A critique of contact-tracing apps

Here are some relevant criticisms from Soltani, Calo, and Bergstrom:

Studies suggest that people have on average about a dozen close contacts a day—incidents involving direct touch or a one-on-one conversation—yet even in the absence of social distancing measures the average infected person transmits to only 2 or 3 other people throughout the entire course of the disease. Fleeting interactions, such as crossing paths in the grocery store, will be substantially more common and substantially less likely to cause transmission. If the apps flag these lower-risk encounters as well, they will cast a wide net when reporting exposure. If they do not, they will miss a substantive fraction of transmission events. Because most exposures flagged by the apps will not lead to infection, many users will be instructed to self-quarantine even when they have not been infected. A person may put up with this once or twice, but after a few false alarms and the ensuing inconvenience of protracted self-isolation, we expect many will start to disregard the warnings.

And:

At least as problematic is the issue of false negatives—instances where these apps will fail to flag individuals as potentially at risk even when they’ve encountered someone with the virus. Smartphone penetration in the United States remains at about 81 percent—meaning that even if we had 100 percent installation of these apps (which is extremely unlikely without mandatory policies in place), we would still only see a fraction of the total exposure events (65 percent according to Metcalf’s Law). Furthermore, people don’t always have their phones on them.

And:

There is also a very real danger that these voluntary surveillance technologies will effectively become compulsory for any public and social engagement. Employers, retailers, or even policymakers can require that consumers display the results of their app before they are permitted to enter a grocery store, return back to work, or use public services—is as slowly becoming the norm in China, Hong Kong, and even being explored for visitors to Hawaii.

 

Taken with the false positive and “griefing” (intentionally crying wolf) issues outlined above, there is a real risk that these mobile-based apps can turn unaffected individuals into social pariahs, restricted from accessing public and private spaces or participating in social and economic activities. The likelihood that this will have a disparate impact on those already hardest hit by the pandemic is also high. Individuals living in densely populated neighborhoods and apartment buildings—characteristics that are also correlated to non-white and lower income communities—are likelier to experience incidences of false positives due their close proximity to one another.

In another study:

Nearly 3 in 5 Americans say they are either unable or unwilling to use the infection-alert system under development by Google and Apple, suggesting that it will be difficult to persuade enough people to use the app to make it effective against the coronavirus pandemic, a Washington PostUniversity of Maryland poll finds.

And here are skeptical remarks from Bruce Schneier.

I also have worried about how testing and liability law would interact.  If the positive cases test as positive, it may be harder for businesses and schools to reopen, because they did not “do enough” to keep the positive cases out, or perhaps the businesses and the schools are the ones doing the testing in the first place.  Whereas under a lower-testing “creative ambiguity” equilibrium, perhaps it is easier to think in terms of statistical rather than known lives lost, and to proceed with some generally beneficial activities, even though of course some positive cases will be walking through the doors.

I wonder if there also is a negative economic effect, over the longer haul, simply by making fear of the virus more focal in people’s minds.  The plus of course is simply that contact tracing does in fact slow down the spread of the virus and allows resources to be allocated to individuals and areas of greatest need.

Comments

Somehow, the South Koreans figured out a way to use cell phone data for pandemic suppression in an effective manner, taking a couple of weeks to have it be functional.

Obviously, the United States is following another approach entirely, in part because the South Koreans did not use any apps, they simply used the already existing infrastructure of a cell phone network. People worried about privacy in this context are amazingly naive, as that cruise ship sailed long before the pandemic. Any concern about false positives or negatives sounds more like another convenient excuse for ongoing failures in responding to a novel viral pandemic.
Liability concerns is another part of this global pandemic that sounds exceptionally American.

Even Singaporeans are mocking stupid Trump followers with coronavirus comic book villains. Auntie Biotic is "head of Trump’s disinfectant-injection task force"

https://www.facebook.com/AGoodCitizen/photos/a.3048482388544498/3048482491877821/?type=3&theater

Funny! Singapore has had 15000 cases and just 16 deaths. Some might call these figures "unbelievable".

we figured the phone app tracing would first be an epidemiological research tool app not a literal contact the potential peep by text app!

90% of TC's covid article posts should begin with, "First assuming S Korea doesn't exist, the authors find..."

If only the US had planned ahead and been an island with only one primary infection vector. Truly a failure of vision.

Commenters more serious than the troll might draw some conclusions from the fact that no other nation in the world managed to do what the South Koreans did.

This is all vaporware till now.

Maybe if Apple and Google were given a million dollar grant, they would be able to provide an app faster. /sarcasm

I saw a bit somewhere about a "contact tracing" app that would treat any two people who spent more than 10 minutes within *50 meters* of each other as likely "shared exposure". It's less than 50 meters from the center of my house to that of my neighbor's houses - neighbors I hyper rarely see or speak with. This is absurd. What, everybody in any apartment building will be deemed "always in contact" even when physically far apart?

Cell phones do not know when two people are in the same *air space* without masks, which is of course what matters.

In the mean time, stay-at-home orders seem to have only a finite effect (they've certainly NOT driven the infection rate to 0 in WA state) I conjecture this is in part because the number of people who are truly essential, and dependency graph of people THEY depend on, adds up to quite a lot of the population. There's a maximum amount of isolation that is possible in our society.

"within *50 meters* of each other as likely "shared exposure"

You probably saw something that said, something like "Bluetooth Low Energy can reach up to 50 meters, so it'll track every phone that comes within that range!" But in practice, you can use the strength of the signal to roughly guesstimate how close another phone is

Isn't the whole point of a test and trace return to normalcy the testing part? YOU might not want to be bothered with a "false positive" notification because your neighbor is sick, but I would feel a lot safer getting a tested to make sure I didn't get the virus if my neighbor had it. Even more so in an apartment with shared hallways and entrances. The key to making these apps work is being able to test anyone that came near someone with the virus. Tests should be free and are a lot lower hardship than locking people down for 2 weeks.

How long until we finally give up and admit that herd immunity is the only way forward? Yes, it would cost between 500k to 1m deaths, but the only other realistic alternative is permanent lockdown. Contact tracing is a pipe dream - too many cases are out there and the US is unwilling to restrict intra-county travel (except for Hawaii and Alaska).

What do you mean give up? Herd immunity is only a valid concept in terms of vaccination, and there is no way we are giving up on that any time soon.

A disease becoming endemic is not, and never has been, an example of herd immunity.

Once enough people get the coronavirus, the median R0 will drop below 1 and thus the pandemic will end. People will still get the virus in localized flare-ups, but it will no longer overwhelm the hospitals. And finally at some point (September 2021, based on the original 18 months estimate?) we get the vaccine and semi-eradicate it.

Endemic means that though the pandemic ends (by a reasonable definition), it also means that the disease never goes away. That is not an example of herd immunity, it is an example of a public health failure, though possibly one that is inevitable at least in some countries.

And it is still unproven whether catching the virus confers immunity over the longer term. There is no way to answer questions about this virus when looking over even a 12 month time span, as the virus is 6 months old.

Given what we know so far, it is inevitable in the US. It would be far less painful to admit so and start talking about getting this over with.

Getting what over with? We hope that being infected confers longer lasting immunity. Much like we also hoped that asymptomatic/pre-symptomatic spread was not occurring.

Apart from a vaccine or South Korean/Australian/Austrian style measures, there may be no getting this over with.

If immunity lasts for at least a year, its good enough, as we'll have a vaccine by then

Are you certain about that?

Dead certain, given the experiments on reinfection in monkeys and the insanely sharp decrease in mortality in NYC. Herd immunity works.

Honestly, anyone using the term herd immunity in any context not involving a vaccine is simply showing they have an economist's level of understanding concerning epidemiology.

Literally on a national level many head epidemiologists have talked about a level of herd immunity building up as a disease moves through the population (not as a goal, but as an unavoidable outcome of other strategies).

That the "Crush the curve, eradicate the disease, and continue lock-down and monitor everyone else's movements full time until this is achieved!" crowd don't want to accept that any level of immunity can exist in a population without vaccination, with a disease simply moving through the population is.... not a sign that they understand epidemiology particularly well.

'don't want to accept that any level of immunity can exist in a population without vaccination' - Define immunity. We live with a disease that kills thousands every winter, even with vaccination. It is clear that the term herd immunity in regards to influenza is misplaced, at best, even when referring to the fairly consistent influenza b that regularly recurs.

The basic argument is between nations like Austria and South Korea and Greece, which have been able to contain this disease to an acceptable level, and nations that clearly are unable to manage corona, like France, the UK, and the US.

Epidemiologists are likely aware of the background of the term - that is, when a disease like measles infects a broad number of kids, in the following year or two, the number of new measles infections is not as high as before, much like the peak and trough of a wave. However, over a longer period of time - 5 or 10 years - the rate of measles infections, and its accompanying problems, is basically the same as if that peak and trough had not been observed in the first place. And to stick with measles - herd immunity only makes sense with a vaccination, apart from that narrow technical perspective.

It is pretty difficult to imagine uniform herd immunity across every locality in a country such as the U.S. I don't think there is historical precedent for it.

Instead, a level of stability may be achieved as long as restrictions not just on international travel but also domestic travel are kept in place for long periods of time. Otherwise, as with measles -- which is much more contagious than covid-19 and is about equally as lethal -- there will continue to be outbreaks in communities that were not previously vaccinated or exposed when an infected person visits there.

Measles was never eliminated from the U.S. through herd immunity but only through vaccination. Once we got too lax about vaccination, it was reintroduced into some populations and flared up again.

No one is proposing elimination of the disease through it moving through the population, just that healthcare systems will not likely be overwhelmed, and that it is more or less inevitable that the disease will move through much of a population before vaccination, however effective initial prevention has been (and that even in those places where prevention has been possible, year end excess mortality may not be too different).

"more or less inevitable that the disease will move through much of a population before vaccination"
None of this will happen if Rt is kept < 1 most of the time.
Of course, this cannot happen because moronic politicians and moronic lolobertarians don't understand the gravity of the situation.

Your uncertain "getting this over with" is not as simple as you can imagine.
You cannot afford to have the health system collapse completely and to ensure it does not happen, you will have to stretch infecting the 70% of the population necessary to at least a period of 7-8 months.

During which at least partial lockdowns will be necessary.
Given your confidence in a vaccine in a ~year, it's better to not infect those people.

Due to exposure variance (see https://thezvi.wordpress.com/2020/04/07/on-r0/), infections will rapidly start falling once you reach 20% infected, as roughly speaking 20% of the population are responsible for 80% of the interactions. Even at 10% you'll start seeing a significant reduction as the people who are most socially active will all infect each other. Once you get to 20% (already done by NYC, Lombardy and perhaps a couple of other cities), you could relax more and more restrictions as the R0 will be naturally suppressed thanks to partial herd immunity. Only reason why NYC cannot reopen everything right now is that people from neighboring states will all flock there (Cuomo calls it "attractive nuisances"), otherwise they could probably remove all restrictions tomorrow and only rely on basic caution. So my prediction is that the US could remove all government intervention (except for huge indoor events, perhaps) by the end of summer, if they carefully plan the herd immunity strategy. Cases would still happen until a vaccine is available, but no longer at the initial explosive pace.

A massively available vaccine is not coming prior to September 2021. Its a heck of a lot of time.

"20% of the population are responsible for 80% of the interactions."
They are responsible for being one party in 80% of the interactions and it's not clear that they are one party in 80% of all transmissions since transmissions are more likely to happen during intimate contact.

It just does not work that way.
You will infect 60-70% of your population, killing 1% of them and severely disabling probably 5% more.
All that for immunity that last a year and you're back to square one.

Better institute a system that will be able to keep Rt<1 most of the time because if we don't get a vaccine, you'll need to do that anyway or civilization collapses.

"You will infect 60-70% of your population, killing 1% of them and severely disabling probably 5% more."

As per NYC serological data, ~2% require hospitalisation and ~0.5% will die. 2% is thus the upper bound on the number of people with any long-term disabilities. And we already have anecdotal data who show great lung function recovery after returning from the hospital.

"Better institute a system that will be able to keep Rt<1 most of the time"

The current system gives us a more-or-less constant R0 of 1.0, which would result in the same number of infections by the time a vaccine arrives, but would also cripple the economy for those 12-18 months. Might as well recognize that and speed up the process.

The fact that you don't require hospitalization does not mean you will not have long term lung sequlae:
https://www.rainews.it/tgr/tagesschau/articoli/2020/04/tag-Coronavirus-Lungeschaden-Forschung-Uniklinik-Innsbruck-6708e11e-28dc-4843-a760-e7f926ace61c.html

We also have anecdotal data of people who were never tested (or hospitalized) that report sequlae 2 months in.

"speed up the process."
You cannot speed this up significantly because even with Rt of 1.3, you'll be forced to lock down again in ~2 months.
Your suicidal tactic will just ensure that a part of the workforce is permanently disabled without solving anything long-term.

"Your suicidal tactic will just ensure that a part of the workforce is permanently disabled without solving anything long-term."

The current R0 = 1.0 tactic will achieve the exact same number of infections by September 2021, but at a far higher economic cost. Presuming we stay at ~1500 deaths/day and a 0.5% fatality rate, we'll end up with 730k dead and 146m recovered by the time we get the vaccine.

Not just imagine if we have an Rt = 0.9 instead.

I would prefer if we had R0 = 0.3 (Wuhan's estimate), but that would require an effective centralized response with well guarded borders. Given quarantine fatigue and a general push for reopening, the US will keep oscillating between spikes of R0 and subsequent lockdowns, averaging out somewhere around 1.0.

"quarantine fatigue" is only a thing with people who don't know a single person that had a severe case.
Once you do know, it doesn't exist for 90+% of that population of people.

Realistically we're fighting over hundreds of thousands or millions of potentially preventable deaths. The idea that immunity will lapse and the disease will remain as bad as ever is catastrophizing and not how things work. Humanity has been through this many times before, society won't collapse even if many millions die.

Nobody knows whether immunity will lapse or not. I find your lack of knwledge disturbing.

There is not a single disease that can reinfect us for which there is reliable data pointing at reinfection being milder in terms of effects.
Humanity has survived epidemics of diseases that are either not deadly or that do not re-infect because immunity is long-term. This one is neither.

Here you go: https://www.cdc.gov/vaccines/vpd/vaccines-list.html. Vaccines only work for diseases against which the body can develop immunity, hence every vaccine that we have corresponds to a disease that provides for long-term immunity.

Do you even understand what "long-term" means?
Do you have the necessary number of brain cells to make a distinction between "civilization can survive a disease for which it has a vaccine that needs a booster shot every 2 years" and "civilization cannot survive a disease that freely reinfects every 2 years"?

If there will be a vaccine in 18 months, we should restrict Rt<1 to avoid unnecessary deaths and long-term health issues.
If there will never be a vaccine, we need to restrict Rt < 1 because the only alternative is full-scale societal collapse.

In game theoretic terms, reducing Rt <1 is a dominant strategy.
Your herd immunity nonsense is junk.

I'm giving you a long list of diseases for which humans can develop long-term immunity. E.g. once you get chickenpox, you are considered immune for life, although you might get shingles later on if you're unlucky.

Now, can you name at least one virus that:

1. The human body can clear out on its own (so, HIV or HPV are out, as almost no one can ever recover fully from then)
2. Doesn't mutate too quickly (so, various flus are excluded)
3. Only provides short (less than 5 years) effective immunity
4. Inflicts serious damage on the first infection (so, existing human coronaviruses are out)

If no such viruses are known to mankind, why presume that SARS-COV-2 will be any different?

Because we know it will not induce long-term immunity (as every single other coronavirus that has infected humans ever) and because it is already known that it "inflicts serious damage on the first infection".

Arguments from induction are invalid in principle (just a useful rule of thumb) yet you nevertheless cannot even formulate a coherent one.

So, to summarize:
a) coronavirus will be able to reinfect you in 18-24 months
b) you have zero evidence that the re-infection will be milder the second time

Herd immunity approaches are doomed. We either control it or society completely collapses.

Far right whites look forward to society completely collapsings ince only the superior race will emerge triumphant.

Becaue only the far white racists will emerge triumphant.

Herd immunity is not guaranteed to exist without a vaccine. A vaccine is also not guaranteed to exist either.

There's no herd immunity to SARS either, but SARS is not with us anymore (except in a lab in Wuhan, waiting to be reborn). Ebola was the same way though they did get a vaccine for one strain of it.

Bonus trivia: How many of you think SARS-CoV-2 was designed in a BSL-4 lab by a research director named Shi, aka Batwoman? Not many of you I'm afraid Yet online you all sound so authoritative, some of you even opining on the GCMs used in global warming to deny AGW exists, wow such geniuses!

SARS had the advantage that, by the time you were infected and contagious to other people, you knew your were sick. That made it possible to identify people and isolate them quickly.

Covid-19 is much more insidious because it has an average incubation period of 5 days (but a max of 14 days). It is perhaps more comparable to polio, which we never got rid of through "herd immunity" and which plagued human civilization for thousands of years until a vaccine was available.

@Ricardo - thanks for the reply and the reference to polio, which led me to this site: https://en.wikipedia.org/wiki/History_of_polio#Early_history

However, your characterization of SARS is inapt a bit, I recall superspreaders in SARS as well. And recall Batwoman (aka Dr. Shi) invented a chimeric virus in 2015 called "SARS-CoV", which is 96% the same as todays SARS-CoV-2, aka the Covid-19 virus. So I would say SARS is a close cousin of Covid-19 (albeit SARS had a 10% death rate I believe, while SARS-CoV-2 is somewhere between 2% to 5%).

Surveillance capitalism is the end game of Thiel-backed Trumpian nationalism. Google, Apple, Facebook, and Amazon pretend to hate Trump but they are growing like gangbusters financially and politically under his administration. The shining example is the trade war with China with exemptions just for tech. Ignore the fake political theater. Watch the hidden hand and backroom deals.

But that didn't matter, because surveillance capitalism will bring a safe, and secure, society.

Germany hasn't managed to do it yet either and has wasted time deciding whether to have centralized storage of information and ultimately adopted the approach favored by Apple and Google.

So even Germany is writing a Liebesbrief to Big American Business.

The French, predictably, continue to fart in the general direction of American big business, whether Amazon (who closed their French operations in a snit after being told they had to protect worker safety) or Apple/Google.

The Australian (COVIDsafe) app only records an interaction when phones are within 1.5 metres (5 feet) for 15 minutes.

The issues with the app so far is that you need Bluetooth on all the time which drains my battery. It has also interfered with other Bluetooth using apps like diabetes monitoring software - though hopefully they can sort out those bugs.

To some of the above: The right App will save us!

Gimme a break.

Ubiquitous smartphone surveillance is the absolute wrong way to proceed. We should be monitoring chokepoints. This is the most bang for the buck. Airports, harbors, borders, etc, places where man, animal, and plant transit. Everything else is way too disruptive and raises all kinds of constitutional issues.

A virus with an IFR as low as 0.2% is making many in the world consider radical measures. This is impressive hysteria.

IFR is at least 0.8% with the current US population structure and it's not as if the deaths are the only consequence of the virus that highlights the need for those radical measures.

The most cogent comment is from Bruce Schneier in the link:

"To me, it's just techies doing techie things because they don't know what else to do."

I love the things computers allow me to do, but I increasingly see the love of tech consistently overriding critical thinking. The danger is that the love of tech leaves us with no more than the "collateral damage" of a ton of useless, shiny apps.

Besides, has no one discovered the only good part of the current moment -- having the mobile turned off?!

I think the primary purpose of track and trace is that its onerousness helps the compulsory-vaccination business model.

Not to brag but when the Google app came out on tracing I was ahead of the curve when I suggested a simple voluntary tracing based on public spirit and voluntary exchange of information as to who are the people you most interact with, as found in South Korea would be more effect. At that time I was told the random strangers at the supermarket would be excluded, but I had correctly intuited that these people by and large don't count (and anyway if they show symptoms they too will be quarantined eventually).

So, not to brag, I was right after all.

Does anyone have any experience with New York City? 1/3 of cases are in tri state commuting area where 1) social distancing while commuting and using trains, busses, escalators, stairs, elevators, tiny coffee shops, offices, etc. is not possible
And, 2) 90% of New York City residents would not be able to understand distancing (aside from language barriers) due to living their lives in a petri dish called NYC with its own culture.
(You cab also predict that 1 in xxx New Yorkers will steal your device.)

Contact tracing without isolation provisions/centers, will not work. Self isolation, with few caveats having a basement for example, doesn't prevent the family getting infected. Asking people to display a greater duty of care for strangers, than for their family, is difficult.

Body cameras are the answer. Our memories lie to us. And so do location apps: they don't provide context. Body cameras provide both location and context. A picture is worth a thousand words. Law enforcement personnel wear body cameras and our soldiers often wear body cameras. We know Pence didn't wear a mask at Mayo because he did wear a body camera (the photographers that follow people like Pence are in essence a substitute for a body camera). We know where you've been and what you were doing there. We'd know where Joe Biden and Brett Kavanaugh had been and what they were doing there if each had worn a body camera. Surveillance cameras are all around us, so what's the objection to wearing them. And we'd know who the body camera holdouts are, just as we know that Pence was a mask holdout - and because we know it, Pence is no longer a mask holdout. And we know that Pence is a scoundrel: he had his wife lie for him and we know it thanks to Pence's body camera. Body cameras are expression of one's citizenship (i.e., civic consciousness). Wear one with pride. What do you have to hide?

They write speculative fiction. I doubt they can know whether the Apps could significantly decrease transmission nor whether the inevitable abuse will be sufficiently malign or pervasive to warrant early government legislation/regulation. Is an App like this required to show efficacy if it is making a medical claim? (FDA) Does HIPPA come into play?

I don't think many of those objections talk about what contact tracing really is. It doesn't measure passing contacts - you need to be more than 10-15-20 minutes in close contact. It doesn't really matter if it's not perfect - no measure is perfect and we're talking about ways to cheaply reduce R.

And half the objections are speculation on how this might me misused. Which, well, it's true, but we're comparing "might be misused" with "very real costs if we don't use it", and I find the argument weak.

Why is Metcalfe's Law relevant?

I cannot see it as applying to the virus as that is often (potentially) more of a broadcast event (eg cough or sneeze) rather than a single person-to-person contact.

I cannot see it as applying to using smart phones to determine proximity (eg using Bluetooth) as each communications connection need only last a few seconds or several fractions of a second. Thus proximity could be determines by a sequence of one-to-one connections covering (eventually) all of the in-range smart phones.

Despite this, I too have serious doubts of the practical utility (absolutely or in competition with manual methods) of contact tracing Apps.

Keep safe and best regards

Well, compared to the status quo where basically everyone’s freedom is restricted for months and months, widespread contact tracing that only temporarily restricts the freedom of people who’ve been in contact with the infected would be preferable even if most of those people never get infected themselves.

The problem with overreaction to the app warning is about two things. First is calibration of the warning - the warning should be scaled proportionally the probability of infection. There are technical means to aproximate the distance (by strength of the bluetooth signal or by additional ultrasound communication), there might be additional rules for the app to deduce if the contact was in closed space (this can be done locally at the phone at the time of contact - it can use GPS - but it does not need to store the input data) and probably lots of other rules. The probabilities all the meetings can also be summed up. The assigned probabilities of different kinds of meetings should also be revised all the time via testing. The warnings should have different levels based on the probability.

The second problem is the calibration of the reaction to the warnings. There should be published some simple advice about what to do after an app warning, and how to combine the warning with possible symptoms. The text can be quickly revised if there are massive overreactions etc. The text should be adjusted to local (probably state level) conditions and should lead people to basically three things: limiting physical interaction (for the lightest warnings) and waiting for symptoms, complete self-isolation (for the severe warnings) plus testing to be sure depending on the local availability.

The app data should also be used by the testing facilities to select the most probable cases.

Overall the app data is just additional information. Sure people can react wrongly to that information - but this is not an insurmountable problem it is a matter of education and of people naturally adjusting out of their own interest.

And beside that - that automatic contact tracing should only be viewed as a mean to increase efficiency of the contact tracing investigators not as something to replace them.

You need the trifecta of tracing, testing and isolation (ideally, centralized isolation) for this to make sense.

Tracing and testing without centralized isolation means people will continue to spread the virus among family members or other people they interact with if they break quarantine. Tracing and isolation without tracing is miserable because it means anyone who came into contact with an infected person is isolated for two weeks, even if they never got infected. And testing and isolation apparently requires millions of tests to be performed every day to be effective, which will never happen.

“I wonder if there also is a negative economic effect, over the longer haul, simply by making fear of the virus more focal in people’s minds.”

Great question, similar to one that Heather MacDonald has also asked and thinks she knows the answer to (spoiler: yes, there will be).

https://www.google.com/amp/s/spectator.us/paranoid-style-covid-19-central-park-safetyism/amp/

Comments for this post are closed