About Me: Background and Biases 2021-04-04T14:19:31.220Z


Comment by tkpwaeub (gabriel-holmes) on Covid 4/15: Are We Seriously Doing This Again · 2021-04-21T00:55:16.171Z · LW · GW

I could also add #3 - non-pharmaceutical interventions are still perfectly legit.

Comment by tkpwaeub (gabriel-holmes) on Covid 4/15: Are We Seriously Doing This Again · 2021-04-21T00:53:12.836Z · LW · GW

Fair enough. I believe that it's been noted that in some places (Australia) the stronger version of (2) really does hold. I'm not sure if (1) still holds, but it's conceivable that it could. Either way, you're comparing two very small numbers and my meta-point is that it's still something that bears analyzing. Thanks for keeping me on my toes.

Comment by tkpwaeub (gabriel-holmes) on Covid 4/15: Are We Seriously Doing This Again · 2021-04-20T09:47:59.750Z · LW · GW

Yup, especially if you keep up the non-pharmaceutical interventions.

Comment by tkpwaeub (gabriel-holmes) on Covid 4/15: Are We Seriously Doing This Again · 2021-04-20T09:41:55.218Z · LW · GW

Well, if we are involving Congress there are even better remedies than litigation.

Comment by tkpwaeub (gabriel-holmes) on Covid 4/15: Are We Seriously Doing This Again · 2021-04-20T09:40:07.384Z · LW · GW

I agree that it was probably a mistake to pause for a whole week. However, they were correct to look into the matter. May I reframe this a bit? (I still plan on writing my own, more comprehensive piece on all this - your posts are helping me sort this out).

Deep breath - here goes.

  1. We are asking people to get vaccinated for the common good - because there's an emerging consensus that all the vaccines help mitigate transmission.

  2. It really does appear that for some vaccines (the ones using adenovirus vectors), there's an increased risk of a very rare type of blood clot for some people (women 18 to 55). The personal risk of death from Covid for that group is lower than it is for the general population.

In other words, we're asking some people to put their lives on the line for the common good. Yes, the absolute numbers are small, but that shouldn't be a distraction. Women between 18 and 55 need to be offered mRNA vaccines first. Everybody else should consider it their civic duty to get an adenovirus vector vaccine. And women between 18 and 55 who get one of the adenovirus vector vaccines need to be fully informed and at the very least sincerely thanked - possibly with money.

Comment by tkpwaeub (gabriel-holmes) on By which mechanism does immunity favor new Covid variants? · 2021-04-15T01:30:31.651Z · LW · GW

This reminds me - have you read about some of the connections between covid outcomes and cytomegalovirus? It's pretty fascinating. 

Comment by tkpwaeub (gabriel-holmes) on Covid 4/9: Another Vaccine Passport Objection · 2021-04-13T08:42:11.719Z · LW · GW

Yup. High stakes free rider problem, in which we are comparing individual risk to collective benefit. And the moral calculus is a moving target.

Comment by tkpwaeub (gabriel-holmes) on Covid 4/9: Another Vaccine Passport Objection · 2021-04-13T08:38:50.817Z · LW · GW

That's why I got a vasectomy

Comment by tkpwaeub (gabriel-holmes) on Covid 4/9: Another Vaccine Passport Objection · 2021-04-12T14:43:09.522Z · LW · GW
  1. Enforcement of vaccine passports (or whatever - the Excelsior pass includes the possibility of recent negative test results, but it does display which "bucket" someone falls in) would be a whole lot easier if they took your photo when you got your vaccine, and then whoever's scanning it could pull it up automatically from a database.  Adding photo ID's to electronic health records is a really good idea anyway.  So is making it easier for people to share pieces of their EHR. There's also a strong case to be made for including GPS to avoid duplication and forgery (e.g. randomly "polling" passport holders and having them report their locations). Another thing that would probably help would be to keep track of statistics. E.g., if it appears that a mathematically impossible percentage of vaccinated individuals are out on the town, that could trigger a second look at specific sites.

    2.  Verification of vaccine status should occur off-site, when you buy your "ticket".  At that point, a second picture should be taken during that transaction. This step should be done by a disinterested third party (e.g. a licensed security guard, a private investigator, a licensed ticket reseller, a non-profit entity, etc). The "bouncer" at the door can then verify that you belong with the ticket - saving you the trouble of having to carry the passport around with you.  Walk-ins can be offered a rapid self-test, of the sort that's becoming more widely available (should be easy enough if there's a bathroom on-site). 

    3. Not to beat a dead horse, but the Biden administration is absolutely correct in insisting that any such vaccine certification solutions be open source. The relevant API's can then be shared with private sector apps (use your imagination) and other government agencies. Verification of immune status should be something that's done continuously and discreetly in a way that's indistinguishable from magic. 

    4. As silly as this might sound, it would also make sense for the "passports" to have a "self attestation" option if people want to indicate that they don't have symptoms, haven't traveled out of the country, etc. It would incur heavy penalties if you blow off a contact tracer (for example).  This would also be great for contact tracing, scheduling vaccination and testing appointments, contact tracing, equity, future-proofing, and quickly levying penalties for violations od public health mandates (for the win!)

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-08T20:35:52.257Z · LW · GW

If you're not concerned about enforcement why bother with the security theater? Might as well just trust people when they say they're vaccinated. The marginal benefits of the apps is negligible, compared to the cards. Anyone with the chutzpah and resources to forge a CDC card isn't going to have a hard time forging a QR code and a driver's license.

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-08T11:20:48.053Z · LW · GW

To clarify - the most humane, least risky use case that would satisfy these desiderata would be have a modular system where we

  1. Verify off-site that DanArmak is vaccinated (or has allergies, or a heart condition, or is immunocompromised)

  2. Verify on-site that you're DanArmak

  3. Give the user the option of sharing information with the state or local health department so they can contact you or your physician about virus exposure (not just covid), food poisoning, whatever.

This could be widely adapted to a variety of situations, depending on how rigorous one wants to be about verifying someone's ID.

If we DO stick with apps, the best approach might be to give everyone a QR code (including those who haven't gotten vaccinated or tested). Separately, provide multiple options for verifying the codes (strict verification of ID, census of how many people are vaccinated once capacity is exceeded, etc)

Looks like some people may already be moving in this direction.

Again, easier to implement if the software is open source.

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-06T17:01:04.120Z · LW · GW

Interestingly in New York State it appears aren't allowed to store anything about the verification per GBL 899-aa and 899-bb. That's about as close to a "no warranty" statement as it gets.

Comment by tkpwaeub (gabriel-holmes) on What are the best resources to point people who are skeptical of getting vaccinated for COVID-19 to? · 2021-04-04T23:10:06.640Z · LW · GW

Same here. Fretting about how to convince, threaten, or cajole people into getting vaccinated is a solution in search of a problem, for now.

Comment by tkpwaeub (gabriel-holmes) on About Me: Background and Biases · 2021-04-04T14:22:12.374Z · LW · GW

I also believe that you should take civil service exams, for your own edification and to help ensure that government workers like me are at least as qualified as you. Think of it as a civic duty, on par with voting, paying taxes, and serving on a jury.  

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-04T13:41:57.758Z · LW · GW

Privacy concerns could be addressed by periodically deleting data more than ten days old. 

Comment by tkpwaeub (gabriel-holmes) on In plain English - in what ways are Bayes' Rule and Popperian falsificationism conflicting epistemologies? · 2021-04-04T13:28:13.155Z · LW · GW

I'm not sure Bayes' Rule dictates anything beyond its plain mathematical content, which isn't terribly controversial:


When people speak of Bayesian inference, they are talking about a mode of reasoning that uses Bayes' Rule a lot, but it's mainly motivated by a different "ontology" of probability. 

As to whether Bayesian inference and Popperian falsificationism are in conflict - I'd imagine that depends very much on the subject of investigation (does it involve a need to make immediate decisions based on limited information?) and the temperaments of the human beings trying to reach a consensus. 

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-04T00:51:40.458Z · LW · GW

I appreciate the explanation of the downvote (no harm no foul) and I'll try to tweak it if I get a chance. I probably do have experiential information that's hard to unpack (without starting to break confidentiality, based on my line of work - which is annoying, because I absolutely never wanted to play the "I could tell you, but then I'd have to wipe your memory tomorrow" card, but here we are). 

I do think the potential for escalation of conflict is a real concern; and that's another reason for keeping the implementation as discreet as possible. For a restaurant, I could imagine a combined vaccine status/capacity logging/contact tracing app that has the same look and feel to everyone involved as a handy way of making reservations in 30 minute increments once the capacity reaches a certain level. This would involve giving everyone a QR code, and I believe this would probably be easier to enforce because it's a lot easier to catch "two people being in the same place at the same time" 

My main "meta-points" in all of this are:

a) Keeping the code open source, like Jeffrey Zients suggests, is incredibly important.
b) We should try to come up with a set of guidelines of what it would mean for vaccine passports to be a failure (e.g., no measurable effect on case rates, evidence of rampant forgery, etc)

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-03T21:50:57.905Z · LW · GW

So I guess another approach that would make vaccine passports palatable to everyone would be if we just went ahead and gave everybody something similar to the Excelsior pass (either on their phone or printed), where we'd attempt to implement that underlying logic of "x% vaccinated or y% capacity" in real time. The venue would stop allowing new people in when neither of those conditions are met, and nobody would see anybody's personal information. And it would take away any motive for the prospective attendee to cheat. This could also be leveraged for contact tracing, and perhaps the expectation would be that you get a test if you don't feel well, or if a contact tracer tracks you down. 

I don't think it would be too difficult to deploy something like this (all-in-one contact tracing/vaccination tracking/compliance app)

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-03T15:47:13.677Z · LW · GW

Yeah, so I guess my point is that in the spirit of "less wrong", making a beeline for aggregate statistics appears to me to be the "least wrong".

There's also somewhat promising evidence that there's going to be enough self selection that Bayes' Theorem will have our backs even without incentives. Kinda like how there are a lot of uh, people like me in movie theaters on December 25th.

Comment by tkpwaeub (gabriel-holmes) on Bureaucracy is a world of magic · 2021-04-03T14:10:18.341Z · LW · GW

I haven't read every other comment, but has anyone else brought up Clarke's Third Law?

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-03T09:44:29.849Z · LW · GW

Especially when there's a case to be made for quarantines as a way to stop the next pandemic.

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-03T08:14:40.037Z · LW · GW

I feel like there's some sort of yet-to-be-articulated "impossibility theorem" here. Some sort of mash-up of the project management trilemma and Shannon's theorem

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-03T08:07:07.143Z · LW · GW

Even if we made fakes easy to spot, and the gatekeepers were able to spot them and report them every time - it's worth considering that the first link in the chain of events leading to the death of George Floyd involved somebody spotting and reporting a forgery. Is this REALLY the path we want to go down?

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-03T08:01:00.202Z · LW · GW

Thanks. That means a lot to me. I feel like a lot of things depend on "who's in the room" when decisions are made, and all too often it's the people who are stuck with implementing things that are left out.

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-02T18:31:26.618Z · LW · GW

It turns out that non-profits in New York State don't need to obtain licenses to resell tickets. So it would be pretty easy for an event organizer in New York City to, say, sell a bunch of tickets in bulk to a large hospital system that uses MyChart. The hospital system can then resell the tickets to its own patients, as a promotion - possibly with a corporate sponsorship. 

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-02T18:28:22.735Z · LW · GW

Right, so whatever direction we go with this, it's really important that the application be open source, as Jeffrey Zients has suggested, so that businesses can add on security features that identify the person. 

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-02T18:17:19.715Z · LW · GW

I've been thinking about this too. Say we let 

 expected net personal benefit of getting vaccinated
 expected net collective benefit of getting vaccinated.

It's possible to imagine a scenario where the former is negative and the latter is positive. That would leave us with a genuine "free rider" problem, and a rather high-stakes one at that. The most prosaic solution to such problems is to pay people to do the thing that they wouldn't otherwise do rationally. 

I don't think there's anything terribly objectionable about the government incentivizing people to take on personal risks they wouldn't otherwise take. After all, there are lots of government jobs which are inherently dangerous - and the government "incentivizes" them by paying people decent sums of money. 

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-02T15:52:25.959Z · LW · GW

Sorry, I seem to have reached my character limit, so I'm continuing this in a reply to myself - I think this is important. 

There's no reason to think that incentives wouldn't serve to maximize . Bayes' Theorem relates  and  as follows:

The beauty of an incentive system is it would maximize both factors in the numerator while the denominator can stay more or less fixed (e.g. by the capacity of the venue). Same reduction in risk for people attending, at minimal cost. 

Again, I want to emphasize that I can imagine a role for digital vaccine certificates in such a system, but the idea of vaccine passports presented at the door, just seems untenable. 

Comment by tkpwaeub (gabriel-holmes) on Covid 4/1: Vaccine Passports · 2021-04-02T15:40:21.047Z · LW · GW

Thanks for bringing up the topic of vaccine passports and doing it in a thoughtful way. Before getting started on the stuff I can actually claim to have some insight into, just a few quick points of order about the Excelsior Pass:

  1. It's not really a vaccine passport, or even an immunity passport, it's a broad "good to go" credential based on whether you've been vaccinated or you've had a recent negative test.
  2. As best I can tell the 30 day limit is just an artifact of when you download it. You can reset the clock by downloading it again the next day, and there's no limit on how many times you download it.

Also an important point about the extent to which there might be something rolled out at the federal level: the Biden administration has already indicated that except possibly for international travel, they don't intend to mandate anything of the sort for domestic use. Exact language matters  a lot here. The feds are using the term "digital vaccine certificate", rather than "vaccine passport", and I think the connotation is important. The former suggests something more agnostic as far as use-case. The latter suggests something much more rigid, and something that's only really appropriate for travel (as opposed to something you'd have to carry around with you anywhere you go). According to Jeffrey Zients, “any solutions in this area should be simple, free, open source, accessible to people both digitally and on paper, and designed from the start to protect people’s privacy.” I boldfaced "open source" because I really think that's key here. People, and by extension governments composed of people, make mistakes. An idea can be good-ish, and people can put in work developing an app, but as long as we keep this open source we can salvage any work that's been done without throwing the baby out with the bathwater. Not surprisingly, this is a position shared by the American Civil Liberties Union

Now to the aspect that I can actually claim insight into - "fraud concerns". They're valid. Before delving into why they're valid, I want to pose the question: what's the "end goal" of vaccine passports? Is it to 

a) Ensure with 100% accuracy who is or isn't vaccinated? or

b) Maximize the percentage of vaccinated people at a given public venue?

What I want to argue for is that (a) is a fool's errand and that there are better, more cost-effective ways of getting (b). In the process I'm also going to show why "open source" is a key characteristic of any digital vaccine certificate. 

About me: I've been a civil servant working in state government for almost a decade. In that time I've been exposed to a lot of different things that state government does to enforce a lot of different things that could be viewed as rough analogs to what's being suggested with vaccine passports. The recurring theme is that fraud and forgery are common, and become more so when people have a high incentive to commit fraud and forgery for personal financial gain (or avoiding personal financial losses). Indeed, in a managerial accounting class one might learn about the "fraud triangle":

fraud triangle

It's not hard to see that all three of the above elements exist for "vaccine passports", were we to focus on that as our main tool for accomplishing the basic goal of (b) above. Here are some of the examples where fraud is surprisingly common (warning: a lot of the links below are to New York Times articles, and you may vary well blow through your monthly quota if you don't have a subscription - high quality information isn't cheap).

1. Fish About 40% of all fish is mislabeled. The problem periodically resurfaces when it's examined by investigative journalists or law enforcement, and there's very little evidence that it's going away, despite our best efforts to enforce it. One of the more disturbing forms of this is when escolar is passed off as tuna. Escolar is considered somewhat of a delicacy, but when consumed in the same quantities as tuna, it causes explosive diarrhea

2. Taxi meters - unscrupulous taxi drivers can and do tamper with their meters. A noteworthy instance of widespread fraud occurred in 2010. Cyrus Vance uncovered it in a massive sting operation that presumably involved a lot of overtime. 

3. Gasoline: This doesn't happen as much as the above forms of fraud, but only because we have a government agency devoted to its prevention. But suffice to say that without a large number of government agents acting as secret shoppers, this kind of thing would happen:

4. Supermarket scales: another thing that state governments need to keep an eye on. Because people cheat.

5. Firewood: Moving firewood is a pretty big no-no everywhere in the country, and for good reason - it helps move pests from point A to point B, potentially leading to devastating blights on local flora. All levels of government are involved in enforcing this, both with secret shoppers and with periodic checkpoints. 

6. Fraudulent motor vehicle inspections:  motor vehicle inspection shops have a fairly high motivation to pass paying customers on their inspections. Occasionally this is unearthed by undercover operations

7. Underage sale of alcohol and tobacco: One of the exceptions to child labor laws is the use of minors to pose as underage customers trying to buy tobacco and alcohol. 

There are loads more examples - the key takeaway is that people can and do cheat. Maybe not a lot of people, but enough people do it enough that enforcing it is (pardon my language) hard as fuck. It's a game of whack-a-mole with a limited number of resources, and it involves sting operations and secret shoppers. And it's difficult to see how simply putting something on a smartphone changes the fundamental dynamic of one human checking another human's credentials. Even when merchants do catch attempts at forgery and act in good faith, things can go sideways, as is abundantly clear from the unfolding, increasingly agonizing story about George Floyd.  The fact that the fake was easy to spot wasn't the issue; the escalation was the problem. Anyone tasked with enforcing this needs to be carefully trained on de-escalation tactics, and in a country as awash in guns as ours, that's a pretty big ask. 

So that's where I'm coming from with my skepticism about the feasibility about any large-scale deployment of "vaccine passports". It seems to me that the game isn't worth the candle, for this particular method of obtaining the goal of (b). 

So before I get into two alternatives that I thought of - I think we can all agree that we should be hoping that we're able to convince everyone to get vaccinated quickly and efficiently enough for their own sake that the need for all of this is obviated. It's good to see that the Biden administration is making a concerted effort to do exactly that

It should be noted that both of the alternatives I'm suggesting could easily incorporate some sort of digital vaccine certificate, just not used in the way that some people have suggested for vaccine passports.

  1. Batch verification of vaccination status: Here's how this could work. Say someone (the "organizer") wants to facilitate a group of people ("attendees") in one place. Instead of verifying the vaccination status of each individual at the door, the organizer turns to a third party to check the vaccination status of the attendees, but with a catch. The third party only reports back the number of vaccinated attendees. There's no reason that a "digital vaccine certificate" couldn't be used to make this easier for the third party (which is why I emphasized the importance of this software being open source). What kind of third party? Either a local health department or a licensed ticket reseller might do the trick. In the latter case, the event organizer could sell the tickets to the licensed ticket reseller, who would check the percentage of vaccinated prospective attendees. If the numbers don't work, the event is canceled and  the reseller absorbs the loss, which is easier for ticket resellers, since that's sort of what they're for anyway. The important point is that there's very little motivation either for fraud or violent escalation with this approach. Moreover, it acts as a better safeguard for personal information. Finally, I suspect this would also be better from a contact tracing perspective. A contact tracer would only have to deal with the ticket reseller.  To reiterate, digital vaccine certificates could very much have a role to play in a system like this; "vaccine passports" not so much.
  2. Incentives: By now everyone has heard about the free doughnuts at Krispy Kreme, and I'd love to know if it supports the math that I'm about to outline below. As far as I can tell, reputable sources confirm that incentives work. Even the studies that claim they don't work still end up demonstrating that they increase the number of people who get vaccinated, even when they don't meet their targets, which suggests that if the financial incentives were larger, even more people would get vaccinated. And they don't all have to be monetary. A baseball stadium could offer discounted season passes to people who get vaccinated - heck, they could send them directly to whoever's doing the vaccinating (quick, CVS/Walgreen's/NYU Langone/whoever is reading this - get a ticket resale license!). What happens next? Well, vaccinated people will be more likely to step out.  And here's where Bayes' Theorem comes into play. Specifically consider the following probabilities: 

     probability of someone attending an event
     probability that someone happens to be vaccinated
     probability that someone is attending an event and is vaccinated
     probability that someone is at an event given that they're vaccinated
     probability that someone is vaccinated given that they're at an event