Hacker News Reader

Comment by voidfunc | Hacker News Reader

I'm not sure how to feel about this simulation... I basically did nothing and burned through COVID within 3 months and things went back to normal by mid-summer. By the end of the year, despite mid-summer government confidence being terrible, I was told "The people trust the government" with a 70% approval. I mean I killed what 10% of the country, but hey... the people have short memories!

What is this supposed to be teaching exactly?

If the model is correct then I think a lot of people would say we should have just let the disease burn out and deal with the deaths. I'm not sure that's what the designers intended.

View on HN

I'm not sure it wouldn't have worked. Sweden had something like 80% support to only have event limits at 500 and no other restrictions, with people arguing in public forums that those dying deserved it and that it was worth it.

I had a doctor outright scold me for wearing a mask to the hospital here (checkup for crohns disease which I take immunosupressants for) and didnt see a single other mask in the hospital until December 2020. o.O.

Smaller countries with limited access to the wider discussions in English can turn very weird.

>> I was told "The people trust the government" with a 70% approval. I mean I killed what 10% of the country, but hey... the people have short memories!

> Sweden had something like 80% support to only have event limits at 500 and no other restrictions

But even now Sweden has 0.15% of the population killed by covid-19, that is like 100 times smaller than the 10% that the GP got in the simulation.

Also note that as of right now, Sweden has half the number of deaths/capita compared to Czechia, and is below the EU average, despite having had the least restrictive policy.

This game perpetuates the gigantic fallacy that government intervention is the biggest factor determining spread and outcome, when in reality it's geography, structural differences, and enormous amounts of luck.

I would also say it's cultural. Swedes keep a distance without you telling them, Spaniards do not.

The biggest difference between Sweden an Spain is the presence of multi generational households in the latter.

That's a good point, too.

Those sorts of explanations are all totally ad-hoc and have no real scientific validity. The English are famously stand-offish yet have worse results than Sweden. Texas and Florida are culturally very similar to other parts of the USA, yet unlocked without experiencing any change in their situation whatsoever and have seen similar results to other US states.

You can find counter-examples like that for more or less any posited influential factor on COVID outcomes, except for obesity. Government interventions in particular don't correlate well with anything. The models are all built on the assumptions that they should have a huge impact, but real data tells us differently - the assumptions the models are built on are unsound.

IMHO it's really a pity the game authors used this to influence the Czech parliament. These models were never validated against reality before being thrown at policymakers, instead the (sad to call them this) "scientists" simply assumed that because the underlying theories are simple they must therefore also be correct. I've had actual scientists tell me "of course lockdown works, it's basic germ theory" which demonstrates a pretty chaotic confusion between theory and reality, but that sort of attitude can be found everywhere in the academic modelling world.

> IMHO it's really a pity the game authors used this to influence the Czech parliament.

Remember that politicians are in the tradeoff and bullshit business. They assume every they are told is bullshit, they repackage it and than sell it to us. My guess is if they don't like what the scientist and economist say, they just get another group they like more. It's probably a mix of the advice they get, with the tradeoff they want to make.

> "of course lockdown works"

Of course they work! The question is how much it affect the transmission ratio and the economic cost, and if people will follow the restrictions.

I like that this was call the "Corona Game" and not the "Corona Simulator". Some people noticed that in this game "doing nothing" does not drop the popularity to 0, and I think that it's very unrealistic. Other people noticed that closing the schools has a big effect in the game, but the different parts of the government here in Argentina disagree https://news.ycombinator.com/item?id=27783454

Of course they work!

Why do you say this when there are many analyses saying they don't have any impact. At any rate that doesn't really need studies, because places that either didn't lock down, or did and then entirely removed restrictions (like Florida, Texas), did not see any large clearly artificial jump in case numbers. And yet suppressing such spikes was the sole justification for lockdowns.

That's why, when investigated, all claims I've seen that lockdowns work are either mis-analysis of the data (Flaxman et al and friends), or, justified on theoretical grounds like "of course they work, theory says they work".

Perhaps lockdowns are overrated, I think nobody knows for sure the correct weight of it, but I'd be really surprised that it has absolute zero effect.

It also depend on other factors. I guess most people in Florida and Texas move in their own car. Here in Argentina most people use public transport.

For example in normal times I have once a week to commute from one building of the university to another building that is like 7 miles away. I can take a bus, the trip is 90 minutes, and at rush hours it has like 100-150 persons.

I can also make the middle part of the trip using the underground/metro, so I pass under the mess of transit in downtown and the trip is only 45 minutes long. Now I have like 150+150+150=450 new friends to share the air. (And the underground/metro cars are bigger and more crowed, so perhaps it reach 200 and the total is 500.)

Many persons here in normal times must do similar or longer trips everyday, twice per day at rush hour. (Perhaps bus+train+bus, or bus+train+undergroung and the can't skip the train because they live too far away.)

(And the same morons that open in summer the window because the air conditioner is not cold enough, will close the window in winter because the wind is too cold.)

So in different places, a lockdown may have very different effects.

I agree it's really surprising that they have no effect. Yet that is what the data seems to be telling us, and it's not specific to certain regions. The number of places is limited because the number of places that got rid of or didn't implement lockdowns is limited, but the story is consistent between them, including in Sweden where they also use public transport.

I believe the fact that this invalidates what was seen as robust theory is one reason (but not the biggest reason) why people are resistant to that conclusion. Yet, that is what the data demands of us.

In reality there are lots of ways to patch up germ theory to handle the ineffectiveness of lockdowns. For instance, contact tracing studies claim that almost all infections are spread in homes, care homes and hospitals, i.e. places where sick people have to be. That makes logical sense: you would expect viruses to spread in places where sick people are, and you would expect sick people to be in a bed somewhere because they feel sick. The idea that lockdowns work is strongly predicated on the belief that asymptomatic or pre-symptomatic people are highly infectious. However there's actually very little evidence of this, and quite a lot of evidence that they're not.

Hey, just wanted to say thanks for continuing this thread, I agree with pretty much everything you said.

Lockdowns simply don't work, because a functioning society where people don't starve to death requires massive amounts of "essential" workers who have to be exempt from the lockdown rules. But a lot of people seem to think Uber Eats is run by an army of small elves or something...

> Sweden has half the number of deaths/capita compared to Czechia

> when in reality it's geography, structural differences,

Why did you cherry pick Czechia when you have neighbouring countries which are much more comparable in term of climate, culture, geography, density, &c.

Finland: 176 deaths per 1m hab

Norway: 146 deaths per 1m hab

Sweden: 1437 deaths per 1m hab

> with people arguing in public forums that those dying deserved it and that it was worth it.

I would be very surprised if anyone actually argued that.

> I had a doctor outright scold me for wearing a mask to the hospital here

This is HIGHLY irregular. Right before the 2020 summer all Swedish hospitals started with corona security measures: you could not enter the hospital before talking to someone (in full protective gear) and at the very least having your temperature measured. If you were allowed inside and masks were available you would be required to wear one.

Source: visited hospitals and care centres multiple times with sick family members.

Even in a worst case scenario it would be impossible to kill 10% of the country. The worldwide infection fatality rate is under 1%. That would rise a little of the healthcare system was overwhelmed but getting to 10% isn't remotely plausible.

It would rise a LOT if health care gets overwhelmed, as we saw in Italy.

And it’s not just COVID… if the hospitals are overrun many other things will have much higher mortality, like trauma cases.

None of that gets you even close to 10% mortality. In any given year only a tiny fraction of the population needs advanced hospital care to survive trauma or other diseases.

IIRC Lombardy death rate reached 17% once hospitals were overwhelmed.

https://www.thelancet.com/journals/lanpub/article/PIIS2468-2...

You do not recall correctly. You have confused case fatality rate with infection fatality rate. There are a large number of infections which are never officially counted as cases due to limited testing and minimal symptoms.

Well, with that line of reasoning we cannot trust any corona statistics, can we?

Anyway, the effect of failing to "fallten the curve" in Lombardy are well documented. And this happened in December 2019 before we had a global pandemic.

Trust has nothing to do with it. Based on seroprevalence studies and other factors the CDC has estimated that only about a quarter of infections were officially counted as "cases". The undercounting was even worse early in the pandemic when tests were hardly available. So your point is not valid.

https://www.cdc.gov/coronavirus/2019-ncov/cases-updates/burd...

But that's US statistics, this happened in asmall area in Northern Italy.

Even if we go with your numbers, we get an average of 4-5% and since we know there was a huge spike in the middle of the outbreak it is not unrealistic to assume they were above 10% at some point.

It is impossible to kill 10% of the country in the game as well. (Czechia has population 10M, not 1M)

That's 1% with fully functioning ICUs and other stuff. You can _easily_ reach 10% fatlity if the hospitals are overwhelmed.

This happened for a short time in tiny part of Italy and the death rate skyrocketed.

Less than 7% of official COVID cases require hospital care. It is not possible to reach 10% mortality at the population level even with 100% infection rate.

If the two metrics used are loss of life vs cost of pandemic, the cost of a life needs to be included in the final score I think.

In Australia, it appears some early models predicted between 50k to 150k lives lost assuming no response. A statistical life is valued at over $4m, meaning a cost of $200bn to $600bn. Also, with 60% of the population infected, that would mean over 12 million people infected - we still don't have a good idea what the long-term issues will be - but one study in Australia found ~5% of people experiencing "long covid" symptoms, which would mean at least 600,000 people with long-term complications - which in our hypothetical Harkonnen society we'd likely just shove them into a hole somewhere and forget about them, otherwise we'd have to deal with their social welfare (yuck!) and strain on the health system.

No doubt the cost, responses, short and long term impacts, etc, will be something economists and sociologists will debate for decades to come.

https://www.smh.com.au/politics/federal/thousands-of-predict...

https://en.wikipedia.org/wiki/Value_of_life#Australia

https://www.abc.net.au/news/2021-06-24/long-covid-nsw-resear...

Edit: On thinking about it, I think the value of life is based on insurance / compensation purposes, perhaps the value of a life in terms of total GDP contribution would be different? Anyway, a subject more for the actuaries and economists I think. Interesting to consider nonetheless, which I think may be the overall point of this game.

>In Australia, it appears some early models predicted between 50k to 150k lives lost assuming no response. A statistical life is valued at over $4m, meaning a cost of $200bn to $600bn.

That is the value of an "average" life. But the average age of death from covid is over 75, so the actual economic cost would be much less (which is accounted for when using https://en.wikipedia.org/wiki/Quality-adjusted_life_year, the standard way of measuring these things).

I was thinking that the average age of death would have that effect but I wasn't sure what it was called, so I appreciate the pointer to QALYs. Your point is well taken that the total economic cost can't be based on an 'average' life given the typical demographics impacted by the pandemic.

There's a write up here[0] that in part attempts to compare QALY's lost to strong lockdown vs light lockdown. References this game in fact.

https://astralcodexten.substack.com/p/lockdown-effectiveness...

> This paper finds that the average COVID victim might have lived another 8 QALYs (quality-adjusted life years - a measure of years of life saved, in which years when you are very sick and can’t do anything count as less than a full year) if they hadn’t gotten COVID.

I take issue with this.

The average age in the UK of death was 83 I think, compared to an average life expectancy of 82. Most had more than one comorbidity. Do we really expected these people to live until 90?

In addition to this a lot of the early deaths were in care homes were the residents aren't expected to live 8 years (less than 3 years from a very quick search). That would mean that non care home residents would need to average an ever higher amount of life for that to be true.

Sorry, that seems kind of ridiculous.

The life expectancy of people that die at 82 doesn't count against them like that. Glancing at a UK life table, life expectancy at age 83 is 7 years for men and 8 years for women.

Throw in the youngers and it is clear enough the math should work out fine.

An average includes "youngers".

The expected life expectancy you give isn't for people with an average of two comorbidities.

In our thought experiment, the younger people all have more years of life expectancy than the older people.

I doubt this - it is impossible to kill 10% of the population in the game. As anyone can see on the final screen, max number of deaths in the game is about 150000, which is 1.5% (Czechia population is approximately 10M)

[game author here] Worst case scenario in the game is ~1.5% of the population. The estimated fatality rate is 0.5%. This number is increased to 1% if the hospitals are overwhelmed (note that there was a strong consensus in the country that this needs to be avoided).

Note that only 1/4 of the infections get detected in the game the game so the ratio deaths/detected infections will be 4% (the deaths/actual infections is 1%) - likely too optimistic in the scenario of overwhelmed testing capacity.

If you do nothing in the game, you kill 1% of the population during the first wave and you kill another ~0.5% in winter after the natural immunity of some people expires. Again, the real country was simply not willing to go that far.

Well, letting the thing burn is actually the fastest way to end the pandemic. The only thing faster than that would be additional intentional spreading (covid parties?) and/or vaccinations going on at the same time. This applies to any pandemic really, as long as the disease leaves some immunity. It's epidemiology 101, for example the SIR model flatlines when the R gets to the herd immunity level.

But do you want the other [than pandemic ending] consequences? Like in your game-based example, 10% of the population dead?

Is it? Places with little forced mitigation like Brazil are still having waves. Some portion of the populace will be cautious and take precautions, and transient herd immunity within networks of people out and about will prevent some less cautious people from getting it during a given wave.

To truly burn it out in real world conditions you’d need to actively infect people.

Biochemist here. No it is not, and that's not epi-101. By letting it rip you're not just maximizing your fatalities, but you're also maximizing the rate at which the thing mutates, which maximizes the probability that your "recovered" pool becomes vulnerable again.

>but you're also maximizing the rate at which the thing mutates

No country is a closed system. We see now that a previous immune response is better than nothing against the Delta variant. I'm wondering what will happen to the island nations that kept the virus at bay while not having a high vaccination rate, won't they have many more deaths now with the more infectious and more deadly version that other countries that "let it rip" like Sweden have at least partial widespread immunity to?

True. But suppose you had no vaccines, then roughly the same mutations will play out whether the epidemic lasts a week or a year.

(Also, what the sibling wrote)

If you have healthy, low-risk people intentionally get the disease (from a single variant which you keep in good stock) and then strictly isolate/quarantine until they get better, that'd be a potential strategy. It's probably the best way to get overall immunity high enough that COVID would burn out... ...of course, this is basically the same as a super crappy vaccine. Vaccination would be way better. But back when people were poo-pooing the possibility of a fast vaccine development, it almost felt tempting to just do this and get it over with.

Because the virus wouldn't be spreading beyond the host that got it, the risk of mutation would be small.

(IMHO, we should've just massively accelerated vaccine development... even just a couple months would've saved lots of lives. If you're willing to try the strategy in my post, then accelerating vaccine approval would probably be no worse morally and probably a lot better.)

I played the game by restricting everything except a full lockdown and by doing nothing. The first time I had less than 200 deaths, the second time almost 150 k. The government survived in both cases, actually a little better in the latter. I'm not surprised because I don't remember a single government having to resign because of covid, even where the pandemic hit hard. Somebody had lost elections but it's difficult to assess the impact of covid mismanagement of the outcome of any election.

Well if you manged to kill 10% of the country then it is completely inaccurate. The maximum CSF I have seen estimated for coronavirus is 3% and I have a feeling that is quite an overestimation.

Yeah I killed 111k people immediately and then it flatlined, I was a little surprised. Then again, I killed 111k people.. my second time I did what I thought we (the US) should do with a full lockdown and killed like 50 total.

Depends when you do it. Obviously if you go full Madagascar at the first sign of trouble, you’ll crush it, but it’d pretty absurd behavior (every false flag would trigger lockdown).

I waited a little by accident and then went full lockdown and got a total of 130k deaths

Reply on news.ycombinator.com