Hacker News Reader

Comment by soVeryTired | Hacker News Reader

The pattern is fairly consistent across countries [0]:

[0] https://www.ft.com/__origami/service/image/v2/images/raw/htt...

That is the most useful graph on this subject so far. Go look at it now. Most countries have exactly the same growth trajectory measured from 100 cases. That's solid data about how this behaves.

Singapore and Hong Kong have been able to slow things down by strong measures, but it's still a straight line on a log graph, just with 1/3 of the growth rate. Italy is on the same line, just 8 days ahead of Europe and 13 days ahead of the US.

Only South Korea shows flattening so far.

I still don't get how people - including apparently some governments - don't get or don't believe it. Getting this graph is as simple as 1) taking the daily measurements, and 2) plotting them on log scale (and technically 3) shifting the lines to overlap, but that's just cosmetics). The trajectory of the entire West being the same as Italy's is immediately apparent.

It's much worse than that. Italy's health system is already collapsing at 20,000 victims. The population of Italy is 60 million. That has to level off at some point due to saturation, but that's at somewhere in the millions or tens of millions.

Worse still, Italy has one of the better healthcare systems. This doesn't bode well for the rest of Europe.

Well Lombardy does (number 2 in the regions of the EU). I believe that some Italian commentators have said that one of the biggest reasons for the total country lockdown was that the health systems of Southern Italy were much much worse than Lombardy, and would definitely not be able to cope.

FWIW, at the moment it's only lombardy's health system which is collapsing, because most of the victims are concentrated there and the growth rate there was _much_ faster because the lock down was only imposed late.

The growth rate is the same in other regions but delayed, and there is some hope that the rest of the country might be able to cope with it better if the lock down measures are effective.

Hopefully. But the health care system in Lombardia has to be the best or at the top of entire the country. Southern Italy will struggle with fewer cases.

In part, because they think they already understand the situation, and the nature of that understanding precludes taking things like that seriously.

“It’s just the flu”

“It only kills 2% of people”

“The flu, cancer, car accidents, malnutrition, etc have killed far more people”

“People were worried about SARS, Swine Flu, and H1N1, and they turned out to be hardly kill anyone.”

“There number of cases here is small. Why are you panicking over nothing?”

“People need to stop panicking. People are going hysterical. The media is trying to make people panic. Just look at them hoarding toilet paper”.

“People are trying to ruin the economy over nothing”

“The experts — epidemiologists — say there’s nothing to worry about” (I’ve had this said to me and I was given a link to a news article that spun statements from some epidemiologists to make it sound that way. They were saying that they didn’t know how the situation would pan out, and this was presented as if it meant they didn’t think there was any issue).

I imagine that most people aren’t aware of the graphs. I imagine most people would be dismissive of them for reasons like the following:

They simply believe the apparent normalcy in front of them more than some abstract maths. They don’t grasp the notion of exponential growth, so it seems fanciful. They also don’t understand how it’s possible to project into the future like that. Maybe they think it’s just fear-mongering speculation. What concrete evidence do they actually have for the situation in Italy? And why should they believe what’s going on in a different country will apply to their country. Maybe their country has a good health care system, and they assume that Italy doesn't.

> I imagine that most people aren’t aware of the graphs.

And even if they are, they aren't aware of log scale.

Definitely. Noticing it's log scale (or anything non-linear) takes a trained eye.

I've worked with, presented, and consumed a lot of scientific data over my career and I still have a tendency to assume plots are linear no matter how many times I tell myself to always check scale first.

Any half-decent plot aimed at non-specialists will have "log scale" written in the title or surrounding description. The one linked by 'soVeryTired didn't, which is where FT dropped the ball. But it does not matter here.

The relevant axis on this chart is the time axis, the x axis. The chart is linear in it. You see a bunch of nearly identical trend lines, and you have to just count how many days a given country is behind Italy. It doesn't matter what scale the vertical axis is in.

> Any half-decent plot aimed at non-specialists will have "log scale" written in the title or surrounding description

I have to wonder how many non-specialists would even know what “log scale” means.

I didn't mean a non-specialist as in an average Joe; I meant an above-average Joe with some interest in one or more of: economy, physics, electronics, mathematics, biology. There's loads of such people.

>log scale

There's your answer. To 90% of elected officials, a "log scale" sounds like something a timber company might weigh their product with.

Because of normalcy bias: https://en.wikipedia.org/wiki/Normalcy_bias

"Normalcy bias can cause people to drastically underestimate the effects of the disaster. Therefore, people think that they will be safe even though information from the radio, television, or neighbors gives them reasons to believe there is a risk."

People just don't want to believe a disaster is about to happen. They will grasp at straws to explain why everything is going to be fine.

That's usually not a problem because disasters are rare, but it will catch us under-prepped when an actual disaster happens.

I have one issue with or maybe just question about this graph. Why it starts with 100 detected cases? It seems like arbitrary round number, but it also seems to be the smallest arbitrary-looking number that makes SK overlap other countries. Why is that? Why is the initial slow spread in SK irrelevant? What caused it to be initially slower than in other countries?

This doesn't really matter, you're just doing that to make it easier to compare. Shifting South Korea so it doesn't overlap the other countries is possible but not terribly useful.

They do seem to be missing Denmark though, which should stick out somewhere top-left of Norway.

They do seem to be missing Denmark though, which should stick out somewhere top-left of Norway.

I was just talking, via text, to a friend in Norway. They're going for major restrictions on contact and the people there are not worried enough yet. I sent her that graph.

Norway is going into full lockdown (stay at home, don't hoard). Borders closing except for export/import from Monday. Hut vacations now getting banned as well.

Which is a good plan and similar to what Denmark is doing, ie. stay at home as much as possible, everyone who can work from home does so, closing the borders for all non-essential traffic, and so on. There is a whole system being set up for compensation to concert venues and other businesses who are severely hit by the lockdown and blanket ban on gatherings of more than 100 people.

Open air walks are ok, as long as you keep a reasonable distance to other people. Only go shopping if it's absolutely necessary.

And yet some people are still going to crowded bars, like nothing's going on. It's amazing how stubborn some people are.

There's a distribution of who gets it when. The bans will be for those last 20% who just refuse to care about anybody else.

Denmark is now in today's version. https://twitter.com/jburnmurdoch/status/1238913530973339648/...

It's somewhat arbitrary, but you can't start from "patient zero" because that's not usually known. And the detailed circumstances of the first few patients matter. By 100 or so, the disease is in the population and starts to behave statistically. That's what's called "community spread".

It’s somewhat arbitrary, but also, countries can’t detect the presence of coronavirus when just a single resident has it. 100 is a good starting point for “this country has detected the virus within their borders — what’s going to happen?”

People infected with the coronavirus tend to vary wildly in how many other people they infect from people who stay home with a fever and infect nobody to "super spreaders" who many infect hundreds of others. That means the spread rate when you don't have a statistically significant number of cases is going to be all over the map naturally and not really comparable.

This could also be a measure of the rate of test kit production by country.

The graph can be reproduced with more recent data here [1]. It is based on WHO data, and lags a bit behind. If you want to see more data and graphs, see [2].

[1] https://ourworldindata.org/grapher/covid-confirmed-cases-sin...

[2] https://ourworldindata.org/coronavirus

Reply on news.ycombinator.com