Is the "diseases and pandemics" thread doing more harm than good?

[grant hutchison]

Well, the title says it all.
And you can probably guess my own feeling from the fact I've made this post.

Grant Hutchison

[Shaula]

I think it was useful when people were using it to ask for help understanding technical reporting. But that was basically dependent on one or two people (mostly Grant!) sharing their medical knowledge and experience. Trouble is as the volume of reporting has gone up and peoples' familiarity with the subject has increased its kind of turned into something else. I think picking the brains of professionals in the area is great and kudos to them for helping out. I think that having their expertise written off by anyone with access to Excel and the internet has to reduce their motivation for that and without that support the thread is indeed more harmful than useful.

[Roger E. Moore]

I think it was useful when people were using it to ask for help understanding technical reporting. But that was basically dependent on one or two people (mostly Grant!) sharing their medical knowledge and experience. Trouble is as the volume of reporting has gone up and peoples' familiarity with the subject has increased its kind of turned into something else. I think picking the brains of professionals in the area is great and kudos to them for helping out. I think that having their expertise written off by anyone with access to Excel and the internet has to reduce their motivation for that and without that support the thread is indeed more harmful than useful.

Here here. I started the thread because I wanted to share some news I was seeing about Ebola and Zika, which worried me, and suddenly boom now we have a pandemic. I still like sharing news and I enjoy reading medical opinions now that my personality disorders have been squashed, to everyone's relief. So I do still like having it around, but... YMMV

[schlaugh]

It's been like any other smorgasbord - but without the Swedish meatballs. Lots of good content mixed with some material that didn't really fit in the category, but broadly I find the thread educational. And a good reminder that science, especially epidemiology, is both complex and complicated. As said in the thread - it's messy.

[George]

I think it was useful when people were using it to ask for help understanding technical reporting. But that was basically dependent on one or two people (mostly Grant!) sharing their medical knowledge and experience. Trouble is as the volume of reporting has gone up and peoples' familiarity with the subject has increased its kind of turned into something else. I think picking the brains of professionals in the area is great and kudos to them for helping out.

Indeed, and the amount of time folks like Grant and others in the medical field have given to this thread is greatly appreciated. [I thought the virion was a living organism before this thread started and I have no problem reminding others of my short comings; before honor comes humility.]

MI think that having their expertise written off by anyone with access to Excel and the internet has to reduce their motivation for that and without that support the thread is indeed more harmful than useful.

You may not be alluding to my Excel work, nevertheless my graphs, and the Excel work of others, are meant to make use of what can be done in an effort, at least, to add value to the thread, though this is secondary to my own personal interests. Others, IMO, are also trying to use data to look ahead. Isn't that what science does and isn't this a science forum? Such things may need greater scrutiny than, say Planet 9, but that's Science. My guess is that time and effort to apply proper scrutiny has become a greater-than-normal burden on a few great participants, and I regret that. It would be nice to have a few top epidemiologist join us, and they should, but I won't hold my breath.

Regardless, to help make your case regarding you Excel concerns, the "hockey stick" chart from the New Atlantis demonstrates a use of statistics that could easily, and inappropriately, cause some to fear that this pandemic was exponentially getting out of hand. The updated chart shows a very step decline. [I just updated it again since the data is in weekly segments, and the latest projection is for an even steeper decline. Integrating both Covid-19 curves will produce the estimated total no. of deaths and, and the current rate, it should come close to the 80k estimate (60k, I think, is the more official US estimate) on the graph.

If, by chance, it is my graphs that are not useful, or worse, don't worry about my feelings; good science isn't about feelings and my skin is reasonably thick. Criticisms of them are welcome and I will either understand the scrutiny applied or just quit presenting them. [I mention this partly due to the lack of comments by some who should understand them the most (good or bad).]

[George]

It's been like any other smorgasbord - but without the Swedish meatballs.

Nice metaphor since the Swedish data plays a meaty role.

[schlaugh]

Nice metaphor since the Swedish data plays a meaty role.

That may be the first time in my life that I made an unintentional pun.

[Swift]

<snip>
Others, IMO, are also trying to use data to look ahead. Isn't that what science does and isn't this a science forum?

(Please note - not in moderator purple....)

Aye, but there's the rub...... I'm not picking on you George, but that sounds a lot like what we hear from a lot of ATMers. They take a few bits of information, combine them together to "prove" some idea they believe in, and when they get questioned about it say "Hey, but isn't that what science is all about". Science is also about knowing the limitations: the limitations of one's data, of one's ability to analyze the data, of comparing that data analysis to what others in the field have come up with, and not forcing the data beyond those limitations.

We all seem to agree that at this point there are as many, if not more, unknowns about this virus than knowns, that the data set is very incomplete. Unfortunately, we've seen considerable cherry-picking of that data by certain individuals to prove some point they were trying to make. As I say too often, sometimes the mainstream answer is "we don't know", even if that is unsatisfactory. However, on CQ, "we don't know" is not permission for wild speculation.

[Noclevername]

It started as a useful thread "because it's science". It's no longer science, mostly. The thread has largely devolved into arguments about what the data even is, or how it should be used.

[Shaula]

Others, IMO, are also trying to use data to look ahead. Isn't that what science does and isn't this a science forum? Such things may need greater scrutiny than, say Planet 9, but that's Science.]

You know those people who come in here and tell us aerospace engineers are all wrong and the lunar landings couldn't have happened, then go on to 'prove' it using 'logic', basic trig and some choppy footage they found on YouTube? That's probably what we look like to medical professionals in that thread.

Visualising existing data to help support conclusions drawn by professionals is good. Doing our own analysis on data we don't really understand is bad.

Edit to add: And this just struck me. If someone wants to provide analysis on GR, QM or aerospace topics here (and especially if they are arguing against the current consensus in physics) one of the first things we often say is "You really need to study this topic in more depth, it usually takes a motivated person a year or two to get up to speed - and if you want to do this ask and people can provide resources to help you do that". Seems like that recognition of appropriate levels of expertise doesn't apply anywhere except physics, eh?

[Jens]

My feeling is a little bit different. A lot of people are (naturally, I think), worried about this, and we get a lot of information. We don't only get information from experts. In many cases, we get information from journalists who are also not experts, interpreting what experts have to say. And everyone is (rightly, I think) wondering about how things are developing. And it just isn't that easy to get really good information (partly because there are a lot of unknowns) so I think that people will naturally consider various scenarios, and I think it's great to have Grant around to point out the limitations of different ideas. I realize it is a drain on his time and energy, and certainly frustrating, but we're in a situation (most of us) where significant limitations have been placed on us because of this, so although I realize there are issues with the thread, I do think it's is useful.

[profloater]

Well the science references on this thread have, I feel, helped me understand the situation much better. I do ignore a lot of the comments which are not based on useful references. I am particularly grateful to Grantfor his medical advice and I think I understand his frustration at amateur epidemiology.

[grant hutchison]

I understand that people are anxious and feel that they've lost a locus of control. This tends to make some people feel a little paranoid, at least intermittently, and also drives some people to work away, more or less obsessively, at things they can control.
This is not helped by the fact that "we don't know" really is the only valid answer to many of the questions people want to ask, and by the fact that, in medicine particularly, "we don't know" manifests itself by the generation of frustrating little clouds of conflicting data.

To me, the current state of discussion on the pandemics thread brings together all these aspects. And it seems to me that people are to some extent addressing their own anxieties at the expense of other people's--by hunting around and posting marginal tit-bits of information with inappropriate interpretations attached, doing their own home-cooked "analyses" by combining complicated data in simplistic ways, and suggesting that "we don't know" reflects some sort of failure or negligence on the part of people who are working hard and conscientiously on the problem. Winding each other up, in other words.

The little "survival guide" I posted was intended as a way for people to get some locus of control back by taking a systematic approach to new data, particularly if they find it worrying, and perhaps thereby cutting down on the amount of alarming misinterpretation being shared on the thread.

Grant Hutchison

[Cougar]

Yes, Grant's contributions are well taken and appreciated, but I find it odd that the same appreciation is not proffered for Ken's contributions in this thread. In fact, he gets infracted and suspended for what appears to me overly restrictive moderation. I mean, I don't think he suggested to anyone that injecting disinfectant might be a treatment for Covid19! I've always found Ken's postings to be well considered, based on a very solid educational background, thoughtful, and thought-provoking. This is a science discussion community after all. I am reminded of another very smart, very well educated contributor who was banned many years ago - Carl Friedrich Gauss, I think his handle was. I wanted to hear his opinion on things. But no, he was promptly excommunicated, and I think the site was the poorer for it.

But this is the sort of contribution that Ken gave to the coronavirus discussion. This seemed quite interesting to me. And he was suspended for later going a little off topic? And I certainly didn't quite follow his more recent suspension - for clarifying some statistics? It seems to me this site has jettisoned some valuable resources in the past. Evolutionarily, I don't think this is a good survival strategy.

Looking at the case curves, I am starting to think the "SIR" model is making some importantly wrong assumptions. The SIR model has s be the fraction of susceptible people, i the fraction of infected people, and r the fraction of recovered. Of course s+i+r=1. The key equation assumes that each of these classes involves identical people behaving as per their class, and the transmission is scale invariant, so there are infection and recovery rate coefficients A and B that don't depend on s, i, or r. Then the equation one gets is
di/dt = Ais - Bi
on grounds that you need binary encounters between the i and s subgroups to get new i members, and the i members are resolving at a fixed rate B. The key parameter is R0 = A/B, and the result that connects those who never got it (s) from those who have recovered (r) is
s=e^[-R0*r]
and since i --> 0 eventually, in the asymptotic limit we have r =1 - e^[-R0 * r] as the number of resolved cases after the epidemic is over. An important feature of this model is that the final r is not 1, but for typical R0 ~ 2, it is something like half. The model is so crude one cannot conclude much more than that from it.

Of course what one really wants to know is not the final r, but rather, the final r*D, where D is the death fraction. But assuming D does not depend on r (which is only true if the health care system is not overwhelmed), you can know D independently, so r ends up being the key statistic anyway. When D is ~1%, as may be true for COVID-19, one cannot tolerate a final r that is like half, so R0 has to be attacked directly. Hence, lockdowns.

This also means that the SIR model is no good at all for what we have. It's only good for tracking what happens if you don't do anything, since then you can treat R0 as a fixed parameter. What we actually have is a situation where R0 receives feedback from i, because when you have a lot of cases, you take action that reduces R0. There's no easy way to model how R0 responds to our actions, so we have no choice but to use trial and error. We just do stuff, and watch the resulting case rate, and if we don't like what we have, we do more. So the key thing we need to know is not R0, but rather, what is the time lag between what we do, and when it shows up in the case rate. That is controlled by the B parameter, it has nothing to do with R0 because it doesn't depend on the A parameter at all. The A parameter is the inverse of the timescale for cases to resolve, which seems like about 3 weeks from infection to resolution. So we lockdown, and look at what happens to the case load 3 weeks later. Herd immunity plays no role because it's not high enough and that's not what controls the case curve-- we can rewrite all our equations with r=0, which simplifies them a lot because then s=1-i and the equation is
di/dt = (A-B)i - iAi
(I write it that way in case i is a vector and A a matrix). Also, we can't control B, so we should define a new time coordinate, call it x=Bt, and the equation with low levels of herd immunity is
di/dx = [R0 - 1] i - i[R0]i
But here's the real point-- the SIR equation is designed to treat situations where both r and i are quite high. That would be a catastrophic situation for COVID-19 unless there is huge asymptomatic numbers. Assuming that is not the case, we should also assume i << 1, and the equation becomes
di/dx = [R0 - 1] i
which has a very simple exponential growth solution if R0 is constant. But it isn't constant, that's the whole point of the new way of thinking about this equation that I am espousing. We should instead think of R0 is a function of i (and probably a 3-week lagged function of i, but let's leave that out for now). The simplest way to make R0 depend on i is to say R0 should be made to drop as i rises, so let's say R0 = R(i0 - i) where i0 is a hypothetical level of infection that is so bad everyone makes it their entire priority to prevent any contagion whatsoever.

This model has two interesting benchmarks for infection levels. One is i_max - 1/R, which is the infection level that stays fixed (and might be somewhat like what we are seeing in almost all western nations right now), and half that is the level where the infection rate is the inflection point. I would argue this model is better for what we are seeing, because the inflection and peaks are not coming at points of very high i, which is what you get in the "SIR" approach. But note the key difference-- in this model, once i gets to the constant level, it stays there-- it does not drop like in the SIR model.

[Shaula]

Yes, Grant's contributions are well taken and appreciated, but I find it odd that the same appreciation is not proffered for Ken's contributions in this thread. In fact, he gets infracted and suspended for what appears to me overly restrictive moderation. I mean, I don't think he suggested to anyone that injecting disinfectant might be a treatment for Covid19! I've always found Ken's postings to be well considered, based on a very solid educational background, thoughtful, and thought-provoking.

I don't often disagree with you, Cougar, but I definitely have to here. In general the assessments presented (and Ken was not the only one doing this) were built on analysis that was not fit for purpose. And they drove a thread that was useful (by helping people read and understand the popular articles and the technical articles about the disease) to one that was not useful (because all of the energy was being put into arguing about unreviewed, non-expert analysis of a complex topic). That's not tolerated for physics threads and it shouldn't be tolerated for medical threads.

[grant hutchison]

I'd suggest that, if the thread has a purpose at all, it's to help each other understand the complicated and rapidly evolving literature relating to the current outbreak, and perhaps to point out the things we actually don't know, why we don't know them, and how we might come to know them.
Pushing simplistic "analyses" based on poorly understood figures and demonstrably flawed assumptions is something else again. And when it's combined with remarks like:

So no, I'm not impressed by what the experts are saying, we need a great deal of oversight on our own part because the numbers they are throwing around not only don't agree, they don't really make much sense.

it has departed from any sort of attempt to understand the conflicting data being presented, and has moved instead to poisoning the well.

Grant Hutchison

[Roger E. Moore]

I'd suggest that, if the thread has a purpose at all, it's to help each other understand the complicated and rapidly evolving literature relating to the current outbreak, and perhaps to point out the things we actually don't know, why we don't know them, and how we might come to know them.

Yes, please. This has been an enormous help to me in interpreting confusing documents flying a bit over my head. My thanks to Grant for all your help. Please keep the thread going.

[George]

(Please note - not in moderator purple....)

Aye, but there's the rub...... I'm not picking on you George, but that sounds a lot like what we hear from a lot of ATMers. They take a few bits of information, combine them together to "prove" some idea they believe in, and when they get questioned about it say "Hey, but isn't that what science is all about". Science is also about knowing the limitations: the limitations of one's data, of one's ability to analyze the data, of comparing that data analysis to what others in the field have come up with, and not forcing the data beyond those limitations.

We all seem to agree that at this point there are as many, if not more, unknowns about this virus than knowns, that the data set is very incomplete. Unfortunately, we've seen considerable cherry-picking of that data by certain individuals to prove some point they were trying to make. As I say too often, sometimes the mainstream answer is "we don't know", even if that is unsatisfactory. However, on CQ, "we don't know" is not permission for wild speculation.

Agreed.

It would be nice if there was a way (separate thread?) to allow respected mathematicians/physicists to apply their skills with statistical analysis to help address respectable claims of others, hopefully experts. It's not hard to see that NYC is struggling far worse than Santa Clara and perhaps there may be "patterns" that may surface in an analysis that might help us explain those differences. I can appreciate such pattern-seeking can quickly begin to lean toward the ATM even when the data analysis is correct and not actually contrary to mainstream. I also appreciate the idea of keeping the main thread less speculative in analysis for several reasons.

So....would it be wise to have a separate thread that allows honest analysis that might be deemed too speculative and detrimental to the integrity of the main thread? This thread would be "woven" around the main thread. For instance, consider the advantages if I could periodically post into the main thread a simple link with a one-sentence comment to allow a way to keep my graphs off the main thread and into one that allows others more freedom, if they so choose, to address the implications that come from, say, rate changes.

[geonuc]

I'll echo the general appreciation for the useful and measured input from our medical professionals. But as this is chiefly an astronomy forum, I find the amateur analyses of the various aspects of this pandemic less than useful. Frankly, if I want in-depth analysis beyond what I'm reading on reputable media sources, I'll look to other forums where there is more relevant, knowledgeable expertise. Reddit has a COVID-19 sub, for example. I'm sure there are others.

[Swift]

<snip>

So....would it be wise to have a separate thread that allows honest analysis that might be deemed too speculative and detrimental to the integrity of the main thread? This thread would be "woven" around the main thread. For instance, consider the advantages if I could periodically post into the main thread a simple link with a one-sentence comment to allow a way to keep my graphs off the main thread and into one that allows others more freedom, if they so choose, to address the implications that come from, say, rate changes.

This would have to be a question for the Moderation Team, but my first response would be "No". That sounds entirely too much like the many proposals we've had over the years for a more speculative, less restrictive ATM sub-forum, that would allow for all the speculation without doing the work to demonstrate it.

And, as Geonuc points out, we are primarily a physics/astronomy/space exploration forum. We don't really have the expertise to police such a forum.

Lastly, I don't think any of this is solved by a separate thread. The problem, at least as far as I am concerned, wasn't too many posts or too much content in one thread, or off topic posts. The problem was posts that were too ATM for CQ, wherever they were located.

[grant hutchison]

Frankly, if I want in-depth analysis beyond what I'm reading on reputable media sources, I'll look to other forums where there is more relevant, knowledgeable expertise. Reddit has a COVID-19 sub, for example. I'm sure there are others.

Hear, hear. If people really want to discuss epidemiology in detail, they need to find a place containing epidemiologists, and not try to reinvent the wheel, using unsuitable components, on an astronomy forum.
I am not an epidemiologist--but I have sent a happy summer training in epidemiology in a specialist centre, and have a tiny peer-reviewed publication history in that discipline. With that small experience behind me, I absolutely do not feel qualified to deliver any kind of epidemiological verdict on what's going on at present. So I've restricted myself to what may admittedly seem a very negative role--pointing out any basic errors I notice, or drawing attention to why things are more complicated than they may at first seem.
Yet I'm seeing the work of clever and vastly experienced people, for whose previous work I have immense respect, being dismissed out of hand on the basis of a little bit of simple modelling. I think we'd be better off trying to think about why their data are the way they are, and why they have formed the opinions they have.
There's an excellent opinion piece here, which eloquently sets out what I believe as well. The authors say a lot of wise things, but this one seems salient to me:

When the dust settles, few if any scientists — no matter where they work and whatever their academic titles — will have been 100% correct about the effects of Covid-19 and our responses to it. Acknowledging this fact does not require policy paralysis by local and national governments, which must take decisive action despite uncertainty. But admitting this truth requires willingness to listen to and consider ideas, even many that most initially consider totally wrong.

Grant Hutchison

[Roger E. Moore]

Frankly, if I want in-depth analysis beyond what I'm reading on reputable media sources, I'll look to other forums where there is more relevant, knowledgeable expertise. Reddit has a COVID-19 sub, for example. I'm sure there are others.

YES. What he said. Not here, let Reddit do that.

[Shaula]

It would be nice if there was a way (separate thread?) to allow respected mathematicians/physicists to apply their skills with statistical analysis to help address respectable claims of others, hopefully experts.

Do you think there is a lack of these skills in the medical community? Or that mathematicians/physicists have some special insight into this kind of analysis?

Many of the best statisticians I know have biomedical backgrounds. They routinely deal with far more complex and messy data than physicists are used to. I have a physics background and I am happy to admit we don't bring anything special to this party.

I have to agree with Swift too - you are basically asking for an ATM-life area. Which has been resisted for physics and astronomy, our core areas, so should definitely be resisted for areas where most of the forum are complete amateurs.

[grant hutchison]

Yes, you want a medical statistician for a medical statistics job. It's not just a matter of doing some sums, it's understanding the nature and limitations of the data. My Wife The Professor occasionally did research that interfaced with physicists. Having not the slightest physical or mathematical intuition in her entire body, she was initially a little intimidated by them--but after a while she noticed that they were very much out of their depth when it came to the subtleties of collecting and analysing biomedical data, stuff that was entirely second nature to her. It's just a different skill set--but it's an error to imagine that the difference doesn't exist.

Grant Hutchison

[profloater]

I think it is worth mentioning that a completely different skill set is at work here for the decision takers in society. They have to balance the medical needs against the psychology of groups and individuals in society. I am referring to what is politically possible in various kinds of society, and with the scale factors that are obvious for cities and towns. Politics is the art of the possible. If you imagine modelling an epidemic, The estimated behaviour of groups and individuals under instruction and with their own personal judgements taken into account greatly complicates the short-term and long-term outcomes. And it is surely too soon to judge the outcomes in this pandemic!

[profloater]

In other words, the “behaviour” of the virus may be governed by physics until it gets inside a person, then it is complicated and the behaviour of people is governed by emotion with a touch of rationality plus attitudes to law enforcement and various time dependent changes, so the physics part is limited but still interesting.

[Swift]

I think it is worth mentioning that a completely different skill set is at work here for the decision takers in society. They have to balance the medical needs against the psychology of groups and individuals in society. I am referring to what is politically possible in various kinds of society, and with the scale factors that are obvious for cities and towns. Politics is the art of the possible. If you imagine modelling an epidemic, The estimated behaviour of groups and individuals under instruction and with their own personal judgements taken into account greatly complicates the short-term and long-term outcomes. And it is surely too soon to judge the outcomes in this pandemic!

Let's not use Feedback for this discussion.

[profloater]

Let's not use Feedback for this discussion.

Sorry! I got, or get confused about which thread to use.

[George]

I have to agree with Swift too - you are basically asking for an ATM-life area.

That's understandable, no doubt, and I respect that even if our analysis is astute, we don't know what the experts know and understand.

Which has been resisted for physics and astronomy, our core areas, so should definitely be resisted for areas where most of the forum are complete amateurs.

But what about when someone, assuming a professional not a blogger, makes what appears to be a math error? For instance, if an authority claims an expected death rate of, say, 0.1%, but based on the actual no. of deaths and the likely no. of cases the math reveals a more likely result of something much closer to 0.5%. Would it be not proper to have a place, even the ATM, to at least work through it and look for an explanation, like the no. of false positives/negatives for the testing being used?

[Shaula]

But what about when someone, assuming a professional not a blogger, makes what appears to be a math error? For instance, if an authority claims an expected death rate of, say, 0.1%, but based on the actual no. of deaths and the likely no. of cases the math reveals a more likely result of something much closer to 0.5%. Would it be not proper to have a place, even the ATM, to at least work through it and look for an explanation, like the no. of false positives/negatives for the testing being used?

Contact the journal it is published in? Go back to the source and work through that example?
How about a medical forum where you are likely to get more informed opinion on whether you have caught an error or just don't understand the analysis?
Pretty much anything other than post a series of simplistic toy models and assumptions about the data set then proceed to defend the output of this as if it were something precious by attacking the peer reviewed work of professionals.

Or, in the specific example you are talking about, recognise that you are trying to do analysis you are not qualified for and either a) get the right background to do this or b) leave it to the professionals. I'm sorry but your insistence that we (as a group of people united only by our interest in astronomy and physics) have something profound to contribute to a complex topic when most of us have not taken the time to study it to the requisite level is exactly the attitude you see in the CT section every time someone uses a little bit of maths and 'proves' that Apollo couldn't have made it through the radiation belts or that the flag means it was filmed on Earth.