The New Coronavirus Growth Debate | Hacker News Reader

The New Coronavirus Growth Debate

bloomberg.com

32 pointsimartin2k53 comments

Here's the real source of the debate:

"I look out the window and nobody has the virus. Nobody I know has the virus. I don't know anybody who is talking about anybody they know that has the virus. Virus? What virus? Oh, you mean the virus in China, the one sitting next to the their weird government and the air pollution. I guess that could make it harder for McDonalds to source their happy meal toys. Hey, did you hear about the latest thing that happened in the primaries?"

It's weird, and it may even be psychologically impossible, to look out the window at a sunny day and imagine that there's an explosion happening. You can look at the numbers, but even then, can you really believe it on a deep emotional level?

Humans are able to imagine such things. It may be the very thing that separates humans from other animals. We have the ability to imagine, understand, and reason about things we cannot physically see.

Well, enough people seem to believe it that I had trouble buying hand soap a couple days ago.

It had not even occurred to me that people might be buying up hand soap due to coronavirus. After all, we are advised every year to wash our hands frequently to reduce our cold and flu risk, and to wash our hands after using the restroom. Thus, I expected that most people already regularly buy hand soap, and so being told it is also good with coronavirus would not lead to a noticeable increase in demand.

Apparently, I greatly overestimated the basic hygiene practices of my fellow citizens.

Maybe they're buying extra, since it's nonperishable and they want to avoid heading to the store.

The problem is that statistically the confirmed number of cases is only 1 in 2 million in the US at the moment. Your chance of getting hit by lightning in any one year is higher at 1 in 700k. So at this stage it is very hard to comprehend.

It's when you consider the prediction that 40-70% of the adult population are going to be infected if it's not halted that you have to take it seriously. Noah Feldman while interviewing Harvard epidemiology professor Marc Lipsitch who gave that figure admitted that it just didn't seem real that we are looking at 1 million+ deaths in the US if that level of infection occurs. Very sobering thought.

As far as policymakers are concerned, it's the 40-70% that needs to be focused on as preparations are made. We're talking a few, maybe three months before this thing has blown up from an insignificant 1 in 2M and impacted most people's lives indirectly. What people have a hard time imagining is that something can go from insignificant to significant so quickly.

Yes, the people in China, Hong Kong & Singapore could because SARS is still within living memory and it left a deep cultural imprint. The US hasn't had a pandemic in living memory and so it feels so abstract. I'll bet the reaction to the next pandemic isn't so muted.

>But many medical professionals think in terms of what are called “normal” statistical distributions. If someone visits your office with what appears to be a typical flu case, it is usually exactly that.

These two sentences are completely, absolutely unrelated.

Obviously, it's a flawed argument, because the x-axis in this distribution would make no sense.

...but you clearly know what they are trying to say: Among the population of people today with flu symptoms, the vast majority of them are influenza and not covid19.

I wonder if this is still true in places like Wuhan. I also wonder if we'll see a decrease in influenza next season due to the all the people adopting far better hygiene habits this year.

For the people who think it will grow exponentially (seemingly most epidemiologists), China's job in taking an uncontrolled outbreak and at least temporarily controlling it seems quite surprising. How was China able to control it, and why can't that happen in other countries? Their efforts are really quite extraordinary.

"I think people aren’t paying close enough attention. The majority of the response in China, in 30 provinces, was about case finding, contact tracing, and suspension of public gatherings — all common measures used anywhere in the world to manage [the spread of] diseases."

(The above quote and both of these excellent articles are from the same primary source, Bruce Aylward from WHO):

https://www.nytimes.com/2020/03/04/health/coronavirus-china-...

https://www.vox.com/2020/3/2/21161067/coronavirus-covid19-ch...

Will this be enough to pause it for a long period of time? Or will it come roaring back once china lets their foot off the brake? Or will China extinguish the virus in their own country and have to block travel inbound to keep it that way? We shall see.

The answer to your last question is yes, with cases already of Chinese nationals returning from Italy with the virus.

If you go around welding people in to their apartments, well, the virus will probably slow down.

Were this the zombie apocalypse, people would be signing up left and right. Corona virus carriers just don't have the same menacing appearance as zombies, so everyone's still blase like it's climate change (which has a greater potential to kill more people - just not right now).

Rather, people should adopt a S-curve mindset, where the shape of the curve depends on the effectiveness of mitigation strategies.

How about evolutionary growth model:

- severe cases are quarantined

- mild cases keep spreading

- evolutionary pressure causes the virus to become less deadly

- deadliness drops to flu levels at which point it stops being actively tracked

Alternatively it does not spread effectively enough (not enough reservoirs in summer) and dies out in human population.

Or widespread herd immunity reduces its severity, not mutation. (e.g. Polio) In the latter case, we would have some slight persistent number of cases.

What's the false positive rate of the test?

The UK has test 16,659 people and confirmed 85 of them, ~0.5% of those tested, as infected.

How many of them are false positives, if any?

I know in some places two tests are done, by different labs preferably, for confirmation.

Though it might be possible that the false negative rate is higher than the false positive rate.

(I don't know, and have been trying to find the answer, and have failed.)

Best I can do: Assuming that tests performed in different countries are comparable [1], your statistic would set an upper bound on the false positive rate, which could then be applied to other countries. The U.S. has performed 3.6k tests and has ~150 confirmed cases. If the false positive rate is the same, the vast majority of those cases have to be real.

[1] Big assumption.

The assumption may be wrong:

> The rollout of a CDC-designed test kit to state and local labs has become a fiasco because it contained a faulty reagent.

https://www.sciencemag.org/news/2020/02/united-states-badly-...

Depending on the test used, there might be >50% false negatives in there.

They’re both right. We are in a phase of exponential growth. If we don’t check it it will absolutely overwhelm the medical system. But eventually we will have crossed the peak and reach a level more like a base rate. I don’t get the “base raters” position here, it seems they’re living two years in the future rather than concerning themselves about what happens on the way to that point.

If I understand "base raters" correctly, they are not saying anything in particular about this virus, just that in general due to all available factors (including human+government intervention), the world generally stays close to the same.

A month ago, if you asked a base rater and an exponentialist to make guesses on how many cases in China there would be today, I think the base rater would be much closer to actuality than the exponentialist (after all, their new daily case volume stopped growing almost a month ago!).

China was able to pause this outbreak, maybe they can hold it, maybe the rest of the world can too. I'm not holding my breath but it does seem like a possibility.

I see. That seems useless then isn't it?

I mean its fine if you're looking at the longer term picture. But its completely useless for informing how to handle an emerging situation. E.g. if China's decision making was informed by base rater thinking, which amounts here to "eventually things will return to normal" then they would not have taken the measures they took which are leading to that.

I'm just guessing too on what base rater is, but if I'm correct on what a base rater is, then the thinking has use for how normal citizens should prepare (maybe don't sell all your investments to buy rice and beans and N95 masks) rather than what those fighting the virus should do.

If my understanding is correct, the base raters don't say that nothing extreme ever happens, they only say that things usually turn out close to how they have been. I suspect that is true.

But is the number of cases really growing exponentially? The logarithm graphs over at https://www.worldometers.info/coronavirus/ seem to indicate otherwise but maybe that's just China's successful containment efforts skewing the data?

According to the WHO, China is running a massively successful quarantine campaign at the expense of pausing their entire country. However, cases outside of China are still growing exponentially. Check out that plot - it's a straight line on a log scale.

> However, cases outside of China are still growing exponentially.

Arguably because we're still catching up with the initial wave of outbreaks that could develop unhindered. (as was the initial wave in China).

It would be very surprising, based on total population and density alone, if any other country sees more cases than China with maybe the exception of Iran which appears to be in disastrous shape.

https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.h... has separate graphs, both logarthmic and linear, available for China and the rest of the world.

The Chinese graph isn't even showing linear growth anymore. The rest of the world looks pretty exponential, but there are still fewer cases in the rest of the world than there were in China.

This https://www.worldometers.info/coronavirus/coronavirus-cases/... graph shows the number of active cases declining.

Getting data from https://gisanddata.maps.arcgis.com/apps/opsdashboard/index.h...

My gut feeling is that far more people have this virus than the Johns Hopkins app says.

The biggest outbreak outside of China seems to be Iran, with 2,922 confirmed cases, 92 deaths and 552 recovered. This does sound bad since that is technically a 3% death rate, but my gut feeling is far more Iranians actually have it, but are not checking themselves into hospital because they don't want to be put in quarantine and would rather just stay home.

I'm not trying to downplay the seriousness, Corona virus still is something you don't want to get, but it probably won't kill you if you're a reasonably healthy person.

Unfortunately it will probably sweep the world before a cure is found, so we'll all be exposed to it eventually.

Except for some of us who come from the time you had to get permission to get an ID on the Arpanet. In that case, the WHO is not saying mathematically nice things about us. :-(

Why would anyone even remotely skilled in math consider a viral spread exponential rather than logistic or Weibull?

The former stems from communication theory and the observation that few people can be sick forever or more than once. There are also constant factor communication "costs" to the virus expressed as likelihoods of infection due to minimum effective viral load, countermeasures, severity and immunity.

Latter if fomites are a major factor in final spread. (overlapping point scatter gets you Raleigh - you have to be unlucky to be near the source)

Potentially a combination of both. They all look sigmoid.

The stronger the countermeasures, the lower the total maximum.

Non-article. Tl; dr: Some people think corona is going to spread a lot, others don't.

I get the feeling that its a double edged sword. Do you close schools(and more) to prevent the disease from spreading, or just overload hospitals with more patients? Either way, its going to hurt the economy. Which is better choice - is still being decided.

The problem is quarantine is not enough. We cant test fast enough - or at all. So is the only choice to keep people from gathering?

Quarantining the most at risk and those who look after them could be the best solution as far as reducing deaths.

They keep saying don't wear facemasks but if you are a asymptomatic carrier a facemask will limit you spreading it to others. And if everyone wore a mask correctly you then have two barrier protection. The problem is we don't have enough masks, and would quickly run out of them for health workers dealing with confirmed cases.

The problem is that asymptomatic carrier is everyone. Face masks are not free yet, nor comfortable, not everyone will want to wear them, much less correctly.

This guys understands exponential growth! Outstanding...

The closest historical match is probably the 1957-1958 Asian flu epidemic. Airborne spread. There were complications in 3% of cases with 0.3% mortality. 68,000 deaths in the US, 2 million worldwide. Stopped by a vaccine developed after about a year.[1]

Live graph and map of current epidemic: [2]

[1] https://www.ncbi.nlm.nih.gov/pmc/articles/PMC2714797/ [2] https://hgis.uw.edu/virus/

We do not know mortality of this coronavirus truly yet, much less number of cases. It seems to spread somewhat less easily than flu, similar to other coronaviruses already known - like about 15% cases of cold. Potentially spread also shows seasonal nature.

I'd like to see rates and complication rates for common cold caused by coronaviruses.

https://archive.md/yd4me

I'd put myself in the growther camp. In comparing it to ebola and SARS I think it is very different. With ebola the rate of death was very high, and very fast. It couldn't spread quickly due to areas of infection dying out quickly, plus it was blood borne and the Congo is not hyper connected to the rest of the world. For SARS the world got lucky there, again it had a higher rate of death than COVID-19(10% mortality vs. ~3%) but China back then(2003) was a different beast than China 2020. Its a lot more tied in to the global economy and overall now has a higher standard of living ( more tourism, more business flights).

For COVID-19 which is extremely concerning, based on current medical knowledge is that people without symptoms can transmit the virus, and that the virus could take 14 days to present symptoms. Looking at exponential growth in cases at South Korea and Italy show how fast its spreading. Good news is governments are seeming to take it seriously and remdevisir and a few other drugs looks promising to fight it given a vaccine is a ways off.

Pointless article. Posits people fall into two camps, the smart people who understand exponentials, and idiots who don't. Guess where the people in Washington, D.C. fall.

Not at all. If you know the context of superforcasters (See Good Judgement Open) they are the only ones who are consistently good at making forecasts. Even most experts fair no better than chance. And they rely on detailed assessments with a combination of different sources and lots of analytical work.

But what's the base rate here, the base rate of years that have pandemics, or the base rate of exp(t) functions that go up really fast?

If you have read the article it actually mentions explicitly that the 'base line' mindset turns out to be more often correct than expert opinion.

Also I think the term 'growth rate' people is charitable. 'singularity people' might be more accurate because many people in that camp don't really seem to understand that most exponential curves are actually s-curves.

Right... Only an idiot would bet against the fact that weeks ago, when China had tens of thousands of cases and exponential case growth, that they would now have many millions right? But yet the idiot would have been right.

This wasn't a natural flatlining, of course, but it shows that it can be done with this virus (at least temporarily).

Of course, the policymakers can't set their policy based on a forecast that assumes that the rate will decrease due to policy.

It's a good point, HN thinks after 60 years of sucking and being useless Augmented Reality (AR) is going to go exponential this year!

But then someone in the Valley did beat a hell of an old Taxi institution exponentially.

Meanwhile the trump administration is hiding statistics and desperately trying to privatize testing as fast as physically possible so the test counts/result counts are not available as public analytics.

Surprised few have noticed this. It's a bit insidious because the cases are going to explode.

Forcing every test to go though the highly limited CDC was a terrible policy move and was definitely worse than letting the hospitals do it. Trump has some blame here because it should have been changed sooner, but letting hospitals run tests is likely to increase the number of known cases. The former (under Obama I think) FDA head was against the ban that Trump lifted.

Very accurate division between the two mindsets. Considering the overwhelming base-rate bias, the contrary growth bias is the one that is in need of a larger voice.

Also, there's some reasons to think this may be different than these other outbreaks. One reason is that plenty of smart people suspect this was an accidentally released bioweapon, given the ground zero was so close to China's only bioweapons facility.

1) no they don't.

2) even if they do, that doesn't mean a single thing without any kind of supporting evidence.

Reply on news.ycombinator.com