Contact Tracing in the Real World
lightbluetouchpaper.org
lightbluetouchpaper.org
Covid-19 in S. Korea @5:13
https://www.youtube.com/watch?v=BE-cA4UK07c
My opinion is those apps work because S. Korea remembers SARS-1 (2002/2003), and people there are cooperative. Also, if you do full contact tracing early enough, there aren't an overwhelming number of patients like NY.
S. Korea is the only country that literally "flattened the curve." See unbelievably flat graph @6:50 in link above.
(S. Korea had the same problem as Toronto with SARS-1, where the hospital doctors and nurses were wiped out by the first few cases. In S. Korea, it was the first SARS-1 patient wandering around town like a cruise missile to several doctors and hospitals that spread the disease, as nobody identified the magnitude of the problem.)
> And how will things work with an orderly supermarket queue, where law-abiding people stand patiently six feet apart?
In the SF Bay Area, here's how social distancing works at grocery stores:
1) one entrance/exit with security guards at the door
2) queue outside with shoppers 6' apart
3) only 50 people allowed in at one time, but distance is not enforced in the store except 6' in checkout lanes. However, American grocery stores are large, so the spirit is being followed. Recently, I've seen saran wrap over POS terminals. Bank ATMs also need that prophylactic. :)
4) shoppers are responsible for washing their carts and hands, which is a gap.
Taiwan has also done so. It has a similar story, burned by SARS, also.
They basically don't even show up on graphs that show positive test counts for other countries.
I've had people argue that this is containment rather than flattening the curve, but I think it's both, it's containment, which is roughly the limit condition for flattening the curve.
Now they are showing us the future, with new outbreaks showing up.
Hong Kong, for its part, reported only 4 new cases on Sunday and has been testing aggressively - I think they're around 15,000 tests per million people now.
MOH provided an annex[1] breaking down new cases by transmission type in the latest update. The data shows that local cases had been increasing, but they're in a slow decline - for now - and most of the new infections are in or related to dorms.
1. https://www.moh.gov.sg/docs/librariesprovider5/pressroom/pre...
I suppose that's one way of looking at it.
But using the "hammer and the dance" metaphor, if you start the dance (containment) early enough, there is no need to bring the hammer (because no big spike to flatten). https://medium.com/@tomaspueyo/coronavirus-the-hammer-and-th...
https://www.washingtonpost.com/world/aggressive-testing-cont... or https://outline.com/bPStnx
https://en.wikipedia.org/wiki/2018_Nipah_virus_outbreak_in_K...
Credit card processors and stores working to turn off signature verification (the olds will just have to deal) would make a bigger difference than infrequent cleaning anyway.
I'd say the biggest vector of disease is people not washing their hands before or after visiting the supermarket, and the free-for-all aisle traversal. It would probably take one infected person to sneeze in an aisle to infect lots of other people traveling through said aisle. or infected employees touching various products with their snotty hands.
I can't tell if that's significant, but as a general rule I assume supermarkets are places to visit as infrequently as possible, and to wash my hands before and after.
For the most part it does seem people are changing their behavior though, around here. we pass each other with appropriate distance, and we stand in line with appropriate distance.
Actually, at least per Vox, it was more about tracing problems learned during MERS in 2015:
And both sides of the political spectrum are behaving this way, the right-wing downplays the risk and postures that a lockdown will lead to a worse outcome, and many dems pretend a lockdown is the only thing that could have realistically been done, looking to SK like "oh, well they're draconian, that could never happen".
That said, I’d gladly install a decent security verified app that tracks me for the next 3 months. I’d be happy to self isolate again with my family (as we’re doing now) if I came in contact with someone. But the big issues are:
- ensure tracking is available only to those who need it and doesn’t get shared with any commercial interest.
- my employer supports WFH or short term paid sick leave (maybe with gov’t assistance) for those 14 days.
Get those two things and I think the majority of the US population would opt in. That would be sufficient.
Sure there’s some risk that people are infected who won’t opt in. But that opt in system of “we’re all in this together” is much better than a forced gov’t tracking system.
Privacy aside, it seems to me that there are two main issues that Bluetooth-beaconing-smartphones will struggle to solve:
1. Determining whether two smartphones "hearing" each other's beacon constitutes contact between the owners of those smartphones. (The blog post we're discussing brings up several examples, and the "orderly supermarket queue" of shoppers patiently waiting outside while observing a 2m distance could generate a massive number of false positives.)
2. Ensuring widespread adoption. If only 1/6th of the population is participating (which are the numbers Singapore is seeing so far) then the chance that an encounter between two people is captured by the app is 1/36th -- far too low to be of much use. There is also an upper limit to adoption because not everyone owns a smartphone or has a device which would be compatible with the contact tracing app.
I don't think these distances are absolutely safe and it's impossible to say exactly what interactions occured.
The data from Bluetooth beacons are probably as good as is possible to get.
If you're not prepared to test all of those people (and potentially multiple times) then what action should the potential contacts take? Should they all isolate themselves for 14 days because they were in the same carpark as a confirmed case for a few minutes?
(to be clear, I'm operating a 'touch means take' policy, but its clear through simple observation that most aren't capable of thinking through the rule (or even following it))
As mentioned it's probably not possible to get better than that with a technological solution but it is much better than not using technology. It's also difficult to define what 'better' would be. Seems to me that all those people should indeed be careful and self-isolate.
I'd also note that the comment you're replying to didn't really address your two original points. Instead it stated some technically-correct but conversation-misdirecting facts.
At the moment we are all isolating for months. So yes if 5% of the population needs to isolate because they were 10m away from someone who had it then so be it. I'm perfectly fine with that.
But in any case it's all on a curve, the further away you get the less likely a person will be infected.
And yes it's a curve so the risk depends on the minimal infective dose. If you can get infected by a small number of virus particles (which is apparently the case) then you need more distance.
I assume a covid-19 contact tracing system would use this, among other things, to filter out contacts that don’t meet the established criteria for high enough risk of transmission.
Currently we all have to change our behaviour around everyone, all the time. Even if you are in New York that probably means there are “false positive” behaviour changes (behaviour changes for no reason) all the time – simply because we lost control of the situation and cannot currently contact-trace our way out of that situation.
The idea behind these contact tracing apps is that they make only really sense once you have to outbreak well under control (through drastic measures). The goal is then to relax measures (probably not to abandon all of them – handshaking is probably dead and the recommendation will remain to stay away from other people in public and forget large gatherings for a long time, as well as a lot of travel …) and be much more targeted with who you impact through such contact tracing.
Obviously they cannot replace proper contact tracing, but they can be a helpful tool in the tool-belt and increase the effectiveness of standard contact tracing and testing.
If there are, say, 500 detected active cases in a country like Germany (which is probably around the limit of where contact tracing still works, even with many more people working contact tracing and much expanded testing capability) contact tracing apps can provide an additional signal and increase the speed of standard contact tracing approaches, especially since we know that much of the spread is pre-symptomatic.
Even if this tool overestimates the infectious contacts someone had hundredfold and we assume a ridiculous R0 like 10 (so someone infects ten people on average but a hundred times that many people are actually detected by contact tracing – so 1,000 people altogether) that would still only place at most 500,000 people in quarantine (assuming an upper limit of 500 detected active cases that can be handled before you lose control again and have to place the whole country under lockdown).
That’s a huge number but probably not as impactful at all as the current measures that are in place. (Obviously you need to put measures in place to make that possible: mandatory, automatic and paid sick leave for anyone who is told to stay at home for a week or symptom-end plus one week. Stuff like that. Some countries are probably better prepared for that than others.)
Even this blunt instrument would be much better than the current situation.
If contact tracing were enabled and everyone at any risk self quarantined this nightmare would soon be over. Instead people will justify why they're low risk, interact with others and continue the spread.
> First, it isn’t anonymous ...
I believe this is a miss understanding. We loose anonymity once someone is diagnosed, there is no way around that. But we should not skip on anonymity for everyone else. Else this whole project is doomed by the onset as we are simply building a massive 1984 style infrastructure.
> Second, contact tracers have access to all sorts of other data such as public transport ticketing and credit-card records ...
Which is debatable, but not a major problem as long as this is a largely manual ad-hoc process regulated by law and so expensive to execute that we only do it during emergency situations. It's totally different once we start building systems and processes around this to make it cheap and it ends as the new normal.
> Third, you can’t wait for diagnoses. ...
Surely a problem we need to improve upon. Likely a major logistics problem? Maybe something we should ask the military to take a look, after all logistics is something they might have an idea how to do right.
> Fourth, the public health authorities need geographical data for purposes other than contact tracing ...
Why would this need to be part of contract tracing? Currently this is done based on confirmed cases. Surely they are largely localized, especially since the stay-home orders.
> Fifth, although the cryptographers – and now Google and Apple – are discussing more anonymous variants of the Singapore app, that’s not the problem. ...
I agree, trolling, denial of service and school kids are going to be a problem. Could likely be solved by adding a formal gate for infected. Signing the test result with a public-private key scheme should do the trick to prevent any non-legitimate results being used to trigger events.
> Sixth, there’s the human aspect.
Interestingly I would have said this is the technology aspect. How many contacts will we miss because bluetooth low energy is not up to the job?
> Seventh, on the systems front, decentralised systems are all very nice in theory but are a complete pain in practice as they’re too hard to update. ... Relying on cryptography tends to make things even more complex, fragile and hard to change.
Seems solvable, considering we are talking about mobile apps, mostly always online, managed through central app stores and servers? Sounds doable.
> But the real killer is likely to be the interaction between privacy and economics. If the app’s voluntary, nobody has an incentive to use it, except tinkerers and people who religiously comply with whatever the government asks.
Tend to agree, although we have seen quite some adherence to sensible rules the last couple of weeks. Maybe we will be surprised by our fellow humans, maybe not.
If a solution generates that much sceptisism, that many questions and requires that much energy to convince each others, it's a big red flag.
Operation Overlord was nothing any sane person would attempt if there was a viable alternative.
People pushing the "we must do that or millions will die of this deadly virus" have been proved wrong so far
As you say, we have other measures to save lives, but it seems they are rather deadly on the economy, and human (social) life as we know (and like?) it.
Improving medicine: decades?
Improving vaccines: 12 ~ 18mo
Improving detection: ???
Isolation of elderly: presumably already in place (are you suggesting something more?)
Improving healthcare: ??? (anything specific?)
I don't think any of these are going to make much of a difference to this pandemic. The vaccines eventually, perhaps. The things that have been shown to work well are NPIs (non-pharmaceutical interventions) such as: contact tracing, testing, mask usage, social distancing. The system that Apple/Google will put in place is essentially automated contact tracing. It will be much faster and have more data than manual contact tracing.
There isn't even a scenario I could imagine where one of these apps would do something that I would consider a severe infringement of my privacy.
A few million lives may ultimately hang in the balance, so there ought to be a very compelling argument against not putting these apps into practice and getting ahead of coronavirus.
The problem is that sometimes people who don't yet know they are infected have contact with other people before anyone in the story finds out that they are infected. The apps allow the communication to happen:
1. Even if the people who had contact don't know each other.
2. Faster.
That's how we get ahead of the virus.
Not sure what the downvote is for.
1. The US, despite currently suffering the most, will probably not end up being the country with the most casualties. Once covid-19 takes hold in more densely populated poorer countries, there will be immense suffering.
2. 1918 flu cost up to 100 million lives, 1957 flu about a million lives, and 1968 flu about a million lives.
3. Total deaths worldwide is already >100k, and the known cases are still less than 0.03% of the population.
Would I give up my privacy to save a few thousand lives? Yes, but the scenario is incorrect. There isn't really a privacy sacrifice involved. I'm trying to think of the worst-case scenario in terms of privacy, but there just isn't one. Someone finds out that a device with ID 00:11:22:33:FF:EE spent 15 minutes with a device with ID 01:31:24:53:EF:AE. It's not worth so many people losing their lives for.
source: https://eu.usatoday.com/story/news/investigations/2020/04/09...
OP is absolutely right.
Countries with lower standard of healthcare than the US will have a higher death rate.
For example, in NYC, this chart shows roughly 5500 excess deaths on april 4th. Whereas on that day nyc had only counted 3500 coronavirus deaths.
https://mobile.twitter.com/paulg/status/1248905714124414976/...
Further, those numbers are based on NYC hospitals being completely overwhelmed which sky rocketed the death rate. If a country can flatten their curve what is happening in New York doesn't have to happen again.
1. Yes, there was that dweeb with the NYT opinion piece saying it wasn’t worse than the flu. There’s a reason that ran in the opinion section.
Additionally, many many apps silently gather and sell this data, without your knowledge, and with "consent" being ignored or being relegated to privacy statements that are too vague to understand that they are doing this.
I really hope that if people are concerned at all in any way about the tracking app mentioned here that they also learn that such tracking is being used extensively in industry right now.
I'm surprised how few people on HN seem to know this. If any audience should know and warn our friends and neighbors, it's the audience at HN.
It's like people want to suffer.
> There isn't even a scenario I could imagine where one of these apps would do something that I would consider a severe infringement of my privacy.
do you live in the united states? there have been quite a few privacy fiascos in the last couple decades that the general public found unimaginable until they were exposed.
But they'll just say "the information is only from this source, we pinky promise!".
Also, define how you (effectively!) quarantine sensitive populations with less impact on civil liberties than testing and contact tracing of the general population (which has modest privacy implications that can be reasonably managed).
(1) https://www.aier.org/article/what-sweden-has-done-right-on-c...
it's hard to see how contact tracing of the general population can be done with "modest" privacy implications, unless it's opt-in.
Protecting the people they care for means imposing restrictions on them. And so on for all sorts of other people that would be involved in (effectively!) isolating sensitive populations.
The reason contact tracing has modest privacy implications is that the location history of most people isn't sensitive information (or exists already, with little additional harm incurred by giving it to public health workers). It's fine if you disagree with me on that point, but that's my argument, it isn't super serious if Bill over at the county health department sees that you encountered a sick individual at Wendy's, which is going to be the majority of the data.
I wouldn't necessarily oppose stricter measures for people who do this sort of work. easy for me to say when I don't work in a nursing home, of course. in principle, it would be comparable to the privacy you give up when you have to get a security clearance for your job.
> The reason contact tracing has modest privacy implications is that the location history of most people isn't sensitive information (or exists already, with little additional harm incurred by giving it to public health workers). It's fine if you disagree with me on that point, but that's my argument, it isn't super serious if Bill over at the county health department sees that you encountered a sick individual at Wendy's, which is going to be the majority of the data.
I can certainly understand that my position looks a lot like closing the barn doors after the horses have already run five miles down the road. google has the gall to actually send me an email each month showing all the places it has tracked me going. even if I delete my facebook account, it can still recognize my face in pictures other people take.
I'm not happy about any of this, but at least now, the government has to put on some charade of formality to access this information. giving it the power to directly collect this data in plain view is another step in the wrong direction. it adds more stuff to the pile of things we need to undo to get where I personally want to be.
I don't think your argument is unreasonable, just that we're approaching this from different perspectives. you're thinking about "how many people can we bring with us to the other side?", while I'm thinking more about "what do we want the other side to look like?".
Do you realise that immunity for this virus may not be lasting?
Your plan requires millions more lives be sacrificed just so that the ID of one of your many electronic devices isn't recorded by some other devices. It's already public anyway, by the way. That said, this plan wouldn't happen anyway. People will not willingly be infected. Herd immunity through infection is not possible.
If it looks so, that's only thanks to lockdown. There were days where the US saw daily 40% increase: with that kind of curve, it takes five days from "hospitals 80% idle" to "full", and another five days from that to "five times more patients than we can handle".
And since it takes about a week for infected people to show symptoms, when you put the brake again and re-implement lockdown, you will be looking for another five-fold increase before it slows down.
S.Korea and singapore/hong kong, those are densely populated areas/cities with specific movement patterns among people and enough tehcnology infrastructure to know what do something useful with the tracing information. In some western cities, tracing would end up giving people constant alarms that they might be infected, and that information wouldn't really be actionable or at least you coulnd't trust that people would act on it. Western cities have no memory of other plagues either.
Perhaps goverments should reward distancing and testing instead of punishing offenders. Perhaps even pubs could open, provided they serve the same group of 10 people every monday. Western cities will need to figure out their own policies until a cure is found, and they have to reflect their own character - you can't just copy asian policies praying that they work.