Corona Game | Hacker News Reader

Corona Game

koronahra.cz

65 pointscwmartin27 comments

Got a very good result (bottom left corner of scatter chart at the end) by keeping masks on and banning risky businesses for the whole time and limiting events to 100 or 10 people and keeping schools closed until summer.

Probably wasted money on vaccine promotion. It didn't feel like a vaccine did much difference in such short time-frame and number of immune people rose super slow even towards the end.

When they extend the game till the end of 2021 and beyond vaccines may make huge sense.

Here's something that's been bugging me. The SIR (and SEIRS) model that this game is based on seem to me to have a flaw. Now obviously I'm not an epidemiologist, and I'm sure that epidemiologists have a good handle on this, I just don't know what it is.

The problem is this: if you assume an exponential decay, you're saying there's no point after infection at which the probability of transmission is zero. Probability of infection has a half-life in SIR. But for a virus like covid, we know that's not true: if you're infected, give it six weeks and you won't be infectious any more, one way or another. If you replace an exponential decay with a finite transmission lifetime, you do end up with different epidemic curves: they have steeper fall-off, for instance.

What's the reasoning here?

Doing absolutely nothing apparently leads to an infection fatality rate of 4%, which I'm extremely skeptical about.

While the figure for deaths isn't that high, not sure if the modelling of 'do nothing' would include the extra covid deaths due to overwhelmed hospitals?

Early into the pandemic last year I did a rough calculation multiplying out the death rate per age group by the US population by age and I recall it was something roughly around 5% so a number of 4% doesn't seem too far off. Granted the big assumption was infection rate and hospital capacity remained the same (eg New York was at 0% but other states were not).

Taking in data present today and looking at countries that are doing nothing (Sweden), their total population death rate isn't as high as 4% either.

Infection fatality rate for COVID in the Western world in practice was under 1%. I am quite skeptical of the idea that overwhelmed hospitals would quarduple that.

Do you have a source?

Keep in mind I originally did this exercise back in the early days of the pandemic with limited data. I can't locate all the original page I used but these new numbers work out. Going by previous estimations if everyone in the US got infected theoretical death is 1.9% of the population [1]. In march 37.7%[2] of hospital visits required the ICU and 13.6%[2] of hospitalizations resulted in deaths. So here we make the big assumption that everyone who died of COVID did so in the hospital which means 14.6% of COVID cases require the hospital. The next assumption is that if you don't have access to an ICU with COVID-19 you're going to die. We can increase "hospital" capacity with beds in stadiums and schools and such but we won't have enough equipment for more ICUs, so assumed ICU rate multiplied by assumed hospitialization rate is the worst case deathrate of everyone getting infected in the US at once in the early days of the pandemic, which is 37.7% * 14.6* = 5.4%

I'm not fooling myself with precision so I concluded roughly 5%.

An interesting note after the fact that the early estimated US deathrate of 1.9% tracks modern data pretty well [2] which is 1.8% (605,792/33,769,600).

[1] Sources and math, I used archive.org to get the numbers from March 2020 https://www.worldometers.info/coronavirus/coronavirus-age-se...) https://www.census.gov/data/tables/2019/demo/age-and-sex/201... age-sex-composition.html

  age    pop(M) deathrate theoretical-deaths
  00-39  169.5   0.2%     0.34
  40-49   39.9   0.4%     0.16
  50-59   41.5   1.3%     0.54
  60-69   37.9   3.6%     1.36
  70-79   23.4   8.0%     1.87
  80+     11.9  14.8%     1.76
  ----------------------------
  total  324.1            6.04
  6/324.1 = 1.9%

[2] https://gis.cdc.gov/grasp/COVIDNet/COVID19_5.html

[3] https://www.nytimes.com/interactive/2021/us/covid-cases-deat...

Deaths start to pile up when hospitals can't accept more people which otherwise would have recovered. A rapid transmission rate was bad, but hospital saturation was the real COVID killer.

[one of the co-authors of the game here] What you got is CFR 4%, which is highly plausible. As in reality, what you see as cases, are just confirmed cases.

IFR could be calculated only from the explanation only at the final screen, where the actual number of infections is estimated (it's several times higher than the confirmed cases)

We could legitimately say that in your game, the number of infected people was not 2.1M, but, including reinfections, there were up to 6.2M infections, according to experts’ theories. The game and your results only register the officially recorded infections (based on test data).

Hi, game author here.

First, this URL https://covidgame.info/ loads the game in English by default. There was a plan to popularize the game outside of Czechia but, frankly, we missed the window when major news outlets were interested.

The game source code is here: https://github.com/swehq/corona-game

I added few comments on the other thread: https://news.ycombinator.com/item?id=27763873#27766930

I don't like how there is no option for, "spend billions of dollars on buying enough rapid Covid tests to test every man, woman, and child in the country every single day". Given that rapid Covid tests cost about $5 a piece, this should be an economically feasible option that should be able to sharply reduce the spread of illness. Its too bad that nobody tried it in real life, or we would be able to see if it works.

I would like to have the option to close schools more selectively. In Poland every major surge of cases started roughly 3 weeks after schools were opened. Even if it was just opening of schools just for kids ages 7-9 while they stayed closed for all other kids.

I know it was only few instances and the same relationship doesn't hold for other countries, and also most recent school opening in the summer didn't cause the next wave for the first time. But I think in Poland where we don't party all that much those little unhygienic buggers share way larger portion of the burden than science could establish so far.

I coordinated a testing center where we piloted antigen tests.

Antigen tests are less sensitive, applicable in a much more narrow time frame than PCR tests, and not as trivial to perform correctly as suggested in the media. On top of that, they can be sensitive to external factors as well (such as temperature swings for tests in storage, and fast expiration).

I dislike these antigen tests due to the false sense of security they give people. They have their uses, but they are not a substitute for PCR tests.

There's a reason why influenza antigen tests are not widespread.

Slovakia did this in November [1]. Population ~5.5M, 2.5M tested over a single weekend for the first run.

Afterwards there were a few more weekends with lower turnout and until about month or two ago there were free testing places all over the country. These were during many periods require to cross between regions, go to work or shops. Significant fraction of population was getting tested on weekly-biweekly basis.

I can't find a good chart because most only show PCR tests (for a good reason!) but according to [2] Slovakia averaged 7300 tests per 1000 people.

The initial mass testing weekend was useful to get an overview of distribution of the infections and remove a large chunk of the infected from the pool (despite high false negative rate of those tests). It was also good PR and made people more aware how serious the situation is and reconsider if they would rather wear the mask and distance than go through this.

However the continued testing is now viewed as mostly useless. More money spent or tracing and targeted PCR testing, plus vaccine campaigns could have helped more.

[1] https://www.theguardian.com/world/2020/nov/01/half-slovakia-... [2] https://ourworldindata.org/coronavirus-testing#how-many-test...

It's really cool, it's so powerful that it makes people realize the trade-offs.

The only problem is that in hindsight of course we can do pretty well. I guess most governments could do better after the fact. Too bad you can't play the future with random-generated possibilities. The fact that we don't know what's going to happen would make it in part more accurate.

I don't know the folks behind this, but to me the take-away to "in hindsight of course we can do pretty well" is that really, the whole pandemic should have been addressed with (a lot of) hindsight.

The idea that viruses spread like this isn't new. It happened before. Africa had seen it. Asia had seen it. When Europe went crazy with "they're taking away our freedoms" a lot of my Chinese friends hopped on a plane to China, away from the "plague enthusiastic" westerners.

A lot of what happened over the last year was pure crazy, and entirely foreseeable. It wasn't clear when or where such a spillover event would occur, but what would happen (or not) after that, as a function of actions taken (or not) was painted out and experienced before, by epidemiologists and entire countries. And it was promptly ignored by policy-makers, journalists, and the general population.

I lost a lot of faith in (western) humanity over the last year. I have little hope left that we'll learn anything from it. In fact we're already seeing that some of the obvious promises made with easily verifiable binary outcomes, like better income or security for frontline workers, are being dropped.

I recently watched a youtube video from May 2020, and in the end the person on screen hears noise, goes to the balcony and starts clapping with everyone to celebrate "our heroes".

That didn't fucking age well.

</rant>

That's not what I meant. You can err on either side and both directions have their costs, which are much clearer in hindsight. There is a reason why the game doesn't predict the future. Because we don't know it.

I specifically meant erring on the other side.

"Masks, hands, distancing" somehow costs 498B? Luckily, after a spike of 200k cases a day for a couple of months, it seems most of the population dies or becomes immune, with 3.6M infections (14M including reinfections) and 200K dead by simulation end. Doing the bare minimum and letting the weak die on the streets led to a pretty decent life for the survivors.

According to the description of their economic model, "masks, hands, distancing" is a free option: https://koronahra.cz/about If you didn't do anything else, the 498B must be hospitalization costs. (Letting the weak die on the streets isn't an option, they'll go to the hospital and cost you.)

Full lockdown, except for some weeks: 507 deaths, CZK 2.221B used, 849M school days lost, 25.615 to 102.460 infections

Infectonator: World Dominator was much, much funnier. And it was a bit scarying how realisticly it served as a model. I stopped playing in 2020, for obvious reasons (Flash stopped updating :) ).

The maximal lockdown is way too minimal.

According to the game's about page[1], the model is based on the paper Inferring the effectiveness of government interventions against COVID-19[2]. The maximal lockdown is the combination of all the interventions studied in that paper: limit gatherings to 10 people, close all non-essential businesses, close schools, close universities, and order everyone to stay at home. The only thing more restrictive would be to put everyone under house arrest, which is simply not possible outside of an authoritarian dictatorship.

Moreover, law enforcement isn't perfect at enforcing laws. If you lock down too much for too long, more and more people will ignore the rules. Eventually people will start to protest and rebel. If the government persists, leadership will be removed through impeachment or other methods.

This game (and many other data sources) are discussed in the Astral Codex Ten post Lockdown Effectiveness: Much More Than You Wanted To Know.[3]

1. https://koronahra.cz/about

2. https://sci-hub.st/10.1126/science.abd9338

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

The UK's lockdown policies were much more strict than those outlined.

If you do nothing: 143,728 deaths, CZK 996B costs, 3.9 to 16 million infections

Reply on news.ycombinator.com