This is standard with peer review in general.
Anyone who is intellectually honest would apply the same criticisms to all the “real world” studies showing how great the vaccines are and dismiss those too on the same basis.
I don’t trust any of it myself because, first of all, the study needs to account for the reason for a test. There are many more flaws that could affect the estimates as well, but that is the most basic.
However, ~100 years of science tells us humoral immunity vs only one protein is going to be far weaker than mucosal + humoral vs all 29 proteins. There is no excuse for not realizing this a year ago amongst the “experts”.
Joshua,
I must have missed that you were an official epidemiologist, although I recognized your very insightful questions/responses. I think that medical education is quite limited in some respects. I believe that has been acknowledged by physicians/surgeons themselves, prompting a movement to explore alternate lifestyle interventions, as have Drs. Dean Ornish, Neal Barnard, Caldwell Esselstyn, John McDougall have demonstrated through their clinical trials. I am aware that there are legitimate critiques of nutritional research. And research improvements are always warranted.
I am not particularly bothered by all types of experts and non-experts weighing in b/c after all who is the recipient of these interventions? Why shouldn’t the public, even in non-expert, weigh in?
The competition for grants and idea markets is so intense that there is a tendency to inflate one’s competencies. I concentrate on their reasoning mostly. I admit that there are some grumpy and pompous experts. But I tend to ignore their idiosyncratic flairs.
Sameera –
I don’t think there’s a problem per se with all sorts of people weighing in.
The “problem,” as it were for me, is that people weigh in with great certainty, without considering conditional probability or the importance of control for confounding variables, or understanding important statistical phenomena such as base rate or Simpson’s paradox as discussed in these pages.
The net effect is that as a culture we can’t engage in meaningful stakeholder dialog to assess risk in the face of uncertainty, but instead dig into leveraging problems like COVID to satisfy tribe-based antipathy
Sameera –
Just so you know, my armchair has been officially certified as well.
There is also the prevalence of the culture we live in: one that is argumentative and litigious. Moreover, I am not a big fan of ‘devils advocacy’ which leads to plethora of dichotomies, many of which are amateurishly drawn. Not to mention the equally lavish use strawman argumentation.
> The number and intensity of armchair epidemiologists seems considerably larger than what I’ve see with respect to armchair climate scientists.
It seems like the professional epidemiologists are just (in general) much more obviously incompetent at statistics compared to climate scientists.
Matty –
> It seems like the professional epidemiologists are just (in general) much more obviously incompetent at statistics compared to climate scientists.
I think that’s unfair. For the most part, professional epidemiologists put out work that includes a fairly careful accounting for control for variables, including an attempt to deal directly with, and quantify uncertainty. And in my experience, many professional epidemiologists work with professional statisticians, although then there are the related shortcomings of professional statisticians.
Of course, some of them get that quantification wrong. And no doubt, some border on, or pretty much flat out ignore, uncertainties.
But in contrast, a pretty high percentage of armchair epidemiologists, in my experience, either don’t understand the uncertainties or just flat out ignore them.
Ok, maybe I should weaken my statement a bit and restrict it to vocal professional epidemiologists. It is certainly possible that there are professional epidemiologists who understand the shortcomings of modelling, don’t make overconfident predictions/policy suggestions, and who do valuable work to understand the spread of disease. I do not want to criticise them. The epidemiology I am exposed may be a small sample of a large field, etc…
But also: there has been no shortage of bad predictions coming from high-profile epidemiologists. I’m thinking of e.g. the Imperial College group whose predictions for the first wave failed to materialize, and whose later papers (such as the one by Flaxman et. al that was recently discussed on this blog) were not much better. Accounting for variables and uncertainty quantification don’t help you if your model is some variant of SIR.
I have recent draft of a blog post here https://categoricalobservations.com/2021/08/25/on-some-nonsense-from-mishra-et-al/ where I look at a a particularly appalling example that appeared recently. Philippe Lemoine, whose work I first saw here also has some rather damning criticisms of epidemiologists at https://cspicenter.org/blog/waronscience/the-british-variant-of-sars-cov-2-and-the-poverty-of-epidemiology/ and the ICL group in particular at https://necpluribusimpar.net/lockdowns-science-and-voodoo-magic/ . If it’s obvious to anyone with a data science background that these high profile epidemiologists are really incompetent, then it’s hard to fault the armchair epidemiologists who think they can do better.
I disagree, Imperial College predictions were absolutely fine, because they were predictions predicated on a certain kind of inaction. That the inaction didn’t happen was the whole point of the prediction. “We can’t ignore this because look what will happen if we do”
> I disagree, Imperial College predictions were absolutely fine, because they were predictions predicated on a certain kind of inaction.
Yeah this is a common claim, but but you read the actual paper you’ll notice it presents several scenarios, only one of which is inaction. You can also run the ICL models e.g. for Sweden which implemented this kind of policy inaction, and the result is that the ICL models are not predictive at all: https://twitter.com/phl43/status/1263497666144669696?lang=en
Here is the ICL paper btw: https://www.imperial.ac.uk/media/imperial-college/medicine/sph/ide/gida-fellowships/Imperial-College-COVID19-NPI-modelling-16-03-2020.pdf . Under their most-action scenario, critical care beds peak at around 80/100k (Figure 2), if I interpret the google data correctly, *hospitalization* peaked at around 22/100k in the UK so you can see just how badly off they are. Of course, this kind of thing is very hard to predict, there is much uncertainty involved, I myself didn’t do better in March 2020. But at the same time it isn’t possible to come away from this thinking that epidemiologists are able to accurately predict disease spread.
The idea that sweden represents inaction is also commonly held, but is nothing like true. Sweden represents individual but not strongly government coordinated action.
I see the Imperial College predictions as essentially qualitative. Like “if we do nothing, we will have a very high peak within a few months and a lot of people will get sick and die, hospitals will be overwhelmed”. and “If we try to do something fairly limited, it will have a fairly limited effect on spread” and “the only thing that’s going to work is fairly drastic action including basically everyone social distancing for a full year and lots of closures, and then we’ll still have to just put up with a lot of death and illness”…. Turns out that was correct!
What we wound up having to do was something fairly drastic… months and months and months of people not doing stuff they used to do, wearing masks everywhere, closing restaurants, etc. And we still got … around 1/10 of the death rate peaks predicted by that model in the UK. Also, the dynamics of continuous growth and then triggering people to stay home, and decay, and then growth, and then triggering people to stay home and decay, that has also played out.
The problem is people insisted that the scales on the y axes of the graphs should be exact at a time when people didn’t have enough information to even guess the case fatality rate to within a factor of 10, rather than that the relative height and width of peaks be basically correct under different scenarios.
As Richard Hamming said “The purpose of computing is insight, not numbers”.
If you just erase the y axes of their graphs, you’d make the correct decision, which is to implement a lot of social distancing and try to keep the population as safe as possible. Remember that at the time the alternative was politicians saying essentially that they believed they should encourage people to go out and have COVID parties and just get sick as quickly as possible to minimize the amount of time that the economy was disrupted.
Replying to Daniel Lakeland:
> Sweden represents individual but not strongly government coordinated action.
I agree. However, I’d argue that this kind of individual action should have been accounted for by the ICL group.
> Like “if we do nothing
Who is “we” here? I feel like there’s a bit of a motte-and-bailey going on here, where the bailey is “strong government action is needed so as to not be overwhelmed by the pandemic” and the motte is “individuals need to change their behaviour in order not be overwhelmed by the pandemic”. Do read the ICL predictions; they focus *only* on government actions. If Sweden gets by without heavy handed government action then this *does* leave the ICL group looking rather incompetent.
> and then we’ll still have to just put up with a lot of death and illness”…. Turns out that was correct!
No, if you get “a lot of death and illness” wrong by an order of magnitude then you are *not* correct in my book. Moreover, places without the “drastic action” didn’t collapse completely.
>Also, the dynamics of continuous growth and then triggering people to stay home, and decay, and then growth, and then triggering people to stay home and decay, that has also played out.
Did they though? Do look at Figure 4. Then look at the cases/hospitalizations/deaths for the UK. Then look at Figure 4. Do those look the same to you?
> The problem is people insisted that the scales on the y axes of the graphs should be exact at a time when people didn’t have enough information to even guess the case fatality rate to within a factor of 10, rather than that the relative height and width of peaks be basically correct under different scenarios.
No, you don’t understand my point. The point isn’t “epidemiologists have to be exact under uncertainty” but “epidemiologists have to be open about it when they don’t know things with certainty”. Also “basically correct”? Are you just saying “they predicted that sometimes disease would increase and sometimes it would decrease, and look it did *exactly* that”.
> If you just erase the y axes of their graphs, you’d make the correct decision, which is to implement a lot of social distancing and try to keep the population as safe as possible
Ok so first of all the jury is still out there as to what the “correct” decision is; but all of the cost-benefit analyses I’ve seen (including really silly ones like https://astralcodexten.substack.com/p/lockdown-effectiveness-much-more) point towards it at least being unclear whether lockdowns pass. Y-axis scaling matters a lot! Y-axis scaling is why we don’t lockdown for the flu every year. I can’t believe you’re saying that the *only* important variable here – y-axis scaling of cases/deaths – is not important.
Of course y scaling is important. If they were off by a factor of 10000 it’d be incredibly bad. But at a time when some “experts” were saying things like 0.01% IFR and maybe 10,000 deaths (Ioannidis in US) and others were saying 1-10% IFR (WHO), to have reality come in within a factor of 10 of a prediction for a plausible scenario that isn’t what actually happened is very good, a factor of 100x better than choosing a random expert.
You’re insisting on their “predictions” being “prediction about what will actually happen”. They aren’t they’re “predictions about what would happen in a very specific scenario that we know ahead of time won’t play out, but is still relevant to the relative order of magnitude of what is likely to occur”. Those scenarios involve people doing specific things. They aren’t predictions about “how effective will lockdown be vs private individual action” they’re predictions about “how effective will this level of reduction of individual contacts be” period.
when I say: “If Joe punches Robert he will break Roberts nose…” and then Joe doesn’t punch Robert, and Robert doesn’t get a broken nose… That is not a refutation of the prediction. The only way you can refute their prediction in my opinion, would be to show some place where contacts were not reduced substantially and yet illness and death was substantially more than 100x smaller than their predictions. The factors of 10-20x were just implicit in the fact that experts couldn’t even agree to within a factor of 1000x
I read Imperial College as the following:
We know this disease will cause a LOT of suffering and death, definitely more than a normal Flu (this was a controversial assumption at the time, but turned out to be correct), we don’t know how much more, but we can see from this set of scenarios that in order to get significant reductions in death and suffering, we will need to implement substantially more intervention than what is currently being discussed, which is basically “COVID parties”.
They were 100% correct about that.
The part about whether it needs to be Govt Driven or Individual Driven is not identifiable from their model. Their models are not about “govt driven mandates” they’re about “people actually reducing their contacts” by whatever means.
You can argue that you don’t like their political slant on the wording, but you really can’t argue with them that somehow if people had just kept doing whatever and not drastically cut their close contacts back that we would have been fine and had less than Flu level of illness and death.
If your argument again is that Figure 4 doesn’t look like https://ourworldindata.org/explorers/coronavirus-data-explorer?zoomToSelection=true&time=2020-03-01..latest&facet=none&pickerSort=asc&pickerMetric=location&Metric=Confirmed+cases&Interval=7-day+rolling+average&Relative+to+Population=true&Align+outbreaks=false&country=~GBR
Then that’s because we didn’t implement some kind of strict feedback control system such as the one they were using as an example. Duh. But there was some kind of feedback control, it was just ad-hoc, and you know what? It resulted in booms and busts.
Claiming that they have to predict how people will actually act is just ridiculous. The best we could ever ask for is to have their predictions be consistent with the reality of what *would have happened* if the scenario assumptions had played out, and to within a factor of 10 to 20 on overall y axis scaling, but within a factor of 1.5-2 on relative scaling (size of peak under scenario A divided by size of peak under scenario B).
Another way you could refute their predictions would be to show some place where they started out with one kind of level of response, and switched to another level of response, and then the infections didn’t do anything like what the Imperial College model predicted (in terms of growth rates, or peak heights or etc)… And this is where you could really get a good refutation, because I’d accept just *relative* changes out of whack by factors of 2 or so.
But note that by “level of response” what I mean is “actual interpersonal contact levels” not “which on paper policies did the govt enact”
I did look at the graph for Sweden you linked to. I don’t see it as particularly interesting really. What would have to be done is to do a Bayesian model fit, find the high probability set of parameters that predict what actually happeend for Sweden, and then show how those parameters are objectively very far from what actually occurred in sweden (based on measurements related to the parameters, such as google mobility data or surveys of activity levels or etc).
>to have reality come in within a factor of 10 of a prediction for a plausible scenario that isn’t what actually happened is very good, a factor of 100x better than choosing a random expert.
Did they come within a factor of 10? I’m not convinced they did.
>You’re insisting on their “predictions” being “prediction about what will actually happen”
Read the paper. They have multiple predictions for multiple scenarios. All of them are (in retrospect) overblown. Look at Figure 2. The red line was, as far as I know, never reached. If you make a number of conditional predictions, and all of them are an order of magnitude worse than what happens then you’ve (probably) made the wrong conditional predictions. Imagine I go to the government and say “this cantine serves contaminated food because their water supply has lead. If they serve soup, 100 people will die. If they serve curry, 10 people will die. Even if they serve something like mashed potatoes, 5 people will die.” They serve rice with mushroom sauce, nobody even gets sick – can I hide behind “well they didn’t serve soup, curry or mashed potatoes, the predictions were CONDITIONAL?
> The problem is people insisted that the scales on the y axes of the graphs should be exact
I am not asking for scales to be exact, I am asking them to convey uncertainty about the scales in a way that is consistent with what actually happened.
>Their models are not about “govt driven mandates” they’re about “people actually reducing their contacts” by whatever means.
Their paper literally has the words “Impact of non-pharmaceutical interventions” in the title. I feel like we’re in this weird world where you haven’t even read the paper but are arguing with me about what it contains.
> I read Imperial College as the following:
Yeah read the paper, this was from back in the day where “we can’t let it overwhelm our critical care beds” was still in vogue, not “this will cause more suffering than the Flu and must therefore be stopped”.
> to within a factor of 10 to 20 on overall y axis scaling
I know this is a fuzzy question, but how many times “worse” is covid than the flu?
> What would have to be done is to do a Bayesian model fit, find the high probability set of parameters that predict what actually happened for Sweden
This doesn’t make any sense unless you think we can accurately model pandemic dynamics. There’s no point in (over) fitting some model that doesn’t accurately describe pandemic dynamics regardless of whether we stick the word “Bayesian” into our analysis.
In order to understand whether the model *can* fit pandemic dynamics, you have to search the space of parameters to see if any of those parameter values *do* fit the dynamics. Choosing one set of parameters, running it, and then saying “hey this didn’t look anything like reality” is not very interesting. I can get some resistance vs temperature data for a certain kind of wire, and fit it very well with a polynomial, but if I just choose some other coefficients for the polynomial it won’t fit! big surprise! This doesn’t mean “polynomials can’t predict resistance vs temperature”
Show me that there is no way to choose a time-varying set of parameters representing swedish contact rates that make it possible for the model to get within 10-15% of what actually happened in Sweden and I will agree that it’s a lousy model.
However, my conclusion from what you say is that you simply disagree with their labeling… If they had not said “close universities” and “close schools” and such but simply said “reduce contacts by x” would you have still been so upset? I don’t think so.
As I see it we have “textual description -> degree of contact between people -> illness and death”
you are upset that the implied function “textual description -> illness and death” doesn’t match. I am saying that it doesn’t match because the part “textual description” -> “degree of contact between people” was only ever supposed to be some kind of hand-wavy qualitative guess, but that “degree of contact” -> “illness and death” has the properties:
1) degree of contact A / degree of contact B -> illness and death A / illness and death B is a reasonable function (relative contact changes map to relative illness changes in an ok way)
2) degree of contact A -> illness and death A is a function that is within a factor of 10-20 or so of what happened even though Ioannidis and the WHO offered illness estimates that ranged over factors of 1000x
I think that’s reasonable.
The politics of it all could be objectionable, the difficulty of guessing what people will actually do is real, but no one has shown me that conditional on knowing what people actually did, the dynamics are objectionable and don’t fit.
As for how close they got to reality…
According to ourworldindata.org UK hit a peak daily (7 day rolling avg) confirmed death rate of about 14/million/day in April 2020 and 18/million/day in Jan 2021.
The report you referenced says that under an “Unmitigated Epidemic”, Figure 1 A the UK would hit a peak death rate of 20/100k/day so that’s about 200/Million/day and would have 510,000 total deaths before the epidemic hit endemic phase (everyone who was going to get it had gotten it). In the ACTUAL epidemic which was heavily mitigated (including vaccinating a vast majority of everyone over the last 6 months), SO FAR, they have 132,000 deaths.
200/18 ~ 11x and 510/132 = 3.86 so yes, they are within a factor of 4 of the real death rate EVEN WITH mitigations (which probably means they UNDER estimated the death toll in the unmitigated epidemic if anything), and they were within a factor of about 11 in terms of peak deaths per day (for the mitigated epidemic relative to the projections of unmitigated). So basically if anything it indicates they again underestimated how bad the unmitigated epidemic would be.
>you have to search the space of parameters
Which is, pray tell, what? The space of human decisions? Some (effectively) infinite dimensional space? My model says that cases are in L^infty(days), I can fit it perfectly to cases.
>Show me that there is no way to choose a time-varying set of parameters representing swedish contact rates that make it possible for the model to get within 10-15% of what actually happened in Sweden and I will agree that it’s a lousy model.
My model above (case numbers are bounded on each day) fits the data perfectly, can I publish this in Nature?
>“reduce contacts by x” would you have still been so upset?
Would I have reacted differently if they had said something else? Maybe, what’s the point.
>As I see it we have “textual description -> degree of contact between people -> illness and death”
Again, is this what the paper I referenced actually says? Because “less contact => less spread of disease => less disease” is sufficiently uncontroversial since the germ theory of disease that I struggle to see why it’s an insight anyone would disagree with. If that’s what you take from the ICL paper then I could have told you the same thing even without an epidemiology degree.
>So basically if anything it indicates they again underestimated how bad the unmitigated epidemic would be.
You’re making the assumption that NPIs matter for the number of deaths, which may or may not be true. I’ll grant you that they were within a factor of 10 for their deaths calculation for the UK. However, a large part of the paper was to look at how this would overwhelm the healthcare system, how do those numbers compare?
I’m not talking about your model, I’m talking about *their* model. I honestly haven’t looked at the code so I don’t know what the parameters look like, but I imagine they are things related to the average degree of mixing of different populations as a function of time. It’s reasonable to think that averages over millions of people are smooth functions which change “slowly” even when a shock occurs, rather more over a period of 3 days than over a minute. So within say the Fourier series with upper frequency limit of 1/(2 day) can you find functions of time for the parameters that induce the outcomes to be within a few tens of percent of the ones seen in Sweden?
If so, then the actual results are a possible outcome from the model. If there’s no way for the model to fit Sweden no matter what you do to the parameters, then that’s definitely decisively proven that the model doesn’t and CAN’T predict real pandemics.
Then beyond that, once you’ve found some approximately correct parameter vectors, can you look at survey data and google movement data, and soforth and show that the parameters estimated which track the actual infection also track at least qualitatively and order of magnitude the measured quantities they are modeling? If so, then the model is good, if not then the model has some kind of problem with it in terms of some assumptions. It’s no good to say you can explain say asthma by assuming that the density of air pollution particles is approximately equal to the density of concrete. Similarly it’s no good to say that you can fit the Swedish pandemic but only if you assume all the people are living in an isolation bubble like an immunodeficiency patient.
As for how well they did in terms of hospitalization etc, their prediction under the “school and university closure, case isolation and general social distancing” was that they predicted a peak ICU usage of just under 4/100k occurring in early May 2020, and they actually got a peak of 4.7/100k occurring on April 12 or so and falling off to nearly nothing by Aug.
If you look at their green curve from their Appendix panel B and compare it to the ICU per million people (convert to per 100k by dividing by 10) listed on ourworldindata.org
So basically their prediction for ICU usage under their most restricted type of NPI peaked at the actual value, dropped off a little sharper than they predicted, but was largely spot on the actual results for the period from April 2 through late Aug 2020.
I don’t see how you can call that any kind of failure.
> So within say the Fourier series with upper frequency limit of 1/(2 day) can you find functions of time for the parameters that induce the outcomes to be within a few tens of percent of the ones seen in Sweden?
What’s your point? Of course we can fit a smooth function to smooth data. To paraphrase von Neuman – given 5 parameters, I can even fit an elephant.
> If there’s no way for the model to fit Sweden no matter what you do to the parameters, then that’s definitely decisively proven that the model doesn’t and CAN’T predict real pandemics.
This is a bizarre take, the question isn’t whether we can somehow fit the model to data, the question is whether the model *with the real parameters* for things like “number of contacts per person” fits the data.
>Similarly it’s no good to say that you can fit the Swedish pandemic but only if you assume all the people are living in an isolation bubble like an immunodeficiency patient.
Yes exactly, as mentioned earlier someone took the model, and ran it using the Swedish policies.
> So basically their prediction for ICU usage under their most restricted type of NPI peaked at the actual value, dropped off a little sharper than they predicted, but was largely spot on the actual results for the period from April 2 through late Aug 2020.
I can’t find the figure you’re referring to in the linked PDF, but the ICU numbers in Figure 3B are nothing like what you cite. The peak at around 15, then go around for a bit, and then grow like crazy once measures are relaxed (as they were in summer 2020, which everyone seems to like to forget).
I liked your blog but I didn’t read it (no-one does).
Regarding the link to nec pluribus impar:
This shows that any non-pharmaceutical intervention implemented just before the peak of a pandemic wave will always seem to have been effective. This reminds of the RCT literature on anti-nausea drugs which consistently shows that they don’t work. Anecdotally though, people swear by them (“I had a hangover and took some of drug X and then felt much better”). Nausea is a self-limiting condition for most people so whatever drug they took for it will seem to work.
The general principle is that its very hard to use observational data to explore a condition that naturally waxes and wanes.
Yeah I agree, this kind of stuff is really hard to study! Honestly I don’t think we can get much from observational studies other than having a weird collection of anecdotes to throw at anyone who thinks they understand what is going on.
Matty –
That’s pretty funny
I was thinking of Phillipe when I was criticizing armchair epidemiologists – because I think the stuff of his I’ve seen lacks an appreciation for uncertainty.
And I was also thinking about how many armchair epidemiologists’ take on Flaxmen illustrates my point – because they react to it as if was a PREDICTION (of 2 million deaths or whatever) when what it was, was a conditional PROJECTION. That illustrates how their approach reflects a lack of understanding of conditional probability.
I’ve seen the same countless times with climate change, where “skeptics” react to conditional projections as if they’re predictions, and say that the predictions were wrong when the conditions they were built on didn’t materialize.
>because I think the stuff of his I’ve seen lacks an appreciation for uncertainty.
Yeah I haven’t read much by him other than his criticisms of various epidemiology papers which to me seem spot on; maybe he lacks an appreciation for uncertainty when making predictions himself.
>because they react to it as if was a PREDICTION (of 2 million deaths or whatever) when what it was, was a conditional PROJECTION
Did you actually read the ICL paper? It contains many more “projections” (some of which include various interventions). See my other comment above replying to Daniel Lakeland which should be visible once approved by Andrew. Obviously the “they said 2 million deaths if we do nothing; we did something and got less they must be incompetent” take is nonsense.
Matty –
> Yeah I haven’t read much by him other than his criticisms of various epidemiology papers which to me seem spot on; maybe he lacks an appreciation for uncertainty when making predictions himself.
Again funny. As my father used to say, “That’s why they make chocolate and vanilla.” and “That’s why they have horse races.”
I’m actually referring to his pieces on epidemiology (re COVID).
To be clear, I’m not referring to the statistics or math in particular (I can’t evaluate that). I offered some of my takes here:
https://statmodeling.stat.columbia.edu/2021/07/31/struggling-to-estimate-the-effects-of-policies-on-coronavirus-outcomes/
Let me know what you think.
> Did you actually read the ICL paper?
No.
I’ll read what you wrote and yeah I’m referring to the “they said 2 million deaths if we do nothing; we did something and got less they must be incompetent” take.
I’ve seen that take tons o’ times from armchair epidemiologists.
Replying to Joshua:
> Again funny. As my father used to say [..]
Too cryptic for me to understand :)
>I offered some of my takes here: […] Let me know what you think.
I read some of your comments you linked to(especially the long one), and I broadly agree with your point. It is not possible to really say anything rigorous about counterfactuals (“costs of lockdowns bigger than benefits”) because – as you rightly point out – we don’t know what would have happened if there hadn’t been lockdowns.
That said, I feel like you’re coming at it from an angle of (the following are obviously not literal quotes) “we need to find the ‘best’ policy under uncertainty and maybe we get it wrong sometimes but that’s to be expected”. I don’t think governments have the responsibility of finding the ‘best’ policy, so I do think that anything that establishes reasonable doubt (see footnote 2 below) for the effectiveness of lockdowns should cause them to be rejected (see footnote 1 below). Read this way, the burden of proof is on the lockdowners to prove they are effective; first-order cost-benefit “analyses” like Lemoine’s that go against lockdowns should require the lockdowners to argue why second-order effects nevertheless favor lockdowns. I read work like Lemoine’s in the sense of “this should shatter anyone’s confidence in simplistic takes on lockdowns”, not in the sense of “this shows that lockdowns are obviously not worth it”. But I think Lemoine sees this differently.
Footnote 1: in fact, I think that courts in many places *are* expected to rule this way, but since the scientific literature is so full of overconfidence they think there is no reasonable doubt.
Footnote 2: just like we require “beyond reasonable doubt” to lock up people in other circumstances also. See also Michael Huemer’s http://www.owl232.net/papers/passivity.htm which explores similar ideas in a pre-pandemic setting from a more philosophical perspective.
Matty –
> I don’t think governments have the responsibility of finding the ‘best’ policy, so I do think that anything that establishes reasonable doubt (see footnote 2 below) for the effectiveness of lockdowns should cause them to be rejected (see footnote 1 below). Read this way, the burden of proof is on the lockdowners to prove they are effective; first-order cost-benefit “analyses” like Lemoine’s that go against lockdowns should require the lockdowners to argue why second-order effects nevertheless favor lockdowns.
Ok. So yah, I think this is where we disagree. I don’t see why either side here is saddled with a burden of proof. Particularly because “proof” in this situation is just a fundamentally unrealistic standard. That doesn’t mean that I don’t think that the arguments should be engaged. Indeed, the only “burden” I see here is on both sides, to have a well-reasoned argument. And here’s where I see both sides failing, including what we have with Phillipe’s anslysis.
The same argument is made with climate change all the time. And with such an approach, we can never have collective action against potentially fat tail existential risk. Both context are further roiled by overwhelming political biases.
In the end, you have a lot of people who, through a democratic process, want action taken. That isn’t carte blanche, in my view, but it’s important – especially against a background of a minority with disproportionate power and a historical legacy of disproportionate representation who, IMO, have a enormous sense of entitlement.
To circle that back, the problem that we have here that I”m focusing the most finely on is the basic difficulty of using cross-national analyses to try to assess the efficacy of interventions. There are so many uncontrollable variables there that the task is immense. (Personally, I think longitudinal analysis within countries is a much better way to go as while far from perfect it helps to eliminate certain obvious confounds.) Next order of less sub-optimal would be in comparing most like countries, but that’s really complicated also, and it’s easily gamed by again being selective in the confounds you’re controlling for- (see Phillipe in that regard, btw). And then there’s the complicating factor of controlling for the pre-existing conditions of COVID prior to intervention – which is mostly ignored by the anti-lockdown crowd.
And what I see is, for the most part, epidemiologists who seek to factor all that in, because that’s what epidemiologists do. They test interventions all the time and do so in a framework of controlling for confounds as well as they can. On the other side, we have armchair epidemiologists who not only don’t try to control for those biases (or other statistical biases), but who don’t even understand what they are or that they need to be controlled for. Once again, I draw a parallel to the discussions over climate change mitigation – where so many people are absolutely sure they know what the solution is even though they actually can’t even understand the technical analysis.
> I read work like Lemoine’s in the sense of “this should shatter anyone’s confidence in simplistic takes on lockdowns”, not in the sense of “this shows that lockdowns are obviously not worth it”. But I think Lemoine sees this differently.
We certainly don’t need Phillipe’s work to understand that simplistic takes on interventions isn’t going to get us much of anywhere. But my problem with Phillipe’s work is that IMO, it is overly ideological in nature. That’s fine. But you have to account for the way that your ideology interacts with your analysis. Unfortunately, again as with climate change, we have a lot of armchair experts who claim they’re doing pure science and non-ideological analysis and pointing fingers at the “expert” community and asserting incompetence (usually accompanied by accusations of political bias) – when actually the “experts” IN GENERAL take systematically controlling for biases much more seriously
> Footnote 1: in fact, I think that courts in many places *are* expected to rule this way, but since the scientific literature is so full of overconfidence they think there is no reasonable doubt.
Sure, that can happen. But this looks at the problem from only one side. And there’s another side.
> Footnote 2: just like we require “beyond reasonable doubt” to lock up people in other circumstances also. See also Michael Huemer’s http://www.owl232.net/papers/passivity.htm which explores similar ideas in a pre-pandemic setting from a more philosophical perspective.
I’ll try to follow your link. But I don’t think the analogy applies. We don’t need “beyond all doubt” in some circumstances. And your use of “lock up” isn’t even close to analogous. Maybe, if we responded as did China then we might be able to have that philosophical conversation as to whether “locking people up” during a pandemic is that same thing as putting people in jail for committing a crime – but we can’t even approach that discussion, IMO, until you agree that the analogy of the interventions we’ve had in this country compared to putting people in jail for committing a crime is a horribly tortured one.
I will announce my bias. I find the whole “freedoms” angle around COVID pretty inane. Of course, once we get to the extremes of the whole range of what’s gone on, the discussion becomes extremely important and relevant. But for the most part I find the ways it’s been engaged with from the anti-lockdowners to be a political game, reflective as I said of an over-developed sense of entitlement. For example as seen with the claims that mask requirements by. store owner, say, who wants to limit the risk of employees and customers at a minor expense of other customers, is some sort of violation of basic freedoms.
There’s no reason people shouldn’t play that game, but please, it’s politics. Let’s have a political discussion But don’t pretend its science.
> Too cryptic for me to understand :)
It just means people will always have different opinions.
Joshua
>Particularly because “proof” in this situation is just a fundamentally unrealistic standard.
I don’t mean “proof” in the mathematical sense. Do read the article from Huemer. I think that strong government action – in particular when human rights are taken away – should need to “prove” that it is appropriate.
>The same argument is made with climate change all the time. And with such an approach, we can never have collective action against potentially fat tail existential risk. Both context are further roiled by overwhelming political biases.
The same argument is made in lots of places, it was also a big part of the communism/capitalism debate (central planning, collectivization, etc..) which the communists lost.
>In the end, you have a lot of people who, through a democratic process, want action taken. That isn’t carte blanche, in my view, but it’s important – especially against a background of a minority with disproportionate power and a historical legacy of disproportionate representation who, IMO, have a enormous sense of entitlement.
Which minority is this, old people?
> And what I see is, for the most part, epidemiologists who seek to factor all that in, because that’s what epidemiologists do.
Is it though? The ICL group doesn’t seem to factor in much other than government decisions. Maybe everything I see selects for overly simplistic takes, but as far as I can tell it’s more or less impossible to make accurate predictions.
>And your use of “lock up” isn’t even close to analogous.
>but we can’t even approach that discussion, IMO, until you agree that the analogy of the interventions we’ve had in this country compared to putting people in jail for committing a crime is a horribly tortured one.
The analogy to house arrest wasn’t too far off in plenty of European countries. I think we can all agree that several fundamental rights were restricted by most states during the pandemic (free movement, free association, right to protest, etc..) such rights are typically only restricted as part of criminal proceedings. Australia has recently restricted the human right to leave a country. This is serious stuff. You can’t do this in a democracy (here I mean a state that respects human rights) without being *very* sure you’re doing the right thing. Just like we let some criminals go when we aren’t 100% sure they are criminals even though this leads to mistakes being made because “don’t wrongfully lock anyone up unless you’re 100% sure is part of a democracy.
>>I will announce my bias. I find the whole “freedoms” angle around COVID pretty inane […] , reflective as I said of an over-developed sense of entitlement
Probably you don’t understand the argument then? How would you like it if I told what I thought you should do and who you should associate with and then got together with my buddies to force you to comply? Scale it up, and you have government. Or maybe you live in a place like the US that just hasn’t restricted freedoms so much? Again, do read the linked Huemer post. I’ve lived in plenty of dictatorships growing up and I was quite shocked to see how quickly the respect for fundamental rights got lost once a few people got afraid of a virus. Made me understand how quickly something like this can go, luckily we don’t do the same stuff like in the first half of the 20th century anymore but “fear taking over politics” is definitely a force to respect.
>There’s no reason people shouldn’t play that game, but please, it’s politics. Let’s have a political discussion But don’t pretend its science.
Sure, leave science out of politics! This is exactly what I would want; that people wouldn’t pretend that their politics is somehow science. Like most epidemiologists/”scientists” I see.
Matty –
> I don’t mean “proof” in the mathematical sense. Do read the article from Huemer. I think that strong government action – in particular when human rights are taken away – should need to “prove” that it is appropriate.
Sorry, that’s not any better from my perspective. “Proof” of what is “appropriate” is necessarily subjective, and thus no less possible than proof in a mathematical sense (maybe even less so).
> The same argument is made in lots of places, it was also a big part of the communism/capitalism debate (central planning, collectivization, etc..) which the communists lost.
As to whether centralization or collective action have “lost” in some total sense isn’t what I had in mind. There are degrees of centralization and collective action which, IMO, certainly haven’t been established to have “lost.” Even on a larger scale it’s a hard argument to make looking at China recently. But really, that’s an entirely different discussion.
We’ve taken collective action in this country as well, under a non-communist state, when facing a fat tail existential risk – such as the war effort when people accepted rationing on a massive scale.
Bur more relevant, my point is that the question of whether sacrifice for the sake of community welfare is in zero sum relationship with individual freedom is also a subjective evaluation.
Perhaps because I’ve spent a good amount of time in Asia, I’m well aware that different people feel differently about that question. But even in this country it’s often a moving target. I’m guessing you’re probably in the “libertarian” range and so this might not apply directly for you but certainly the question the relationship between community standards for behavior and individual “freedom” shifts according to context across the political scale in this country, as in abortion versus COVID interventions.
> Which minority is this, old people?
I’m talking about the polling that shows majority support for COVID interventions, even strong interventions, and in some ways in particular among those who stand to be most affected (such as those most likely to be essential workers, or those who would likely be laid off with no unemployment compensation). As we know, polling data aren’t conclusive, but they’re relevant information here.
> Is it though? The ICL group doesn’t seem to factor in much other than government decisions. Maybe everything I see selects for overly simplistic takes, but as far as I can tell it’s more or less impossible to make accurate predictions.
Well, we’re both going to have a tendency towards over-simplification here…. I think we agree it’s more or less impossible to make “accurate” predictions. But the same would apply for climate change in the sense that “accurate” is subjectively assessed. My point is that epidemiologists are more likely to at least try to quantify confidence intervals. Of course they’re going to be wrong to some extent but I think it’s better than people who don’t even try, and even worse, don’t even understand the importance of controlling for confounding variables or the potential statistical problems such as base rate or Simpson’s paradox. So for me the relevant question is when is accuracy good enough for action – recognizing that inaction is a form of action when a threat is present (which I think you’re basically ignoring).
> The analogy to house arrest wasn’t too far off in plenty of European countries.
From here it’s going to be hard for us to progress. As far as I’m concerned, putting someone in jail without a standard of proof as punishment for merely being charged with committing a crime, is an entirely different context than what we’ve experienced in this country (I’m not particularly aware of what you’re referring to as “house arrest” in Europe but fear that would be another can of worms, so I’d rather stick to this country). That doesn’t mean I’d dismiss slippery slope scenarios – they’re important. But we can run slippery slope scenarios in either direction. And the first place to engage this discussion, IMO, is with a realistic characterization of what we’ve faced. Then we could move on to the far ends of the distribution.
> I think we can all agree that several fundamental rights were restricted by most states during the pandemic (free movement, free association, right to protest, etc..) such rights are typically only restricted as part of criminal proceedings.
This only works if you look at “fundamental rights” from a frame where individual rights exist in some zero sum relationship with community welfare and responsibility to the community. I don’t feel my fundamental rights were restricted in the least. You’d probably think that cell phone tracking would be an infringement of your fundamental right but someone living in Korea would likely feel very differently about that. I think that being able to walk into a grocery store where the risks from COVID have been reasonably mitigated is a fundamental right, and not being able to do so would, in some sense reflect a sub-category trade-off between fundamental rights or freedoms. I think that being able to stay home from work and collect unemployment and not have to go to work when there’s a raging pandemic and choosing between risking being fired or risking binging COVID home to infect a 90 year old MIL who lives with you is a form of freedom.
> Australia has recently restricted the human right to leave a country. This is serious stuff. You can’t do this in a democracy (here I mean a state that respects human rights) without being *very* sure you’re doing the right thing.
You seem to think it’s your prerogative to dictate to other people what rights they should and shouldn’t have, based on your own set of priorities, that likely shifts with context. Do humans from Mexico have a fundamental right to cross the border into the US? It’s funny how right now, the fundamental right of Afghans to seek freedom and risk their lives to flee across borders is subject to a different standard than someone from Honduras.
> Just like we let some criminals go when we aren’t 100% sure they are criminals even though this leads to mistakes being made because “don’t wrongfully lock anyone up unless you’re 100% sure is part of a democracy.
??? Surely you agree that many people in this country (particularly those with no money) are imprisoned despite a lack of 100% certainty that they’re guilty of the crime they’re accused of.
> Probably you don’t understand the argument then?
That’s a bad place to go – that I disagree with you because I “don’t understand the argument” I’d like to avoid going there.
> How would you like it if I told what I thought you should do and who you should associate with and then got together with my buddies to force you to comply?
It happens all the time in various ways. And sometimes I think it’s entirely reasonable.
> Scale it up, and you have government.
I don’t argue that there are limits to my acceptance of that. But the fact that slippery slopes exist (in infinite directions, btw) doesn’t really help us to address what we’re really faced with now.
> Or maybe you live in a place like the US that just hasn’t restricted freedoms so much?
Honestly, I’d rather be living in New Zealand right now. And get this, it’s actually not because I value freedom any less than you. Try to wrap your mind around that. It’s not because I value freedom any less than you. If you can’t accept that, then there’s really not much more to discuss here. But I challenge you to understand that I can have views different from you on this even though I value freedom no less than you do.
> Again, do read the linked Huemer post. I’ve lived in plenty of dictatorships growing up and I was quite shocked to see how quickly the respect for fundamental rights got lost once a few people got afraid of a virus.
So I’d suggest that you have a bias. No less than I do. But the point is to examine the interaction between our biases, not to fall into the trap of thinking that our own bias – reality.
> Made me understand how quickly something like this can go, luckily we don’t do the same stuff like in the first half of the 20th century anymore but “fear taking over politics” is definitely a force to respect.
Again, slippery slopes exist, but so do trade-offs. There are non-dictatorships that have enacted stricter restrictions than we have in response to COVID. My state had some fairly stringent responses (NY STate) and I don’t feel my freedoms were materially restricted in the least. My neighbor (I live in a largely rural area) might feel differently than I do and call Cuomo a dictator. Who’s right?
> Sure, leave science out of politics! This is exactly what I would want; that people wouldn’t pretend that their politics is somehow science. Like most epidemiologists/”scientists” I see.
Not pretend their “science” isn’t politics. Like Phillipe, in my view.
Replying to Joshua:
I think we more or less agree on most points.
> recognizing that inaction is a form of action when a threat is present
This comes down to philosophical issues we probably won’t resolve here, but see the Huemer paper I linked.
>As far as I’m concerned, putting someone in jail without a standard of proof as punishment for merely being charged with committing a crime, is an entirely different context than what we’ve experienced in this country
I’m not in the US, but “you’ve been exposed to someone with corona, so now you do 2 weeks quarantine without a test because we say so” or “no leaving the house without a ‘valid’ reason” or “if you protest we will arrest you” or “we will close the beaches and police it with fines and drones” or “if you have a family dinner with more than 5 people, you will get heavy fines” or “up to 10 years in jail if you skip quarantine on entry”(footnote) have been part of of the European covid response. I think the analogy to house arrest is entirely appropriate. Just because you do it to everybody doesn’t make it any better.
Footnote : This was reported in the media for the UK, but it may have been sensationalisation.
>But I challenge you to understand that I can have views different from you on this even though I value freedom no less than you do.
I understand your position, but you’re using “freedom” differently to myself. Say you’re being harassed by someone, and take them to court to leave you alone. However, you can’t provide evidence that convinces the jury that you’re being harassed. Then likewise your “freedom” is being restricted because you worry about being harassed. However, in Western countries it has been the view that better your freedom is restricted 10x like this than that the court wrongfully locks up the harasser once.
> So I’d suggest that you have a bias.
Obviously I have biases. And it’s not like I don’t understand your position, though re: “I’d rather be living in New Zealand right now.” see New Zealand right now, also if you think harsher lockdowns would have achieved this see Europe.
> There are non-dictatorships that have enacted stricter restrictions than we have in response to COVID.
Well sure, but what is a dictatorship really? Plenty of “democracies” have violated human rights before, just because people vote for leaders does not make what they do appropriate (see all your wars).
> My neighbor (I live in a largely rural area) might feel differently than I do and call Cuomo a dictator. Who’s right?
Neither of you, if you ask me :P Your freedoms were (obviously) materially restricted, but you see this as a worthwhile tradeoff. Your neighbor does not. Cuomo was not a dictator, but this does not mean that New York state did not violated human rights just like e.g. Russia is (probably; Putin is very popular) not a dictatorship but nevertheless violates human rights; if you disagree re: Russia pick some other country you like. Dictatorship/democracy is a false dichotomy.
Matty –
> I’m not in the US, but “you’ve been exposed to someone with corona, so now you do 2 weeks quarantine without a test because we say so” or “no leaving the house without a ‘valid’ reason” or “if you protest we will arrest you” or “we will close the beaches and police it with fines and drones” or “if you have a family dinner with more than 5 people, you will get heavy fines” or “up to 10 years in jail if you skip quarantine on entry”(footnote) have been part of of the European covid response. I think the analogy to house arrest is entirely appropriate. Just because you do it to everybody doesn’t make it any better.
It’s hard for me to comment on that since I don’t know anything about the conditions concurrent with what you’re describing. I try to avoid forming opinions on this kind of thing in the abstract, as I don’t know how I’ll feel about it without being there. I don’t think that universal opinions on this is very useful. Sure, I can think of situations where that would be well beyond what I think is reasonable. Or I can think of situations where I’d feel I should just suck it up and endure for the sake of my community, and stop feeling so sorry for myself and feeling so entitled.
I understand your position, but you’re using “freedom” differently to myself. Say you’re being harassed by someone, and take them to court to leave you alone. However, you can’t provide evidence that convinces the jury that you’re being harassed. Then likewise your “freedom” is being restricted because you worry about being harassed. However, in Western countries it has been the view that better your freedom is restricted 10x like this than that the court wrongfully locks up the harasser once.
I think the main difference is that I see “freedom” as more of a balance. I don’t see restricted movement as necessarily a loss of freedom because I might gain freedoms as a result.
> Obviously I have biases. And it’s not like I don’t understand your position, though re: “I’d rather be living in New Zealand right now.” see New Zealand right now, also if you think harsher lockdowns would have achieved this see Europe.
Truth told, I thought about living in New Zealand prior to COVID. But again, I do see things differently than you. I look at the extreme measures being taken right now in New Zealand within the context of 6 months of relative safety, not wearing masks, etc. Again, I see these issues as a balance, I think, less than you do.
> Your freedoms were (obviously) materially restricted,
There’s my point. You think that beyond the shadow of a doubt my freedoms were materially restricted but I don’t see it that way – because (1) I still had many ways to enjoy my life and (2) I gained advantages (or at least considered that the time I well might do so) by virtue of shelter in place orders.
> but you see this as a worthwhile tradeoff.
A tradeoff of *freedoms.* With the potential of a net advantage in terms of *freedom.* Like the people in New Zealand (on the whole) might feel. Let’s say people in South Korea, who have to submit to all kinds of controls monitored through their cell phones. You might see that as a restriction of freedom but many of them very well might look at us in the US and think that our freedoms have been limited by a much worse pandemic trajectory.
Again, I don’t see a zero sum relationship between individual freedom to do whatever I want, and community restrictions on what I can do, as *necessarily* a zero sum relationship – in terms of freedoms, in term of advantages, etc.
> There’s my point. You think that beyond the shadow of a doubt my freedoms were materially restricted but I don’t see it that way
You are seeing this wrong. It is obvious that your freedoms were materially restricted, because the definition of “freedoms” that I am using (something like “things you can do without being sanctioned by the state”) is one where it is obvious that your “freedoms” were materially restricted. Things like “the freedom to meet with people in your own home”, “the freedom to protest”, “the freedom to move from one place to another”. It’s a question of definitions; if you don’t see it that way you are using a different definition of “freedoms” than I am that somehow incorporates “what things you can do within your level of risk tolerance” or “what things you can do while feeling good about it”.
Matty –
> You are seeing this wrong. It is obvious that your freedoms were materially restricted, because the definition of “freedoms” that I am using (something like “things you can do without being sanctioned by the state”) is one where it is obvious that your “freedoms” were materially restricted.
If by your defintion my freedoms are restricted but by my defintion they aren’t, they why is it obvious my freedoms are being materially restricted?
If public health officials say everyone should stay home except to shop and seek emergency care, because in their estimation it will reduce illness and death and help prevent illness among hero healthcare workers and keep them from being overwhelmed, then I will have no desire to go out unless it’s an emergency. Are my freedoms being restrictes? I don’t feel that way.
And if the government then says if I do go out I’ll be questioned and possibly arrested are my freedoms being restricted? I wouldn’t feel that way because I wasn’t going to go out anyway. And if I do go out ill be infringing on a healthcare worker’s “freedom” to do their work saving lives without indie risk to their own welfare.
In the very least, it’s not a restriction I care about. It’s not something that animates me. Why would a someone get animated about not being able to go out in violation of what public health officials soad is responsible behavior? By my standard, only if they have an inflated sense of entitlement or are overly motivated to be a freedom fighter, would they feel their freedoms have been limited. But of course no one else has to have my standards.
And then there’s the whole issue of which and whose freedoms are being restricted. Sometimes, by restricting some freedoms you expand others. And there are complications like when governors tell businesses they can’t require customers and employees to be vaccinated, in the name of “freedom” which is a restriction of a business owner’s ability to run their business and to do what he/she thinks is the best way to run a safe business.
In short, I think it’s complicated.
>If by your defintion my freedoms are restricted but by my defintion they aren’t, they why is it obvious my freedoms are being materially restricted?
Because I’m using my definition. I see no use in arguing over definitions here. You are free (no pun intended) to use a different set of definitions if you like, under which it is no longer obvious your “freedoms” are restricted :)
>then I will have no desire to go out unless it’s an emergency. Are my freedoms being restricted? I don’t feel that way.
Is coffee without milk the same as coffee without cream? But obviously if nobody has an interest in going out, then no use in making a law against it because nobody will go out.
> In the very least, it’s not a restriction I care about. It’s not something that animates me
Well yeah, same here, but I go the other way in just ignoring the restrictions. But they *are* annoying. But to be honest I don’t think this argument is particularly strong; I read it as “someone else’s rights are being violated why should I care”.
> Why would a someone get animated about not being able to go out in violation of what public health officials soad is responsible behavior?
Because public health officials are obviously idiots in most places. How would you feel about public health officials instituting another prohibition, or banning rock climbing, or banning “unhealthy speech”? Because this kind of argument proves far too much. I mean why get animated at all over anything, just put up with whatever gets imposed on you and be happy that the people in control know better I guess.
>or are overly motivated to be a freedom fighter, would they feel their freedoms have been limited
Sure, everybody who disagrees with public health officials imposing their (which happens to be close to your) risk-level onto everyone is just entitled, that’s it. Sorry if I’m being facetious, but isn’t it strange that all around the world, public health officials have had different standards for “responsible behaviour”? What are the odds that your particular public health officials got it “right”?
> in the name of “freedom” which is a restriction of a business owner’s ability to run their business and to do what he/she thinks is the best way to run a safe business.
That’s one way to phrase it, another way to see it is that that the government is refusing to use its monopoly on force in certain situations (punishing those who lie about their vaccination status). But it’s complicated, I agree. I mean how would feel about banning a circumcision requirement for employees, is it bad that this is (presumably) not allowed?
Matty –
> But obviously if nobody has an interest in going out, then no use in making a law against it because nobody will go out.
Well yeah, but for some people the interest in going out is to have an interest in going out. I”m not saying all, of course, but for many making a bit deal about restrictions is an animating force. They can say they’re rebels fighting against tyranny. Kind of elevates one’s existence to be fighting against tyranny. I know I’m being condescending but I think the phenomenon is real.
> I read it as “someone else’s rights are being violated why should I care”.
No, I’m not saying that (I can understand why it would read that way). It’s not that I don’t care that other people think they’re rights are being violated. But I’m saying there are different views. Often I see the argument being made in an absolutist way – as if there aren’t others who have, basically, a diametric view.
> Because public health officials are obviously idiots in most places.
Well, I see that differently. Public health officials make mistakes. For a variety of reasons. But no shit.
Everyone makes mistakes, particularly in the context of a fast-moving and unprecedented health crisis where there’s a tone o’ uncertainty – including with respect to just how many people will resist the advise of public health officials merely because of politics. Including just how much misinformation is being pushed our at enormous rates through social media.
>
How would you feel about public health officials instituting another prohibition, or banning rock climbing, or banning “unhealthy speech”? Because this kind of argument proves far too much. I mean why get animated at all over anything, just put up with whatever gets imposed on you and be happy that the people in control know better I guess.
I just have a hard time finding slippery slope arguments. We can run them in any direction. We could trade slippery slopes. I just don’t where it gets anyone.
> Sure, everybody who disagrees with public health officials imposing their (which happens to be close to your) risk-level onto everyone is just entitled, that’s it.
That’s actually not what I’m arguing. And there are certainly some policies being pushed by public health officials that I DON’T agree with.
I’m saying that something like viewing expectations to wear a mask – which I consider a minor inconvenience – as a matter of tyranny – particularly in the midst of a pandemic, seems off by a matter of scale to me. And as such, it seems to me to be forced, and political, and yes it smacks of entitlement to me.
> Sorry if I’m being facetious, but isn’t it strange that all around the world, public health officials have had different standards for “responsible behaviour”? What are the odds that your particular public health officials got it “right”?
I don’t think it’s odd in the least and I’m not expecting “right.” I’m hoping for less “wrong,” for less sub-optimal. I actually think an expectation that they get it “right” to be unrealistic and yeah, reflective of a kind of entitlement.
That’s not an excuse for getting it wrong. There are plenty of inexcusable reasons for why mistakes have been made. But I’m looking at the full context.
> That’s one way to phrase it, another way to see it is that that the government is refusing to use its monopoly on force in certain situations (punishing those who lie about their vaccination status).
I’m not sure – is that an argument about details of execution or a more general argument>
> But it’s complicated, I agree. I mean how would feel about banning a circumcision requirement for employees, is it bad that this is (presumably) not allowed?
Again, slippery slope or ad absurdum arguments just don’t work for me.
Being competent at statistics in epidemiology studies may just be beyond most – so fatalism sets in for the majority of professional epidemiologists?
Invited Commentary: Dealing With the Inevitable Deficiencies of Bias Analysis—and All Analyses. Sander Greenland – https://academic.oup.com/aje/article/190/8/1617/6189732
I really enjoy how Joshua’s comments on armchair epidemiology instantly generated an example to illustrate his point.
https://www.bmj.com/content/363/bmj.k5094
Do we also need to do studies that getting hit by an electric car going 60 mph is just as dangerous as getting hit by an ICE car?
This practice of making an obviously wrong assumption that ignores all prior information, then wasting a bunch of time and money “testing” it while basing policy on that assumption, is not improving the credibility of medical experts.
We are fast headed towards something akin to the reformation when the catholic church lost credibility with a huge portion of the population. I’ve been saying that since 2015 or so, but this response to covid has accelerated it.
His own comment was an illustration of same. Just the same I agree w/ J’s sentiments, which were as accurate 15 mo ago as they are today – which begs the question: why bother?
You should bother. For the same reason a parent lets their child figure out the puzzle by themselves even though they see them struggling.
That is the only way people using EBM may eventually learn and figure out that using science (incorporating prior information and reasoning where appropriate) works much, much better than brute force.
jim –
> His own comment was an illustration of same.
Do you mean in the sense that I seem certain that there’s a lot of uncertainty?
I mean that you’ve correctly recognized that people “waaaaaay, way, way, way over-estimate their ability to assess the efficacy of interventions in a context with so many vast and highly complex confounding variables”
But this has been happening almost since day one. Nonetheless no amount of evidence of failure seems to stop the flood of claims about what’s effective.
I do not see in this list the analysis of the main problem. Any system which involves people must first and foremost be checked for hidden intentions.
For instance, an ordering of a dead patient into COVID-19 victims or into some other category is a conscious decision of a pathologist who is dependent on the results of this decision. You cannot ignore this cyclical dependencies. Two pathologists in the same hospital may have quite different intentions, what will inevitable produce different “raw” data. Consequently, the data will show dramatic changes when one of them goes on vacation. Even worse, their decisions are also dependent on official criteria which change over time and — what may be more important — on the orders from above that are nearly random.
Even a decision about making a test is a conscious decision that consider the probably results of this test and their consequences.
You cannot hope that you are not analyzing intentionally prepared junk, if the people who have produced your data are dependent on the results of your statistical analysis.
Modern global pandemic inevitable causes global data fraud. It was started with the manipulated socialistic statistics from China. A lot of scientists intentionally manipulate data to quickly produce articles which will generate citations in the overheated field. Many of the “breaking news” are already proven to be completely wrong.
And the intentionally wrong results produced from intentionally wrong data are supported by the most prominent journals reasons because they also depend on the political decisions which will be made on base on these results. (Lancetgate)
Vit:
We do have a section in our paper on data and measurement quality! We don’t specifically mention the problem you talk about with dramatic changes in the data when a pathologist goes on vacation. I’d not heard about that before—do you have an example of it?
Andrew:
“For in much wisdom is much grief: and he that increaseth knowledge increaseth sorrow.” However, I was going to translate this anyway.
I use statistical methods for the quality control and in my area the results of an analysis almost always affect the people who produce the raw data, and these people usually know that these results will affect them, and are not dumb to ignore this fact. The current pandemic has the same problems.
1. Classification rules
The criteria for definition of COVID-19 cases and not COVID-19 cases are not objective. They are depended on subjective criteria which are not stable.
The most obvious cases are listed even on Wikipedia:
a) “On 17 April, following the Wuhan government’s issuance of a report on accounting for COVID-19 deaths that occurred at home that went previously unreported, as well as the subtraction of deaths that were previously double-counted by different hospitals, the NHC retrospectively revised their cumulative totals dating to 16 April, adding 325 cumulative cases and 1,290 deaths.” (https://en.wikipedia.org/wiki/Statistics_of_the_COVID-19_pandemic_in_mainland_China#cite_note-NHC_17_Apr_WH_correction-5)
b) “Russia More Than Doubles April Coronavirus Death Toll”, The Moscow Times, 13 June 2020 (https://www.themoscowtimes.com/2020/06/13/russia-more-than-doubles-april-coronavirus-death-toll-a70564)
There are a lot of more subtle cases which are difficult to recognize unless you are directly involved somewhere in the process. (I talk with doctors.)
The “raw” data concerning COVID-19 cases is affected by all levels of rules, and they are not stable. Even the WHO recommendations were considerably changed several times. Sometimes the local test labors get overloaded, sometimes the insurance companies of a country change rules for compensations, sometimes a specific test facility get closed… Nobody considers such changes by analyzing the data.
Sometimes politicians intentionally change the rules to “fight” the raising or falling numbers. It is possible to change the rules even for the ICU cases. For instance, Swiss nursing homes have received very interesting (and ethically somewhat questionable) orders about providing and not providing medical help for old or chronically ill patients. And these rules were also periodically changed by politicians.
The difference of two pathologists in my example describes this problem on the most basic level. A lot of death causes are not obvious, consequently many decisions are partially based on personal opinions. These personal opinions depend on personal preferences. Which may drastically differ.
This is a typical case for Russia, and I know such cases from other countries. It is possible to find some complaints of the pathologists on social networks, even in English (probably not all of them are already banned).
2. Commercial interests
There are a lot of possibilities to manipulate test results, even if the measurements are produced independently. The people in a hospital which is generously paid for COVID-19 cases will act differently in comparison to the people in a hospital which will be closed, if a single COVID-19 case will be found in it.
Such commercial interests may be local, may be regional, may be country specific, and may change immediately after a sudden political decision.
Sorry, I cannot share my samples because this information is not public. There are some articles from Russia about a hospital where the medical staff was prohibited not only to test patients but also to carry face masks. (This was leaked because of only one person who dares to speak with journalists, despite many severe cases and some deaths. Most doctors and nurses in all affected countries prefer to be silent.)
3. Wrong basic models
Wrong models produce wrong conclusions. This means, the same data may — dependent on the kinds of ignorance or misunderstanding which are hardwired into analysis — prove quite different theories, .
Most models simply assume a consistent mix of randomly and evenly connected people. This is usually not the case. As a rule, the virus propagation networks and the qualities of the clusters are not considered. (https://vit-r.dreamwidth.org/file/90597.png)
Some critical dependencies may be simply sorted out of the equation because of ignorance or because of political reasons.
For instance, a conclusion “The new version X is more contagious” could be corrected into “The schools have again opened their doors for the children”. (https://vit-r.dreamwidth.org/file/88737.png)
A headline “The new wave brings more deaths” may mean “The management of Y Inc. has decided to save money and has recruited many cheap part-time workers who neglect the sanitary standards and take shifts in many nursing homes that belong to this chain”. (https://vit-r.dreamwidth.org/file/89675.png)
Of course, you could not get such information, but the mass media sometimes report about the criminal investigations that were caused by statistical anomalies (which were ignored by science but were noticed by the relatives of “fatal cases”).
Even a recommendation to disinfect hands could produce quite different effects. (https://vit-r.dreamwidth.org/file/89856.png)
This is the cause why all arguments about masks are endless and meaningless when the other variables are not discussed.
(The legend: https://vit-r.dreamwidth.org/file/89281.png)
The problems get worse in cases where medical interventions are involved. You can close schools from a specific day, but a vaccination process is inevitable prolonged through time, uneven on workdays and the weekend, and the effects of a vaccination of a single person are not immediately, but cause complex dynamic changes. (Two images here: https://vit-r.dreamwidth.org/1134325.html?style=light The text is written in Russian, which is today the de facto free speech language, and some of my posts are intentionally protected from the automatic translation.)
I would not prove that the people are different, but most models ignore what the doctors known from the previous Spring. The _A_bnormal reaction on the virus s_A_rs-cov-2 is highly dependent on the risk groups. This means, the correct statistic would say not consider “N patients 85+, M patients 65-84 and L patients 20-65”, but “A diabetes mellitus cases, B hypertonic cases, C multimorbid patients and E patients without chronic diseases”.
Usually, the correct data are not visible after the information leaves a hospital. The correct COVID-19 related information is simply not gathered by authorities. Which inevitably produces wrong conclusions.
For instance, the patients with darker skin color have lower levels of Vitamin-D3, which affects the immune reaction by COVID-19. The vaccination mysteries in the Israel may be solved if we consider that the propaganda against the skin cancer has caused overprotection from the sun, which has caused statistically significant low vitamin-D3-level in the population.
We cannot prove or disprove this theory because the data is not collected. AFAIK, a report form even contains a single checkbox with the description “was vaccinated or was already ill with COVID-19”. This checkbox affects the scientific conclusions about the effectiveness of the vaccine.
There are strict rules for data collection by a phase 3 drug trials. They are now completely broken. This genration of raw manipulated data has many other interesting features, but I better stop here.
There was the science in the previous century which was called “sovietology”, and it had allowed the analysts to make right conclusion on base of intentionally manipulated data. These skills are lost now. The CIA is unable not only to predict the North Korea’s nuclear program, but it also misses the basic economic estimations of this country. The manipulations in the Chinese statistics were not detected at the start of the COVID-19 pandemic, this inability had caused disastrous decisions and catastrophic strategies, which have affected the ways the data is collected and processed in other countries. The wrong data has produced the wrong political strategies, and this cycle cannot be stopped.
Such information is banned from social networks (I was banned on FB even for quite harmless remarks in German) and most doctors would not explain the problems they know about because such explanations could be considered as a propaganda against the vaccination, what could cause not only administrative actions, but also may be considered as a crime according to some recently issued laws.
It would be possible to study some funny effects of some local political decisions, but I do not believe that an article that proves or even mentions the existence of the problems I have described could be published in English.
Vit,
Much of what you say is interesting and I don’t see anything in it that I think is wrong, except the ‘just so story’ about sun protection in Israel. But I do think your comment lacks some perspective. You’re saying that all data sources related to the pandemic contain some inaccuracies, some of which are quite large and some of which are politically motivated. That is true. But you seem to be _implying_ that all data sources are hopelessly compromised and we therefore don’t know anything and can’t trust anything, and therefore any conclusion we make about what does or doesn’t work in controlling the pandemic will be wrong, and that implication isn’t true. We know a lot about how the virus is transmitted, how people catch it, how much risk it poses to various people, and so on. There’s lots that we don’t know, too, but the situation isn’t nearly as bleak as you suggest, when it comes to what we know.
So Phil / Joshua:
do you know of any direct evidence that masks have had any influence on the outcome of the pandemic? I have *always* worn a mask indoors in public, mandate or not, for nearly 18 months. I wear one because the cost of wearing one is small relative to cost of an infection. Just the same, I see absolutely no evidence that masks have had any influence on infection rates.
jim –
> do you know of any direct evidence that masks have had any influence on the outcome of the pandemic?
No. Being overly-semantic for a minute, I’m not sure how direct evidence would even be manifest. And I have no expectation of seeing direct evidence as it seems like an incredibly high bar.
Obviously, it’s a complicated subject for some kind of interventional study. I’ve seen lab-based studies that seem to show the potential for marginal benefit. And of course, there’s also some indirect evidence suggesting a potential marginal benefit as seen through comparisons between locations with different prevalences of mask usage (which is obviously highly problematic w/r/t controlling for confounding variables).
I’m not really expecting anything else. Seems to me that given the potential marginal benefit, combined with the compounding effect of a marginal (individual event) benefit at the population level, it’s worth it to wear a mask when in a public space, indoors or even perhaps under certain conditions outdoors.
As for your Hoover Dam analogy – that seems to me to be thinking of the mask as protection against “incoming” virus? But the theoretical main benefit is w/r/t “outgoing” virus – and maybe only in the sense that it might change droplet or aerosol behaviors even if not (only?) reduce the volume that escapes. I don’t think your analogy quite works even for “incoming” – as there could be circumstances where are you’re looking for is a marginal benefit in a context where it’s less than a “damfull” of virus?
IMO, this all goes back to looking at uncertainty with respect to mitigating “high damage function” risk – perhaps marginally. In this case, it seems to me that the potential downside risk is extremely small except perhaps in very specific circumstances – so it’s kind of a no-brainer.
“As for your Hoover Dam analogy – that seems to me to be thinking of the mask as protection against “incoming” virus?”
As I wrote it yes, but double the protection to 20% (one “mask” for outbound and one “mask” for inbound) and you have the same result. The point is the density of viral particles is astronomically overwhelming, so unless you increase the protective capability by an order of magnitude it’s unlikely to have a notable effect.
“In this case, it seems to me that the potential downside risk is extremely small except perhaps in very specific circumstances – so it’s kind of a no-brainer.”
Sure. As a preliminary position that makes sense. However at some point as the lack of evidence of effectiveness accumulates it has to be understood as ineffectiveness – all the moreso given that even the lab evidence is weak. So here we are 18mo into the pandemic with a variant twice as contagious and we’re still endorsing an intervention that has no known benefit. That’s astounding, because there **are** (were) other options: we could increase production of N95 masks, which **do** have a substantial benefit. I bought a supply of KN95 masks at the outset and have been using them the whole time. I just wash them, they come out great.
> I just wash them, they come out great.
Well, stop that right away.
The way these filtration masks work is that they develop a static electric charge among the various layer of fiber and then as particles pass through they have electrons stripped off as they pass the positive layers, and then become stuck to the negatively charged layers. When you wash them, even once, you basically destroy that entire effect.
Dry heat at 180F for 15 mins or so is enough to eliminate the virus, and ~ 1 day in a paper bag is also enough.
Reuse them by putting them in a paper bag and waiting 24 hours, you can reuse them for a month or more without problem unless you’re wearing them for long time in a fairly polluted environment.
DON’T wash them.
jim –
> As I wrote it yes, but double the protection to 20% (one “mask” for outbound and one “mask” for inbound) and you have the same result. The point is the density of viral particles is astronomically overwhelming, so unless you increase the protective capability by an order of magnitude it’s unlikely to have a notable effect.
I’m less convinced than you that there aren’t circumstances where the ratio of a given exposure to viral particles needed to become “infected” isn’t close to being the amount of particles a person is exposed to, and that a mask might make a difference there.
I’m hoping I’m not appealing to ignorance as that’s a fallacy, of course, but I do think there’s too much unknown to have much confidence one way or the other in some definitive fashion. That’s why I object to the assertion that much of an of this is “obvious.” In light of that, I look at the “fat tail” of risk against the minimal downside risk (I don’t see much of any such fat tail in the other direction of risk from wearing masks). And again, my point is that any marginal gain can potentially compound at the population level
> However at some point as the lack of evidence of effectiveness accumulates it has to be understood as ineffectiveness
?? I don’t understand that. Measuring the effect of wearing masks is hugely complicated. The fact that we haven’t closed the gap over time, in my view, doesn’t mean that masks are ineffective. It just means we haven’t invented some magical way of measuring the efficacy.
> – all the moreso given that even the lab evidence is weak.
What does weak mean? I’ve seen reports on lab experiments that show there is a marginal benefit to wearing masks – the one that Anoneuoid linked to (which does look thorough to a total non-experts) not withstanding.
> So here we are 18mo into the pandemic with a variant twice as contagious and we’re still endorsing an intervention that has no known benefit.
No proven benefit.
> that’s astounding, because there **are** (were) other options: we could increase production of N95 masks, which **do** have a substantial benefit. I bought a supply of KN95 masks at the outset and have been using them the whole time. I just wash them, they come out great.
That N95 masks would have a greater benefit doesn’t really change the equation for me for wearing cloth masks per se. But it does just so happen that I’ve just recently switched over to N95 masks exclusively – mostly on the thinking that they might be more efficacious in protecting against “incoming’ virus.
Of course, I agree that it’s irrational that as a society we haven’t made N95 masks more easily available – in fact free in huge numbers. Other countries have done so. I see zero valid reason that we haven’t.
Actually, the bigger puzzler for me is why we don’t have better testing, tracing, and isolation, and more widespread antigen testing, to be used at home or at least at places like schools. It makes absolutely no sense to me. I think it’s bizarre and very frustrating. The evidence of waning efficacy with the vaccines, and the unlikelihood that we’re not going to reach “herd immunity” via vaccines or otherwise only makes it even crazier that we haven’t developed those other ways of addressing COVID. I had hoped that with Trump gone, we might see more effort in those areas. Alas…
@Joshua
> Actually, the bigger puzzler for me is why we don’t have better testing, tracing, and isolation, and more widespread antigen testing, to be used at home or at least at places like schools. It makes absolutely no sense to me. I think it’s bizarre and very frustrating. The evidence of waning efficacy with the vaccines, and the unlikelihood that we’re not going to reach “herd immunity” via vaccines or otherwise only makes it even crazier that we haven’t developed those other ways of addressing COVID. I had hoped that with Trump gone, we might see more effort in those areas. Alas…
Well, it seems here is our turn on the arm chair
The problem with testing and tracing is that it costly, it could drain resources from other healthcare areas (ie a nurse doing testing is a nurse not doing something else). The efficacy of test-and-tracing depends on multiple factors. Test-and-Tracing would be useless if people are unwilling to cooperate, or, on the other hand, are too cooperative and inform their contacts and quarantine on their own. Finally, as the portion of vaccinated grows, portion of asymptomatic transmission is expected to raise too, so the test-and-trace is going to be less and less efficient.
Im not saying you should not test-and-trace. I have no idea, I dont know the whole context. I just saying it is not such a simple decision.
Mikhail –
> The problem with testing and tracing is that it costly, it could drain resources from other healthcare areas (ie a nurse doing testing is a nurse not doing something else). The efficacy of test-and-tracing depends on multiple factors. Test-and-Tracing would be useless if people are unwilling to cooperate, or, on the other hand, are too cooperative and inform their contacts and quarantine on their own. Finally, as the portion of vaccinated grows, portion of asymptomatic transmission is expected to raise too, so the test-and-trace is going to be less and less efficient.
>>Im not saying you should not test-and-trace. I have no idea, I dont know the whole context. I just saying it is not such a simple decision.
Overall, point well-taken. Yes, it’s complicated and I should follow my own advice and respect the uncertainties.
Just because practices have worked well elsewhere doesn’t mean they’d work well here. Particularly given the high likelihood of a lot of resistence to government initiatives, particularly if they’re implemented by a Democratic administration.
Still – in terms of expense, ideally it would be considered against potential payoff. Of course, that calculation would be complicated but reducing morbidity and mortality would net a huge return and maybe one much bigger than the amount invested. At some point limited resources such as nursing would have to be considered differently as there we’d reach diminishing returns quickly if resources are exhausted.
With the specifics of testing, I was thinking of (repeated) Antigen testing where all you’re looking for is someone who’s infectious. The whole way we’ve gone about testing, where we’re using medical testing (PCR) for public health surveillance is actually another issue that from my comfortable armchair, seems quite puzzling.
Phil:
I suspect there is a fundamental reason why the argument you presented above – that if a common mask stops 10% of particles, that’s better than nothing and should have an influence on infection rates – would fail. Imagine that you’re standing in the canyon below Hoover Dam. If all the water is released at once, how effective is protection against 10% of the flow? Not effective at all, right? You’d have to have protection from about 99.9% of the flow to survive.
To me the evidence suggests the release of virus particles by highly contagious people is roughly analogous to the Hoover Dam model. Large numbers of people infected in a few hours or less in enclosed spaces suggests a massive output of virus particles by the contagious person(s) – so much that common masks would be irrelevant.
Even in California it sounds like the mask is only worn 50% of the time you are in the restaurant, and in most places it is closer to 1%.
There is really no scientific basis for this practice, yet multiple people in this thread are trying to defend it (or at least refusing to call it out for what it is).
Phil,
I have already explained the validation problems in medical studies.
https://statmodeling.stat.columbia.edu/2021/08/22/what-happened-with-hmos/#comment-2000365
If there are clear intents to affect results by manipulating the raw data, your only solution is an independent measurement of the variables in question. You can hope that the people who produce the data will not lie intentionally (and even unintentionally), but you cannot prove this. You cannot even estimate which part of data was manipulated unless you fully control the measurement process.
A small wrong part of data can produce completely wrong results. 1% difference causes pretty small errors in linear models, but all processes we speak about are nonlinear and cyclically dependent. (For instance, a 1% rise of some variable causes a lockdown which affects this variable.)
We know that some data would be inevitably compromised, we do not measure it, and we even cannot estimate this part because the additional data that could allow such measurements is not collected. This means, we only affect but not control these processes, and we cannot predict unintended consequences.
For instance, we know, that a face mask can reduce the mean virus transfer. We know, that this protection is not absolute because we do not stop the air flow completely. We also know, that the probability to catch COVID-19 in a contact with a virus carrier is dependent on time, because humans periodically breathe in a new portion of air.
It is possible to make an expensive controlled experiment, which, say, states that in an open air environment by contact of two persons of the age group X who are placed on the distance Y for Z minutes the specific method to carry specific masks reduces the probability to be infected by 20%.
You issue an order to carry face masks. How would it affect the virus transfer in a real-life mixed population?
The correct answer is: We do not know.
People can carry masks on their chins, and this would reduce the protection effect to zero. People can purchase not the tested masks, but the masks with better filtration, and the effect will be much greater than the experimental results. People can decide that their masks 100% protect them, and consequently these people would reduce the mean contact distance, what would increase the virus transfer by 30%. An order to carry masks may produce a positive or a negative cumulative effect, or may not produce any significant effect at all.
Low vitamin-D3-levels in Israel are reported by vitamin-D3 blood tests. The doctors usually know this, but only a fraction of them could recognize the symptoms of an insufficient vitamin-D3 level without a blood test.
Vit:
You’re bringing up a good point which was discussed elsewhere on this thread, which is that in general it’s kind of impossible to untangle causal effects from individual behaviors. You say that people when they wear masks might then reduce contact distances, or there’s the opposite possibility that wearing masks makes people more aware of the unusualness of the situation and they will then be even more careful. In any data I think a much stronger pattern than either of these will be the correlation across people and over time, that the sorts of people who are more likely to wear masks will also be more careful in other ways, and the times when people where masks are the times when people will be more careful in other ways, hence I expect masking to be strongly correlated with reduced transmission. In that case, even if the causal effect of masking is zero I think we’ll be seeing a lot of it.
Another way of saying this is that the effect of a masking policy could be different from the effects of masks themselves. A masking policy sends the signal to people to be careful about reducing transmission. This might seem silly to you, to have a policy to send a signal, but in a sense that is what all behavior-influencing policies are about.
I’m not saying I like this reasoning, it’s just what it is. I was super-annoyed when they closed the basketball courts last year and took the hoops down, but I can understand the point of view from a policy perspective even if I doubt that there was any important transmission going on at these courts. It’s all very complicated, and I think that we need to think about direct causal effects of particular behaviors and also about effects of policies.
“Another way of saying this is that the effect of a masking policy could be different from the effects of masks themselves. ”
That *could* be, sure. But surely you know it’s just as much speculation as Vit’s suggestions about how people are likely to behave? :) It’s not even remotely worthy of policy consideration. I wish that weren’t true but it is what it is.
Jim:
I don’t think it’s speculation that to say that the effect of a masking policy could be different from the effects of masks themselves. I guess I could even replace “could be” with “will be” in that sentence. The differences go in two directions: First, a policy can have many effects, not just on this one behavior. Second, the policy can have little to no effect at all because a policy can ratify or coordinate behavior rather than changing it.
There’s a “science” or “personal behavior” question: What is the effect of me putting on a mask on my health and that of others? There’s a “population health” question: What is the effect of 10% more people wearing masks on the population spread of disease and health outcomes? There’s a “policy” question: What is the effect of requiring (or not requiring) indoor masks in public places on the population spread of disease and health outcomes? My point in these comments is just that these are three different things. I’m not making any specific claims about what these effects are.
Vit –
> Low vitamin-D3-levels in Israel are reported by vitamin-D3 blood tests. The doctors usually know this, but only a fraction of them could recognize the symptoms of an insufficient vitamin-D3 level without a blood test.
Why would you expect doctors to take an action, which could potentially produce a negative effect (say people taking Vitamin D might think they’re protected and thus take more risks), when the positive benefit is uncertain?
As with masks, so with Vitamin D.
Joshua,
you miss the point. You cannot compare apples with ducks because this is meaningless.
It is impossible to measure an amount of viruses that were inhaled and to predict their distribution inside. This means, it is impossible to measure face masks in real circumstances.
Such substances as vitamin D3, selenium, zinc, etc. are not silver bullets, but simply the bricks for the human immune system. (Details are not important, the biochemical cycles are too complex to explain them here.) An immune system cannot respond properly to a new threat if it does not have enough supply. Nothing more and nothing less.
Optimal levels of these substances in blood are known for ages. A measurement of the vitamin D3 in blood is a routine such as a measurement of the blood pressure, or of the glucose in urine. Good doctors had known about vitamin D3 from the beginning of the current pandemic. The number of tests had increased because it is important to detect patients with dangerously low levels and to prescribe them doses that are much more than the usual recommendations. Sometimes the levels must be increased with injections. This means, it is very simple to raise the level of this vitamin and to hold it near the optimal value. (Different doctors may have different opinions about it, but usually this is near the maximal border of the norm.)
Already the first studies have shown that the low levels of this vitamin mean statistically significant more deaths, but the information about a “wonderful vitamin” was released in the Swiss mass media only in Fall 2020. I think, most people here take this vitamin as a precaution. I have also heard about cases when a patient was too active by such self-protection. (An overdose of almost all substances is not healthy. An illness can be caused even by taking too much water.)
Of course, this is only one protective factor of many others. The number of patients who tries to fight their overweight or their high blood pressure is also significantly increased. However, the vitamin D3 is considered today as the most important factor of survival by COVOD-19.
If you do not know about the role of this vitamin, your life is not important for people who hold the information.
Vit –
> It is impossible to measure an amount of viruses that were inhaled and to predict their distribution inside. This means, it is impossible to measure face masks in real circumstances.
Well, I think it’s obviously extremely complex. So you do what you can to reduce the uncertainties and use the information you gain to judge probabilities. Just appealing to ignorance doesn’t work for me.
> Such substances as vitamin D3, selenium, zinc, etc. are not silver bullets, but simply the bricks for the human immune system. (Details are not important, the biochemical cycles are too complex to explain them here.) An immune system cannot respond properly to a new threat if it does not have enough supply. Nothing more and nothing less.
>> If you do not know about the role of this vitamin, your life is not important for people who hold the information.
I’m aware that vitamin D plays an important rule in our immune response. I take a supplement. In fact I increased it early on in the pandemic.
However, there’s been a fair amount of research on vitamin D supplementation as a prophylactic and/or therapeutic for COVID, and from what I’ve seen there isn’t a compelling case as of yet. If you do a search at this site, you’ll see Andrew did a post.
I’m not inclined to just take your word for it, but thanks for the advice.
Joshua:
Please do not mix a complexity with an impossibility. It is impossible to make measurements with better precision than the precision of the available measurement methods. This impossibility is fundamental and cannot be solved with money, quantity or any mathematical tricks.
Any estimation is meaningless if the estimated value is comparable with the variation of the variable in question, and the distribution of this variation cannot be determined because of imprecision or because of lack of data.
Vit-
> It is impossible to make measurements with better precision than the precision of the available measurement methods.
As it happens, I’ve actually spoken with someone who’s working with a team on developing better measurement techniques. Obviously, they don’t share your certainty that better techniques aren’t possible. What was described to me sounded interesting and seemed plausible.
Again, thanks for sharing your opinion but I’m not inclined to consider your opinion as fact.
Vit –
> The people in a hospital which is generously paid for COVID-19 cases will act differently in comparison to the people in a hospital which will be closed, if a single COVID-19 case will be found in it.
You state that as a universal fact. I’m skeptical. Do you have an evidence basis for making that statement? Are you just saying it happens sometimes and you have evidence? Are you saying it might happen?
If you believe that the uncontrolled people are dumb enough to handle against their interests, I would not disturb your arrogance. You are not alone. This is the ground, why all attempts to mix democracy with socialism have inevitably failed.
You may remember Orwell’s “1964”. The information propagation follows today the main socialistic rule: If you do not know some information, you have no permission to know it.
NASA had recently lowered the 2016 global average temperature to advertise the year 2020 as the “hottest year on record by narrow margin”. If you are lucky, you could find some scientifically sounding explanations, why it is correct to change in 2021 the measurements data from the year 2016. I doubt, you could find a single explanation, why this is a fraud.
Of course, I know many cases in different countries. Of course, I will not share this information. Today’s laws consider this as a dangerous propaganda.
Vit:
You say, “all attempts to mix democracy with socialism have inevitably failed.” I don’t think there have been any attempts to mix democracy with pure socialism (the socialist republics were not at all democratic), but the U.S., Britain, France, Germany, etc., are democratic and have mixed economies, i.e. they combine democracy with a mixture of capitalism and socialism, and they’re not perfect but I think “failed” is a bit strong.
Regarding your last point, there are lots of reasons to want to increase the capitalist percentage of our economy and decrease the socialist percentage; I think you can make this case on its own terms without bringing in climate change denial. Climate change is physics, and political rhetoric isn’t gonna stop it from happening.
Andrew,
thank you for your comments. They help to reshape my translation and give me additional insights.
Look, I talk about the information filtration, and you react with an urge to deny a completely unimportant example.
This was simply the latest funny thing I have discussed with the people who know. I could use any other example to explain the same problem. Of course, I could use the funniest story and replace NASA with Trump, but in this case nobody would discuss the abstract problem of the consequences of a possibility to cut off even a President of the United States (who was some time ago considered as the most powerful person on this planet). An unimportant name will cause a strong emotional reaction which would destroy the main theme and which would force all participants to report their personal opinions about Trump.
This is not science, this is psychology.
The same situation is with the face masks. The politics is not a memoryless Markov chain of independent events. People adjust their reactions on new political initiatives dependent on their previous experience (if they do not forget the history).
The politicians of the European World could not say: “Hey, people! The evil virus about which we have talked you several decades is finally here. But — guess what? — we are not prepared. We even do not have for you most basic protection. Piss off and die.”
The politicians had searched for a solution and the modern science had offered its help.
You can find today scientific proofs for almost everything. If there are no ready proofs for the theory you need, you can order them (if you have enough money and enough administrative connections).
The politicians had told the people that the face masks are evil and they had mentioned many scientific articles with undeniable scientific proofs. I remember how Chancellor Angela Merkel had explained to Germans that it is enough to wash hands and to sneeze into the own elbow bend to build a perfect protection against COVID-19. (I consider this as a massive diversion.)
Some people had ignored these insane reasons, some people had gotten curious, some people were imprinted with the early phase science that had denied face masks.
The pandemic did not stop and the spread of the virus SARS-CoV-2 got out off control. The politicians had searched for a “sure solution” and they were pleased with the Chinese dictatorship.
People who know the basics, would never believe any Chinese numbers and any Chinese reports. Other Asian countries such as South Korea had better strategies, but the politicians always understand only the simplest ways and do not give a dumb about unintended consequences.
The mandatory face masks were present among other wonderful Chinese methods.
The same politicians had issued orders, but they were also in need to explain that their actions are for people’s good. They had searched for a solution and the modern science had offered its help.
The politicians had told the people that the face masks are an ultimate solution and they had mentioned many articles with undeniable scientific proofs.
Some people had ignored these insane reasons, some people had gotten curious, some people had forgotten the former science and were imprinted with the new science, some people were caught in the early phase science that had denied face masks.
The last two groups are destined to participate in endless scientific arguments. Logical reasons cannot affect the beliefs that are hidden on the unconscious level of mind.
Sorry, it is meaningless to discuss socialism unless you have some relevant personal experience of living in a socialistic country. You simply would not comprehend my explanations. I think, the U.S. citizens need at least 10 more years to understand the basics.
If you rally are interested in this question, there is a simple but only partially correct explanation about an exchange and a distribution. You can allow the people who control an exchange to be greedy without destroying the system, but you must be sure that the people who control a distribution of any resources among other people do not use their position for personal benefits. (If they misuse their power, you simply have created a perverted version of an exchange.)
Note, I do not mean a simple corruption. Most benefits in a distribution do not have any monetary value because the money itself has in any kind of socialisms not the same meaning what it has in an exchange. It is necessary to properly study Marx to understand this. (Most interpretations of Marx’s works are incorrect.)
There is a very long list of ideas that could solve the problem of imperfect people who are in control of distribution. All these ideas do not work. Only two forces can protect a distribution system against massive misuse: You can select between faith and fear, or choose both of them.
The distribution works perfectly in a monastery or in a religious terrorist’s group because these people have faith in their beliefs. The distribution works quite good under massive terror. However, there is a need for an organization which distributes terror among people who distribute resources among normal citizens.
An interesting example is the North Korea. The Western point of view is wrong. It is now not a socialistic country, but a specific kind of national-oriented monarchy with socialistic elements. These elements are pretty big and consequently the citizens are forced to have faith in their dear Leader and to have fear of the country’s security services. It is not clear, which of the forces is prevalent.
It is possible that a small instability in the country would cause a rebellion. It is possible that any kind of severe problems in the country would only force people to offer their lives to protect the Kim’s rule.
The citizens follow rituals and show their love to the dear Party and to the dear Leader. Sometimes citizens protest against an economic policy, but I have not heard that they have demanded more freedom. It is completely unknown, what they really think about Kim and about the political system.
I have discussed this problem with people who study the country. This means they not only read some documents, they have visited North Korea and they had many personal contacts with (former) citizens and officials. Nobody knows how stable is the Kim’s system. This means, nobody in the world knows this for sure, including the North Korea’s top officials.
If you have sharp observation skills, you could detect the signs of faith and fear in the current situation among you, because the preventive actions against COVID-19 have introduced a lot of socialistic elements.
I have many friends in your country, none of them had noticed any increase of democracy.
Two cents about physics. It is an old saying. If somebody tells a mathematician that the Earth is flat, the mathematician says: “You are an idiot.” If somebody tells a physicist that the Earth is flat, the physicist asks: “What reasons do you have to make this conclusion?”
By the way, you must decide what you are fighting: the Global Warming or the Climate Change. Unfortunately, they are incompatible: There are no hurricanes in a greenhouse. However, there is a solution in Orwell’s “1964”. (^_-)
Vit –
> If you believe that the uncontrolled people are dumb enough to handle against their interests, I would not disturb your arrogance. You are not alone. This is the ground, why all attempts to mix democracy with socialism have inevitably failed.
I’m glad to have my arrogance undisturbed!
Yah, I happen to have real work experience observing people act on motivations outside of merely personal, short-term financial gain at the individual level and even at an institutional level. I don’t doubt that sometimes short-term financial gain dominates.
That’s ‘why I asked if you were asserting some kind of universal principle and if you had evidence in support. Unfortunately, you didn’t answer my question and instead preached your political philosophy – which actually seems mostly a non-sequitur to me.
That you went on to link your opinions on a vast conspiracy regarding global temperatures doesn’t increase your credibility, IMO. I’ve observed debates about global temps quite a bit, and in my view the claims of “fraud” don’t stack up particularly well.
A couple questions:
> Similarly, the uncertainty stemming from low case rates constrains the ability to make informative comparisons across time and space, for instance, to identify the causal impact of specific mitigation measures or environmental drivers, such as temperature or air pollution. Individual heterogeneity and the potential for superspreading events can be accounted for using a negative binomial distribution for modeling the number of secondary cases.
The first sentence makes me think the problem is unsolvable, and the second sentence makes me think at least part of the problem has been solved. Is there a third sentence that might clarify what is happening here? Both those sentences have papers cited, so I assume the details are somewhere in there, but tldr; :(.
> Following this and other failed attempts at prediction, people have mostly given up on forecasting the incidence of COVID-19 beyond a few weeks.
> after a year in it is now more widely understood that the incidence of COVID-19 cases and deaths at a given time and place depends on too many converging factors to allow useful forecasting.
So are these few-week predictions hopeless? What about predictions on average (as opposed to time and place)?
I dont think prediction on more than a few weeks are hopeless, they are just… well, you should not expect then to be accurate. Quoting Galadriel:
Remember that the Mirror shows many things, and not all have yet come to pass. Some never come to be, unless those that behold the visions turn aside from their path to prevent them. The Mirror is dangerous as a guide of deeds.