Well, maybe not so fast. It’s not really the point (more on that below), but take a close look at what he’s doing here and what was actually said in the thread.
Take a look at the y-axes of the graphs he posted above. Now imagine Taiwan on the same scale as Sweden and think about what their curve would look like. As I know you know, it would look essentially like a straight line.
Now, who’s the one being disingenuous here? All of us who have any grounding at all in scientific training understand that reports can be manipulated to give such false impressions, and that it is deeply irresponsible to do so. Let’s just say that one should be careful about drawing conclusions from the Taiwan data set, as the statistical power will be a little less than robust.
But again, let’s look at what was said in the thread. It was a good discussion – really, a more open-minded and fair one than many others that have come along since – about what effects lockdowns have and whether they are worth their costs. The following claim was made:
Before you decide that there is no way to avoid the virus blasting through the population, maybe take a look at the curves of other nations. See what they have achieved without vaccines or antivirals, and in many cases with very weak healthcare systems and poor testing and tracking capacity.
Just with lockdowns and social distancing.
(I don’t know what criteria that site uses, or why Bhutan is “winning” while south Korea is not. But it is a good collection of graphs, and they are the information I wanted to convey.)
Now, I think the analysis offered in that claim would not hold up to great scientific scrutiny. The poster says, essentially, ‘Take a look at this page of curves. Proof positive that lockdowns and social distancing alone are enough to get a handle on this thing’.
I don’t know if that’s supposed to pass for good argument in QZ. First of all, there was no indication from the page of curves or in the person’s post of what specific lockdown and social distancing measures were put in place in the various countries. He himself seems to express some uncertainty at the end of the post. But he seems to think there is a real takeaway from the page of graphs that proves the effectiveness of interventions.
The angle I was coming from, and it’s a stance I have not seen much good reason to change since, is that it’s really, really hard to demonstrate effectiveness of such NPIs, given all the confounders. (And Stranger does a good job in the thread of explaining the reasons why. He seems more prone to caution than I might find warranted, but there’s no one I’ve seen on this board who is so dispassionately analytical.) And also that the course of the epidemics seem similar in most localities, despite variances in approach to interventions, with a surge lasting somewhere from four to eight weeks, probably depending in large part on how heterogenous or not the locality is. (That’s another reason it can be misleading to look at national, or even state-level, graphs for a country like the US, as opposed to analyzing outbreaks more locally.)
A lot of time has passed, so the link to the page of graphs we were discussing no longer shows the same thing, but here’s a snip from the NYT that might serve the same purpose:
I think most people would look at that set of graphs and say they are broadly similar. They at least seem to belong to the same basic sort of family. There are variances, sure, but we’re dealing with natural phenomena so I think we can accept that. It looks like there is something similar happening in all those places. Maybe a little more intense here and a little less intense there, but it certainly doesn’t look random.
So, I reply with this:
Given how similar all those curves look, maybe it’s not ‘just lockdowns and social distancing’.
Now, there’s room in there to rebut me and make a stronger case. You could say ‘Well, they don’t all look similar, I mean you have this outlier here. What do you make of that?’ And then we can explore the outlier and figure out if there are valuable lessons to be learned from it. Maybe there is a country on that page that has indeed figured out to how to stop this natural phenomenon in its tracks, which would certainly be highly useful. Or maybe we look a little closer and find out that there is some other reason why we’ve got the outlier. Maybe their epidemic got a late start and it’s still on the upward part (of what is going to end up being the same-looking kind of curve). Or maybe we find something like…what’s that one small country that always stood out with its high death rate…San Marino? Maybe we decide that the data sample is too small to draw reliable conclusions from. Or maybe we find a country that never really had an epidemic to begin with, meaning they may not have much to say when it comes to figuring out a way to stop an epidemic that is already raging.
So, we could explore those things and maybe move the discussion forward. But what DMC did was reply with snark to ask if I had even been to the page I commented on:
Did you actually go to the link? Do Sweden and Taiwan look similar?
Now, that’s not an honest-broker sort of way to move the discussion forward, to imply that I would have commented on the link without visiting it. I gave it the kind of response it deserved:
Yes, I did. That’s how I was able to observe that they all looked similar.
I ignored the second question, because I didn’t (and don’t) think the point being made was relevant to the discussion in the thread. The thread’s about flattening curves. You can’t flatten a curve when you don’t have a curve. Again, on the same scale as Sweden above, Taiwan is a flat line.
Here’s Taiwan’s death curve:
Yeah, that’s not really a curve we can do much with.
Here’s Taiwan’s case curve:
You might call it more of a blip, given the y-axis. It’s certainly not ‘flattened’.
Here’s Sweden:
When you change the scale like that, it gets kinda hard to see what was going on at the end of May. Here’s Sweden’s death curve:
I’m not trying to tell anyone what to make of those curves, mathematically. But DMC had an attempt, with the curves as they existed in late May. Here was his reply:
You think a Gaussian and a Gompertz function look alike? I thought you were supposed to be a math teacher.
Now, I let that go and exited the thread, because that was not a discussion I wanted to get into. If you ask me, I think that Sweden at the time DMC asked was just beginning the downward part of a distribution that looks not dissimilar to those of most other countries. And again, I thought (and think) that Taiwan at the time wouldn’t even have had enough data to robustly define a function.
But even now – especially then, but even now – I think that casually throwing out a statement that could well be inferred to imply that the epidemiology in different countries follows different mathematical functions is…well, I don’t really know what to call it, but it seems to claim some expertise that is unwarranted, or at least not supported by evidence. I mean, my impression was that scientists and modelers are still working had at figuring this stuff out, and that the matter is far from settled. Maybe I’m reading his statement the wrong way, but it sure seems to me as though he’s implying that the case count graphs indicate something about the mathematical nature of the viral spread in the respective countries.
At any rate, I didn’t answer his question. I ignored it. He has repeatedly, in many threads besides here, made the claim that I believe those two images he posted in this thread look similar. I don’t believe that, because I’m not an idiot. And I never said I believed that.
Look, I’m not trying to push disinformation or avoid an honest appraisal of the facts. I find, and have found throughout, that it’s not necessarily clear what the facts indicate. What I’m trying to do is hold people to honest arguments about those facts. Here’s an argument from that thread:
Social distancing, better hygiene, masks, etc. Sure, the new cases are decreasing - because nearly the entire planet is sheltering in place. The “worst is over” ONLY because of the lockdowns. If we throw off the facemasks, forget about social distancing, and go back to the way things were before cases will spike. As has been demonstrated in various areas that tried just that and turned into mini-hotspots in 1-2 weeks after that.
And here’s an observation on New York:
Yep. That’s happening here in NYC, which has not started to reopen yet. But people are sick of it all and are (irresponsibly) not wearing masks, not distancing, and bars and restaurants that cater to younger groups are re-opening (semi-clandestinely, but they’re open, no doubt about it).
I’d be surprised if we don’t see an uptick in rates of infection (and death) soon.
That uptick didn’t happen right away. Here’s New York’s curve:
I just don’t think these things are near as clear as they have been made out to be on this board. And I don’t think the reasoning behind a lot of the arguments is solid.
That said, I regret the lack of decorum I have exhibited, or let myself fall into when I perceived myself being provoked in the disingenuous kind of way I just described. It’s probably best if I withdraw from most of these discussions, or at least the ones where I doubt we’ll ever have much of a chance to see eye to eye. I think there is value to community debate on issues that don’t seem yet resolved, especially ones of such great importance to nearly everybody. But I don’t know anyone who enjoys being called subhuman, as in this thread; murderous, as in another; or just generally an ignorant fuckstick of epic proportions. I don’t know about the fuckstick part, but I’m certainly not ignorant enough to not see where I’m unwelcome or don’t belong.
