Sweden do-nothing approach good, US/UK/other countries' early do-nothing approach bad. Why?

The Quarantine Zone
321

You are somewhat mischaraterising the phone apps. They are certainly not perfect, but nor are they useless. As provided in countries that use them, most, if not all, don’t track location. So there is no need to keep data on movements. Processing power is close to zero. Phones simply handshake with one another and exchange a key recording the exchange. That can be used after the fact to allow alerts to be sent to phones that made the exchange. Is is essentially a secure and opaque way of automatically exchanging phone numbers.

Distance is potentially a problem. The phones can’t tell if you are on the other side of a wall. They use Bluetooth which is inherently a short range protocol. In principle they probably could measure distance quite accurately over Bluetooth, but the power needs probably go up. The problem will be false positives, which so long as they are not too large in number won’t be an issue.

But using phone apps does really only work when the infected populace is well under control. It helps significantly in stamping out flare ups. If you are only getting a few people a day, or less, tuning up positive - you can find the contacts fast. Clearly there is a point when so many people are still becoming infected that everybody’s phone will be bleeping every day telling them they need to be tested, and it is useless. That isn’t a problem with flare ups, that community should still be in lockdown.

Phone apps don’t replace contact tracing, but they should make it a whole lot easier, efficient and effective.

Here in Oz, although we have the app, and we probably have enough downloads to start making it effective, the news so far is that it has only been used once to trace contacts for one person. Then again, where I live, we have had only one person - who flew in from the UK and was already in isolation - test positive in the last month. For us the phone app makes a great deal of sense. We can open up, and if someone comes down with the virus, even if they were clubbing furiously the week before, we could find most people at risk and get them tested. It potentially allows a much greater loosening of restrictions. But only after things have been brought under control. Not before.

322

No, radio waves “in principle could measure distance very accurately” but not specifically Bluetooth signals. They can only estimate based on signal strength.

I specifically said it was mainly useful when there are very few cases, so sure it will be useful in Australia. Bet it would crash pretty quick if you were getting 2000 cases/day.

323

It is certainly possible that there will be a greater fraction with long term residual impacts from Covid-19 ICU admissions than from influenza, but as of now nothing actually reported that demonstrates that. Generally they are in with ARDS and that is what ARDS does, usually. ARDS generally has multi-organ failure. It is bad. It is scary.

Almost 100% of those admitted to and then discharged from ICUs with influenza leave with residual damage. Longer term, H1N1 less long than others -

Of course SARS-CoV-2 is not an influenza so it may have a different pattern, maybe worse maybe better. We perhaps should look to the few who survived SARS-CoV?

Bolding mine.

A bit more hopeful.

I don’t think we are close to “millions” of Covid-19 ICU survivors yet across the globe, let alone knowing which pattern they will follow.

Anyone stating what will be for years to come is making it up.

So agreement?

324

In terms of efficacy, yes. I’ll not belabour the technical implementation issues. These are not relevant to the discussion.

325

I guess. Not to get into it but distance measuring with Bluetooth was a technical problem before covid-19, so it’s not like “now we might try and solve it”.
What’s The Difference Between Measuring Location By UWB, Wi-Fi, and Bluetooth?

326

For contact tracing you don’t need to measure distance and you sure as hell don’t need to measure location. What you need is proximity. Was someone probably within a few meters of me? With Bluetooth Low Energy, very high signal strength from a beacon typically means that it was within a couple of meters of me. Considering the fact that many people, including myself, can count on one hand the number of people during the past week that were within 5 or so meters from me for more than 5 minutes at a time, contact tracing could be helpful. Since only one app (SwissCovid) is using the Google/Apple ExposureNotification API at this time and since it just rolled out for testing on a limited basis a couple of days ago, it’s too soon to say whether phone based tracing will work. But it definitely overcomes many of the issues with the existing attempts, such as shitty operational performance on iPhones unless the app is running in the foreground on an unlocked phone, which is almost never. It also overcomes most of the privacy concerns with the other apps. This is why many of the governments behind those attempts are now looking at potential integrations with the ExposureNotification framework.

As for everyone whining about scalability, it’s not really an issue, as it is currently limited to single-step tracing. The API is meant to contact people who received a beacon signal from someone who has the virus. It doesn’t attempt to contact people who received beacons from someone who received beacons from someone who had the virus. So, if every single person who was close enough to trigger a proximity beacon became infected, I’d receive 5 notifications a week. I think I, and my phone’s operating system, can handle that. If that number was 2000 week (if you actually share beacons with that many infected people in a week, you have other problems as you are likely an emergency room specialist in New York City), my phone still wouldn’t crash. In theory, they could eventually look into multi-step tracing, as scoring is part of the current process. Perhaps if you were exposed to one person with extremely high exposure scores, or many people with moderate scores, etc. We’re way too early to worry about that, as multi-step won’t work if single-step is ineffective.

327

Uh huh. Well, I’ll stop “whining” when I see it working. Sound fair or is that too obstructionist for your tastes?

328

I’m also interested in seeing whether it will work or not and we should hopefully start seeing some results within 4-6 weeks, if they are able to roll it out widely in mid-June as planned. I posted to show that many of the arguments that have been made against this possible method of contact tracing were factually incorrect or based on a flawed understanding of the technology.

329

I don’t think so. You quibble about proximity detection and distance measuring. The fact is, we want “proximity” of about 6 feet. How sensitive we set that will, as you say, have a massive effect on the number of contact “hits”. And it isn’t a “fact” worth noting that you personally have come into contact with only 5 people this week. A great number of people are not like you.

330

It ain’t a quibble as you’re the one telling us that app based tracking isn’t feasible due to the fact that they have issues with distance measuring. BLE distance measuring is hard, while proximity measuring is easy, so the difference between the two is hardly trivial. The API in question isn’t attempting or claiming to measure distance, simply the strength of the signal subtracted from transmission power. Due to the inherent limitations of BLE beacon technology, proximity is often categorical, not continuous, and will be limited to labels such as immediate, near, far, and unknown. This API is a bit more granular and provides 8 bins for attenuation level based on transmission power - RSSI. The app developer can choose to give each bin any value between 0 and 8, so they get to choose whether to have 8 discrete attenuation level scores, or give multiple bins the same values. Maybe they want everything in the top 3 attenuation level bins to have the same value. Sure. Perhaps they want the lowest 4 bins to not have any risk associated with them. Can do. That particular score is also included in their exposure risk value formula, which is easily weighted however the app developer would like as I showed above.

I have no idea who “we” is, but “I” am okay with using categorical proximity values, since we can’t actually measure distance accurately (although Bluetooth 5.1 is pretty damned accurate under certain circumstances, thanks to AoA and AoD). It’s also typically quite good at determining if someone is within a few feet of you. Perfect is the enemy of the good in this situation, so if it sometimes pings someone 15 feet away and once in a blue moon it misses someone 8 feet away, I still find it a better option than throwing my arms in the air and proclaiming “Waste of resources.”

And most of that great number of people will have a very similar experience to me, no or few risk messages in any given week. How many people who actively have coronavirus do you think you spend more than 5 minutes in close proximity per week? If the number is greater than 1 or 2 and you’re not a healthcare professional, nursing home employee, grocer, prison warden, or for some stupid reason, tattoo artist (and any other “essential” service worker), you’re doing something wrong.

Again, I don’t give a flying fuck if you like the app or not. I also don’t have any idea if it will end up being widely adopted and properly utilized or not, but I’d much prefer that the smart people of Apple, Google, and the EPFL work on a solution than not. Since it has an open-source reference server implementation, has fully documented what it will not allow to be tracked (it won’t even allow the app to get coarse GPS coordinates), has short-lived random device IDs that are changed multiple times per hour, and runs fine even on a locked phone, I’ll happily install it if my federal, state, or local government ever get on board. If it’s available, I think it would be great to complement this with traditional multi-step contact tracing. The app is great for things that are missed with the traditional method, such as contact with strangers, while the traditional method is great for identifying hot spots and iterating through contacts of contacts, which the app won’t do.

331

That’s just one of the reasons I think it’s unfeasible. Anyways, this is a hijack in this thread.

332

As to the technical I plead cluelessness but taking your versions at face value DMC,

The goal of app-based contact tracing is for use as people get out more, returning to work, commuting, eating out. The numbers of people who will have many hundreds within 5 or so meters of them for more than 5 minutes at a time, is going to be HUGE. Living in a world (true in some societies) in which some critical mass both cared and trusted enough to download and use the app, the difference between accurately pinging based on 6 feet/5 minutes or a wider rough proximity circle versus including those 10 to 15 feet is a very very big number of “false positive” pings. The area, and thus likely number of people, contained in a 10 to 15 foot diameter is much larger than the 6 foot diameter.

Please remember what is supposed to happen if pinged to having had contact. The person so pinged is suppose to both get immediately tested and, if negative, self-quarantine for two weeks. False positive pings have to be rare, not more frequent than not. If the understanding is that false positives are common very few will act upon them.

Whether or not YOU are okay with such crude categorical proximity values is pretty immaterial. Alarms that go off too often with high of rates of false positives are a major source of poor outcomes referred to as “alarm fatigue”.

Whatever number of flying fucks you do or do not give the only way anyone gets more than one risk message every other week is by not acting on the risk messages they have gotten.

Now IF the true rate of infected individuals is very very low then very few would be experiencing false alarms even with frequent pings when not within 6 feet/5 minutes. In that case prioritizing low false negative rates might make sense.

333

Well, obviously his 5 contacts/week is fucking ridiculous as any kind of estimate of what a contact tracing app would have to work with. If we were back at work, I would directly contact ~40 people per week and be on the other side of a thin wall of dozens more.

This dependence on future technologies that the “follow the science” crowd seems to advocate is exactly what got me so frustrated and annoyed that I got suspended here. I am a “follow the science” type person but I seem to be in the minority that thinks “invent something that fixes the problem” isn’t really part of a plan. That’s something we do while we work with the present situation.

334

…just to be clear here: there is a difference between the “follow the science” crowd and the “follow the tech” crowd. I hate to bring NZ up yet again, but I will :slight_smile: We have a government app here: but its a very simple app, all it does is scan the QR Code of the business you happen to be at. No bluetooth, none of the other fancy things that the other apps have, and our tracing regime isn’t dependent on the app at all, instead relying on good-old-fashioned-legwork.

What the science suggests is the best way to deal with (this) pandemic is to lock down early and properly to flatten the curve. Then lockdown a bit longer to get through incubation cycles to both break cycles of transmission and give yourself time to ramp up testing and tracing. I’m actually with you on the apps. You can’t “tech” your way out of this. I’m all for new technology to help with tracing where possible. But you can’t just jump to the tech and ignore everything else that needs to be done.

335

Oddly, I think we kind of agree here, except on the tech/science distinction. Some people think science says “wait for the tech”.

336

Tegnell, the guy behind Sweden’s coronavirus strategy now admits that mistakes were made in their strategy. He says that if he had to do all over again, he would do something between what they did and what the rest of the world did.

Sweden’s Prime Minister Lofven has promised that there will be an investigation into the handling of the crisis. As Sweden has been locked out of more countries due to their high rate of infection, more of Sweden’s citizens are questioning their strategy.
Man Behind Sweden’s Controversial Virus Strategy Admits Mistakes

Coronavirus: Sweden’s Tegnell admits too many died

337

Coronavirus tracked: Charting Sweden’s disastrous no-lockdown strategy

  • Currently the highest death rate in the world
  • No signs of dropping
  • Stockholm has a lower percentage of the population with Covid-19 antibodies than London, New York, and other major cities.
338

The highest death rate is a meaningless number. It just depends who was tested and how many tests were given. Still far too early to say their do nothing approach has been a disaster. Circulating a virus through faster reduces opportunities for infection in the long run

339

Testing of people who die with corona symptoms? Please give us to cite to show that such testing in Sweden is higher than elsewhere.

Herd immunity? Less than elsewhere, as shown in the article above.

You seem to be grasping at straws, without looking at the facts.

340

It is a fact that Sweden has lower excess mortality than many other countries, which certainly suggests that a good number of cases have gone unreported in those places. Some countries don’t even include deaths outside of hospitals, which seems statistically suspect.