Seems like you’re looking for anecdotes rather than data which strikes me as strange for a purported rationalist. The plural of anecdote may be data, but the quality of those data will be low. If it helps, I had one dose of AZ and had minimal side effects (slight headache, tiredness) that were gone in under 24 hours.
Seems like you’re looking for anecdotes rather than data which strikes me as strange for a purported rationalist.
There’s a hypothesis in my post and I’m running a cheap experiment to disprove it. Absences of anecdotes would falsify the hypothesis.
If the hypothesis isn’t falsified that would justify running more expensive experiments. Having detailed accounts of side effects in turn allows setting up expensive experiments in a more targeted way.
“Anecodates are not evidence” is a slogan that’s quite fundamentally opposed to what this community that founded on ideas like everything is Bayesian evidence is about.
If it helps,
It doesn’t help for the experiment that I care about, but I’m not going to do anything to supress any data reporting.
At the least, I would have expected this to be a poll, so as to get a denominator.
5 data points: my household and two coworkers suffered from (at most) temporary sore arms (4 Pfizer, 1 Moderna). Oh, and a mild headache under 24 hours. Edit:670,000 more datapoints.
Update Aug.1: after the same 5 people got a second dose, one coworker got strong Covid-like side effects for about 24 hours, starting 12 hours after vaccine administration; he basically took the day off in consequence. I’m unsure which vaccine he received. Among the other 4, the worst effect was short-lived, mysterious, temporary pain in various locations (24 hours or less).
At the least, I would have expected this to be a poll, so as to get a denominator.
A denominator doesn’t help with falsifing the thesis. If we do get a bunch of reports of significant side effects and a bunch of reports of no side effects, the conclusion wouldn’t be that the thesis is correct but that it deserves more expensive attempts at verification.
I don’t see anything significant to be gained by having a dominator.
My thought process was:
I could run a mechancial turk experiment to get a representative sample and get representative numbers from that.
That would cost money and I haven’t done it before, so is there an experiment that would falsify the thesis that’s cheaper to run?
If the situations looks like what Kirsch is describing, we there are likely rationalist with signficant side effects. This not being the case, would suggest that the thesis of Kirsch is wrong.
Let’s write question to get to know whether there are rationalists with significant side-effects.
If there are I might even get someone else to run the mechanical turk experiment ;)
Point taken. Even so, I’m concerned about the reputational and public-safety consequences of not simultaneously gathering “no side-effect” reports (conditional on the vaccines being safe).
Even so, I’m concerned about the reputational and public-safety consequences of not simultaneously gathering “no side-effect” reports
If you look at this thread there are plenty of people who do post “no side-effect” reports. When people send me those reports privately, I do post them here even when it’s not the information that’s valuable to me.
Doing anything that would effectively stop those reports from happening would be bad.
The question is whether to do this as a poll or not and not stopping the gathering of “no side-effect” reports.
Even so, I’m concerned about the reputational and public-safety consequences
Optimizing for reputation instead focusing on truth seeking is dangerous. This kind of reasoning killed a lot of people in the last year.
It’s not easy to choose truth seeking but in the end that’s what the rationalist project is about. It’s not like I don’t have impulses in the other direction as well, but then I go back and optimize for truth seeking even when it’s an effort and comes with risk.
To get back to the poll question, there are two choices:
Have an anonymous poll. This opens us up to outsiders coming and creating a lot of false votes.
Have a named poll, this discourages people who might say something inconvient from participating.
Irrelevant nit: the archaic second-person singular of “do” is “dost”, as in “dost thou not know”. “Doth” is the third-person form, as in “the lady doth protest too much”.
Seems like you’re looking for anecdotes rather than data which strikes me as strange for a purported rationalist. The plural of anecdote may be data, but the quality of those data will be low. If it helps, I had one dose of AZ and had minimal side effects (slight headache, tiredness) that were gone in under 24 hours.
There’s a hypothesis in my post and I’m running a cheap experiment to disprove it. Absences of anecdotes would falsify the hypothesis.
If the hypothesis isn’t falsified that would justify running more expensive experiments. Having detailed accounts of side effects in turn allows setting up expensive experiments in a more targeted way.
“Anecodates are not evidence” is a slogan that’s quite fundamentally opposed to what this community that founded on ideas like everything is Bayesian evidence is about.
It doesn’t help for the experiment that I care about, but I’m not going to do anything to supress any data reporting.
At the least, I would have expected this to be a poll, so as to get a denominator.
5 data points: my household and two coworkers suffered from (at most) temporary sore arms (4 Pfizer, 1 Moderna). Oh, and a mild headache under 24 hours. Edit: 670,000 more datapoints.
Update Aug.1: after the same 5 people got a second dose, one coworker got strong Covid-like side effects for about 24 hours, starting 12 hours after vaccine administration; he basically took the day off in consequence. I’m unsure which vaccine he received. Among the other 4, the worst effect was short-lived, mysterious, temporary pain in various locations (24 hours or less).
A denominator doesn’t help with falsifing the thesis. If we do get a bunch of reports of significant side effects and a bunch of reports of no side effects, the conclusion wouldn’t be that the thesis is correct but that it deserves more expensive attempts at verification.
I don’t see anything significant to be gained by having a dominator.
My thought process was:
I could run a mechancial turk experiment to get a representative sample and get representative numbers from that.
That would cost money and I haven’t done it before, so is there an experiment that would falsify the thesis that’s cheaper to run?
If the situations looks like what Kirsch is describing, we there are likely rationalist with signficant side effects. This not being the case, would suggest that the thesis of Kirsch is wrong.
Let’s write question to get to know whether there are rationalists with significant side-effects.
If there are I might even get someone else to run the mechanical turk experiment ;)
Point taken. Even so, I’m concerned about the reputational and public-safety consequences of not simultaneously gathering “no side-effect” reports (conditional on the vaccines being safe).
If you look at this thread there are plenty of people who do post “no side-effect” reports. When people send me those reports privately, I do post them here even when it’s not the information that’s valuable to me.
Doing anything that would effectively stop those reports from happening would be bad.
The question is whether to do this as a poll or not and not stopping the gathering of “no side-effect” reports.
Optimizing for reputation instead focusing on truth seeking is dangerous. This kind of reasoning killed a lot of people in the last year.
It’s not easy to choose truth seeking but in the end that’s what the rationalist project is about. It’s not like I don’t have impulses in the other direction as well, but then I go back and optimize for truth seeking even when it’s an effort and comes with risk.
To get back to the poll question, there are two choices:
Have an anonymous poll. This opens us up to outsiders coming and creating a lot of false votes.
Have a named poll, this discourages people who might say something inconvient from participating.
Excuse me? I’m not suggesting there’s anything wrong with measuring the numerator. Methinks thou doth protest too much.
Irrelevant nit: the archaic second-person singular of “do” is “dost”, as in “dost thou not know”. “Doth” is the third-person form, as in “the lady doth protest too much”.
It’s a higher Simulacra level. Zvi wrote a lot about how COVID-19 decision making at higher simulacra levels lead to a lot of harm.
And in this particular case? How does shunning the denominator increase safety here?