Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

A Third Solution

paulbuchheit.blogspot.com

Post reply on HN
Print view View original

131–140 of 535 posts

Re: A Third Solution

#131

Post by tlb »

jhayward wrote: post #119
tlb wrote: post #55

Earlier quoted context omitted.

A reasonable policy is to go home when the machine at the door to your office gives a positive result. Then get a more specific PCR test, and maybe come back to work. So the cost of a false positive can be one lost day. The false positive rate of lockdowns is 100%.

> The false positive rate of lockdowns is 100%. If that were true lockdowns would make absolutely no difference in R0. Clearly, they do make a difference.

I don't follow your reasoning. Maybe we use a term differently? Here's how I understand them:

- A false positive means that a test shows someone is infected when they are not. For most tests that's somewhere between 0.1% and 2%.

- Lockdown means everyone stays at home. Different from PB's plan, where only people with a recent positive test stay home.

- R0 (technically Re) is the expected number of people each newly infected person spreads it to.

When I say a lockdown has a false positive rate of 100%, it means a lockdown is the same as if you tested everyone but the test always (100%) reported positive, so everyone had to stay home every day.

Re: A Third Solution

#132

Post by Mikeb85 »

newacct583 wrote: post #112
stonogo wrote: post #88

Not only is it completely unrealistic at scale, the specific approach in the blog post is wildly impossible at all. It requires screeners to directly manipulate saliva samples; this is dangerous in a pandemic. The assays referred to (lazily) in a Google Scholar search are almost overwhelmingly antibody assays; this does not allow the screener to differentiate between "has COVID-19" and "had COVID-19". Also, there is…

Not to quibble too much with most of your criticism, but this one seems minor and trivially solved: > this does not allow the screener to differentiate between "has COVID-19" and "had COVID-19" As a policy matter, this doesn't rule out the use of the test in a pandemic management protocol at all, it just changes how it needs to be administered. For example it might require that people who are antibody-positive have a…

> people who are antibody-positive have a standardized note confirming recovery

How do you prove recovery if you were never proven sick first?

As an example, I had all the symptoms of Covid in late February, the same severity many people in my age group described, yet was never tested since our health authority dropped the ball and claimed community transmission wasn't a thing back then.

If I tested positive for antibodies, would I get treated like someone newly infected? The only way to prove recovery is to prove you have antibodies and don't have the virus, so we'd essentially have to test every single member of society.

Re: A Third Solution

#133

Post by Ididntdothis »

dekhn wrote: post #101
Ididntdothis wrote: post #7

Earlier quoted context omitted.

I tend to agree. Maybe medical people should start having opinions on software architecture after reading a few articles. They may have some fresh ideas. Or not?

I've worked in both fields. Few people who are experts in one of those fields can be useful in the opposite field, but there are some exceptionally productive individuals who are. Reading through Bucheit's article, he seems to fall into the 'wow, biology is really easy, you just have to do and the problem is fixed' camp. In my experience, that does not correlate with a good understanding of how to produce a successfu…

"you just have to do "

I have told a lot of people who said this "Just do it yourself. I don't know how to just do it". It's so disrespectful. "Just convert everything to micro services written in Go and all your problems will go away".

Re: A Third Solution

#134

Post by arcticbull »

> This could be a reason why so many otherwise young and healthy doctors and nurses have been killed by this virus.

This isn't well supported by data. In the sense that (a) young people just aren't particularly affected any more so than with the flu (old and sick people of course are much worse off) and (b) in Italy's data, no health professionals under the age of 49 died. There have been some deaths outside Italy but on average the trend mirrors the broader trend: if you're young and healthy, you're just fine. [1, 2]

I get that there's wide-spread panic, but we should allow the data to guide us where it exists. This isn't the time to spout off unsubstantiated fear-inducing commentary.

We need to keep our wits about us if we're going to tackle this effectively.

[1] Oxford CEBM, cached since it appears down: http://webcache.googleusercontent.com/search?q=cache:6rohagx...

[2] Original data here in Italian: https://www.epicentro.iss.it/coronavirus/bollettino/Bolletti...

Re: A Third Solution

#135

Post by tmsh »

If the proposed test takes 10 minutes, as Steve Jobs would say, that's not good enough. Try to make it work in 10 seconds. Like a breathalyzer. Add more liquid to the solution if necessary. I am not a doctor, but am just imagining TSA-style x-ray detectors that you walk through, while issued a disposable container to blow into, and put back into the large spinner machine.

A quick search actually reveals something:

https://www.nsmedicaldevices.com/news/astrotech-breathtest-1...

Re: A Third Solution

#136

Post by jhayward »

arcticbull wrote: post #115
macintux wrote: post #110

Earlier quoted context omitted.

But unlike the flu there’s significant spread from asymptotic people, so the young are walking time bombs.

That doesn't really change anything. To the extent the vulnerable are holed up inside, everyone else is welcome to get it, display mild flu-like symptoms, and develop herd immunity.

You repeat the fantasy that the more vulnerable population is separable from the less vulnerable.

As well you omit the size of the vulnerable - everyone who is obese, diabetic, hypertensive, or cardiovascular diseased, or immunocompromised.

The first is just a fantasy - no one has an actionable plan to do it. The second proves the size of the problem, which exacerbates the difficulty.

Re: A Third Solution

#137

Post by notatoad »

chatmasta wrote: post #78
LiquidSky wrote: post #57

Earlier quoted context omitted.

But that just moves the compliance and logistics problem back one step. Who's doing this massive amount of testing? If it's the government, you'd need armies of workers spread out everywhere. If it's the owners of these buildings, who checks to ensure compliance? Imagine trying to enforce this on every non-residential building in, say, NYC. It would be practically impossible.

Now imagine going out shopping, you’re stopped at the door, and you test positive. What happens then? The government puts you in a car and sends you... back to your apartment? Sounds like a dystopian nightmare, to be honest.

we're in the middle of a pandemic. at some point you've got to accept that dystopia is here, and the dystopian things that are happening are realistic ways of dealing with the situtuation.

you can't reject solutions because they sound dystopian unless you've got better, non-dystopian solutions. and everybody has to stay in their homes at all times and all non-essential services are shut down is not a less dystopian solution.

Re: A Third Solution

#138

Post by adpirz »

pb wrote: post #128
stonogo wrote: post #88

Not only is it completely unrealistic at scale, the specific approach in the blog post is wildly impossible at all. It requires screeners to directly manipulate saliva samples; this is dangerous in a pandemic. The assays referred to (lazily) in a Google Scholar search are almost overwhelmingly antibody assays; this does not allow the screener to differentiate between "has COVID-19" and "had COVID-19". Also, there is…

The saliva stays inside the tube for that very reason. The test specifically identifies the virus, not the antibodies, again for the reason you identify. Currently most public buildings are closed, so adding ten minutes is a big improvement relative to that. Also, it probably took them more than ten minutes to drive to work, so I don't think it's completely implausible.

Assuming everything goes to plan, how will these be staffed and manufactured at scale?

Re: A Third Solution

#139

Post by fzeroracer »

marcell wrote: post #6

> First of all, it’s not “just the flu”. It is something much more dangerous. Catching this virus is a bit like playing a round of Russian roulette. You’ll probably be fine, but you could end up dead I think this is mischaracterizing it. People have to die eventually. One year of existence has a mortality rate of 1%. For a 75-84 year old individual it is nearly 5%. Above 85 it's 14%. [1] The coronavirus infection fat…

So would you be willing to roll the dice then for the sake of opening up society?

In fact, are you willing to die for it? If given the choice between dying of COVID-19 or shutting down society, you're saying that you personally would choose to die?

I'm bringing this up because the difference between this and other causes of death is that not only is this transmissible, but it also has knock-on effects that we currently have no clue about (see: blood clotting). If we don't do what we can to prevent this, it could become far worse than what we originally thought.

Re: A Third Solution

#140

Post by danenania »

LiquidSky wrote: post #57
pb wrote: post #38

Earlier quoted context omitted.

That's why I think testing at the door is the more straightforward way to start. We can reopen factories, office buildings, even shopping malls, but no one gets in without passing the screen.

But that just moves the compliance and logistics problem back one step. Who's doing this massive amount of testing? If it's the government, you'd need armies of workers spread out everywhere. If it's the owners of these buildings, who checks to ensure compliance? Imagine trying to enforce this on every non-residential building in, say, NYC. It would be practically impossible.

It doesn’t need perfect compliance to push down the R0 significantly. Lockdowns in the US are mostly not being strictly enforced, but enough people are complying to have a major impact.

I think economic incentives are also fairly well aligned here. If tests are widespread, a significant segment of the market is likely to prefer locations that are testing to those that don’t, just like the market tends to prefer clean restaurants to unsanitary ones.

Post reply on HN