Hacker News Reader

Comment by sillysaurusx | Hacker News Reader

What does everyone think of the idea that there is an 80/20 rule in play for Coronavirus? Specifically, 80% of the infections come from 20% of the infected. I’ve heard some hearsay about it, but nothing conclusive; I was hoping someone knew either way.

View on HN

Any disease that spreads in clusters will have its infections distributed in at least a similar way. Here's the best write up on that theme I've read: https://www.theatlantic.com/health/archive/2020/09/k-overloo...

I don't know either way about your question, but aren't these sort of power law dynamics common in everything? I don't think it would be surprising if that was the case with this "pandemic" nor would it make this virus special. Just like 90% of the content is created by 10% of the people in social media, or something like that.

It's important because focussing on that 20% might be far more efficient and effective

Yeah, but this is known and taken into consideration, as far as I know.

Like health workers who visit lots of houses, gets tested often.

I don't think anybody knows who are going to be part of that 20% in advance though.

Maybe not who they are, but what they do.

I think we have already somewhat focussed on that 20%, by asking people not to hang out in big groups.

Churches and meat-packing plants are still a concern.

There was comprehensive analysis done after the 2002 SARS outbreak and they basically came to this conclusion.

https://en.wikipedia.org/wiki/Superspreading_event

> Some cases of superspreading conform to the 80/20 rule, where approximately 20% of infected individuals are responsible for 80% of transmissions, although superspreading can still be said to occur when superspreaders account for a higher or lower percentage of transmissions.

From https://www.bbc.co.uk/news/health-54648734 "Covid test-and-trace: Is backwards contact tracing the way forward?"

> "Coronavirus does not spread evenly. Most people who get the virus won’t pass it on to anyone at all. But a small number of people will spread it widely. Scientists estimate as few as 10% of infected people are responsible for 80% of infections."

Based on this, they go on to explain why, if it's true, it becomes worthwhile to do backward contact tracing to reduce R.

A colleague of mine was infected, became symptomatic and tested positive, and results took several days to come back. The whole time he was at home with his family waiting for results, he quarantined at home but had contact with his family. He tested positive, but they never did. More credence to the idea that some are super spreaders and some are not.

Meanwhile I met my family for 2 hours, I started having a fever and immediately quarantined myself to another location. I told them to open all the windows of the flat, they didn't take me seriously. They all got COVID-19 from me.

This also happened to me. I had minor symptoms, but they didn’t feel like anything worse than bad allergies. My wife is in health care and is tested twice a week so I figured if she doesn’t have it then I must not have it either. That was until I lost my sense of smell. At the time I got my positive result back from the lab I had been experiencing covid symptoms for 9 days. I was doing nothing to shield the other 3 members of my household and none of them got it either.

I read that is what the Japanese did from the start (backwards tracing), and it worked well at the time. Something like the 80/20 rule was mentioned, that a minority of infected people were responsible for the majority of the spread.

In Santa Clara county, hispanics are about 25% of the population and about 50% of the cases:

https://www.sfgate.com/bayarea/article/Santa-Clara-County-do...

Split by race, it’s not 80/20. Splitting by occupation or socioeconomic status would certainly reveal a stronger divide than splitting by race.

Probably true, but I think it's because most infections happen in certain dangerous situations (an infected person being in a badly ventilated room with others for a while), not because of some feature of that infected person.

There's a long and interesting article on that kind of stuff

>This Overlooked Variable Is the Key to the Pandemic It’s not R. https://www.theatlantic.com/health/archive/2020/09/k-overloo...

about the number k which is a more mathematically sophisticated take along those lines. Obviously reality is a bit more complicated than 80 20.

Since some people are very mobile and others are not, it is possible. But what would be the point of further discussing it?

> But what would be the point of further discussing it?

If you have a limited number of vaccines, you can vaccinate the propagators first and have 80% of the benefit of full population vaccination for 20% of the cost.

Did anyone else think that 80/20 sounds too neat, especially when you consider that the definition of superspreaders is subjective? It suggests confirmation bias and reminds me of when biologists etc assume normal distributions but when you did into the detail you find normal is actually not normal.

Power laws aren't rigid, domain-specific variables create local fluctuations even in models like population or wealth distribution. It's just a general baseline which we can use to model certain characteristics of a system.

Maybe. It's often easy to choose a model and squint to make it fit. You have to be very very careful when performing empirical method, it's extremely easy to taint data and end up fitting models that you shouldn't have. It's not enough to simply fit a power law and say "it works" - that's bad science.

It's more utilitarian than that. The power law is a tool for inquisition, you need to verify your assumptions when possible. If you can't, the power law is one of many models which you can apply to seek insight into a system.

A good heuristic for knowing which model to apply is tantamount in absence of a testable environment.

This nature paper was the best evidence I seen for this: https://www.nature.com/articles/s41586-020-2923-3?

Reply on news.ycombinator.com