12-21-2020, 11:22 AM
I dug into this over the weekend with some friends and family who work at Stanford healthcare, and came away with the impression that this was really more about incompetent administration, not abuse of privilege (although for the more cynical, you can't completely rule out abuse hidden behind incompetence).
As pointed out earlier in the thread, you can't really blame "the algorithm," which is just a set of rules that spits out an ordered list of 3900 (or however many total healthcare workers there are) people to be vaccinated, because it would be impractical for humans to do that manually. However, humans decide the prioritization rules, and humans should be auditing the results to sanity check them. Apparently this second part was not done, or at least not done in time (the articles suggest that the list was generated on Tuesday, and vaccinations started on Friday - this is why they were able to make a quick change after the blowup, because most of the vaccines were planned to be administered over the weekend.)
Stanford also has not provided any specific details of what prioritization decisions they programmed into the algorithm, outside of loosely referencing the CDC guidelines. This seems like a key mistake, because there isn't one right way to prioritize, but if you don't show the details, people will assume the worst. For example, a lot of the coverage seemed to imply that administrators were taking advantage of their position to place themselves higher on the list. From talking to people, this does not seem to have happened - people in purely administrative functions (like the hospital CEO) did not get prioritized (although interestingly, one nurse I talked to said she would be more comfortable if the CEO had gotten the vaccine to show it was safe). Rather, the reason that senior medical staff got priority was because they tend to be significantly older and therefore more at risk. I would also speculate that the employment records might not make it easy to distinguish medical staff who have high degrees of patient contact vs those who don't, although seems like that would be easy to remedy by asking people to self-declare their degree of patient contact in the vaccination survey they sent.
Finally, I heard several anecdotal accounts of vaccination ordering that made no sense to me. For example, on the same unit, a 24-year-old nurse (no known comorbidities) being selected ahead of a 34-year-old. I guess it is possible that they just bucketed everyone under 35 together as low risk and just randomly selected, but again there was no transparency.
As pointed out earlier in the thread, you can't really blame "the algorithm," which is just a set of rules that spits out an ordered list of 3900 (or however many total healthcare workers there are) people to be vaccinated, because it would be impractical for humans to do that manually. However, humans decide the prioritization rules, and humans should be auditing the results to sanity check them. Apparently this second part was not done, or at least not done in time (the articles suggest that the list was generated on Tuesday, and vaccinations started on Friday - this is why they were able to make a quick change after the blowup, because most of the vaccines were planned to be administered over the weekend.)
Stanford also has not provided any specific details of what prioritization decisions they programmed into the algorithm, outside of loosely referencing the CDC guidelines. This seems like a key mistake, because there isn't one right way to prioritize, but if you don't show the details, people will assume the worst. For example, a lot of the coverage seemed to imply that administrators were taking advantage of their position to place themselves higher on the list. From talking to people, this does not seem to have happened - people in purely administrative functions (like the hospital CEO) did not get prioritized (although interestingly, one nurse I talked to said she would be more comfortable if the CEO had gotten the vaccine to show it was safe). Rather, the reason that senior medical staff got priority was because they tend to be significantly older and therefore more at risk. I would also speculate that the employment records might not make it easy to distinguish medical staff who have high degrees of patient contact vs those who don't, although seems like that would be easy to remedy by asking people to self-declare their degree of patient contact in the vaccination survey they sent.
Finally, I heard several anecdotal accounts of vaccination ordering that made no sense to me. For example, on the same unit, a 24-year-old nurse (no known comorbidities) being selected ahead of a 34-year-old. I guess it is possible that they just bucketed everyone under 35 together as low risk and just randomly selected, but again there was no transparency.