Could you expand on what you mean by finding other people’s suffering entertaining? Does it extend to all sorts of suffering, including, say, the suffering of people without access to healthcare dying of malaria or diarrhea? Or the experience of a parent watching their child die of diarrhea?
jimv
In this post and the last one, you’ve indicated that you don’t see a point to pausing a vaccine’s trial in the face of an adverse event, and that normal processes should be overridden.
I think this may be a case where contemplating Chesterton’s Fence is appropriate.
I’ve seen several people who I would trust to understand this stuff well lay out justifications for due caution in the vaccine development process.
Here’s one set of reasons from a biostatistician (Natalie Dean):
“Widespread use of an unproven vaccine:
Gives people a false sense of security if efficacy is really low
Diverts resources away from other interventions (fixing testing!)
Makes it harder to evaluate better vaccines
Jeopardizes safety
Erodes trust in the process”
https://mobile.twitter.com/nataliexdean/status/1289927701273378816
And another biostatistician (Thomas Lumley):
″...the political situation has raised a real possibility that the President of the USA would try to make the FDA authorise a vaccine without convincing evidence that it works. I don’t use words like ‘disaster’ lightly, but this would be a disaster, both for COVID prevention and for the long-term credibility of vaccination. We need a vaccine that works, and we need people to know that it works.”
https://www.statschat.org.nz/2020/09/14/not-quite-alarmed/
Lumley again, specifically on the issue of a pause:
“Vaccines aren’t intrinsically safe; they are safe when approved because they get thoroughly investigated first.
The immediate question for the trial sponsor and investigators in this case is whether the adverse event, whatever it is, significantly changes the risk:benefit balance of the vaccine candidate, as explained to the study participants (and to the company’s investors). So far, no-one knows, but AstraZeneca seem to be handling this appropriately, and there’s no obvious reason to expect that to change.”
https://www.statschat.org.nz/2020/09/09/adverse-events-some-terminology/
I have no reason to doubt that these experts share all our desires for this disease to be dealt with as quickly as possible. They’re saying that we should still take appropriate care in developing and testing vaccines. There is at least a risk that reducing normal precautions could increase the duration of the disease causing problems, if we’re left with a less effective vaccine, or if the first vaccine is not safe enough so people lose trust, not just in that one but in all subsequent vaccines.
And although the paused trial has been resumed, that doesn’t provide evidence that it “never had a good reason to pause”. In fact, possibly the reverse. The prompt resumption shows that the processes that the investigators wanted to go through were possible to be conducted quickly, lessening their costliness (in terms of time to successful vaccine).
In the case of the pause, I wasn’t saying that it was time, that day, to give the vaccine to the public. I was saying that they should continue the trial while they investigate the one case, until such time as they actually find a problem linked to the vaccine, so as to not lose time. The downside risk there seems to be almost zero.
A direct downside risk of not pausing a trial when you identify an adverse event that shows your vaccine might have harmed one of your volunteers is that you might harm more of your volunteers.
A knock-on downside risk from that is that fewer people might feel confident volunteering if you’re not taking reasonable precautions with their safety, so you might find it harder, or at least slower, to complete your trial.
I would also suspect that there would be contagion of this distrust, making it harder to run trials of any vaccines, both for COVID and into the future.
...”ethicists” who make their prestige and money from expressing concerns and stopping people from doing things...
Might be worth noting that probably no one is strictly stopping you from taking a vaccine of unknown effectiveness and safety. They are just reducing your probability of doing it down to 0.5. If you feel that the tradeoffs favour taking it, then you can volunteer to be in a trial. The world needs people who are willing to put themselves in that position.
are you adopting an ethical position where it is wrong for one person to put themselves at risk to benefit others?
No. Volunteers in a clinical trial like this are putting themselves at risk to benefit others, and I feel that is a good thing.
I am concerned about people’s ability to give informed consent. It seems like a hard problem in general to give prospective volunteers a sense of how much risk they are being exposed to—one issue being that people might tend to assume ‘well, whatever they’re saying about risks, it must be pretty safe it they wouldn’t be doing it’. I feel that the ethical thing is to be as good at ensuring consent is informed as is practicable.
I tend to feel that if you continued to admit volunteers to a trial when a potential serious event had occurred, this issue of consent might get trickier. If you don’t tell the next volunteers at all, then it feels like you’re withholding information, meaning their consent is inadequately informed, so I’d be pretty concerned about that. If you do tell them, but continue admitting them, the message of ‘yeah, something bad maybe happened and we want you to know about it before you agree’ might be read by volunteers as having an implicit content of ‘but it’s not that bad or we wouldn’t still be injecting you’.
Of course you might end up with the information putting lots of prospective volunteers off. In that case, you might lose people who would have been volunteers after a pause.
Or are you asserting that there is more suffering and death in the world where the trial continues than in the world where it stops?
I think this is closer to my sense. My guess is that in expectation, a world in which we take pauses (as short as possible) to investigate potential problems with emerging vaccines for a disease like COVID 19, is likely to have less suffering in the long run. Vaccines are not inherently safe, nor are they necessarily trusted by the public even if they are safe. Actions that might let an unsafe vaccine slip through the net have the potential to cause vast harms in terms of the reduced uptake of lots of other vaccines. Even the perception that a vaccine was rushed without taking appropriate safety precautions can result in it getting blamed for things that it doesn’t cause (see the story of the 1976 swine flu vaccine https://www.bbc.com/future/article/20200918-the-fiasco-of-the-us-swine-flu-affair-of-1976 ). Actions that show that care was taken to vet and understand any potential harms will help to improve uptake across all the range of vaccines.
My understanding is that there was conscription during WW2, so the comparison is somewhat weakened if we’re not talking about volunteers anyway. But in an armed conflict, I think people have a fair sense that the flying bullets makes it “really rather dangerous”, even if they couldn’t quantify it. My problem with how informed consent can be is around issues like we know there are biases where people have a tendency toward a dichotomy: low risk becomes zero risk in their minds. (To the extent that this applies at the other end of the spectrum, high risk would become near certainty, so further makes the war situation different in nature.)
Can you make an argument that if trials are not suspended a loss of trust in vaccines is more likely than if they are continued?
This seems self evidently true to me. I struggle to envisage a situation where people see caution that they perceive to be excessive and become more reluctant to take vaccines in the future.
Good to have a number for some comparisons. From a quick search, I see the WHO estimates that measles deaths globally fell from 535,600 in 2000 to 124,000 in 2017, but rose to 142,300 in 2018. That seems to show that vaccine uptake can make differences in the tens to hundreds of thousands. Yes, I’m comparing globally to one country, but on the flipside, I’d envisage that negative harms from increased skepticism could aggregate over years-to-decades, and over numerous diseases.
Coronaviruses will acquire some mutations, but I think it’s generally accepted that the expectation in general is for a significantly slower rate of mutation than flu.
See, for example, this long-ish read from early September: https://www.nature.com/articles/d41586-020-02544-6
On the specific comparison to flu, it says:
But sequencing data suggest that coronaviruses change more slowly than most other RNA viruses, probably because of a ‘proofreading’ enzyme that corrects potentially fatal copying mistakes. A typical SARS-CoV-2 virus accumulates only two single-letter mutations per month in its genome — a rate of change about half that of influenza and one-quarter that of HIV, says Emma Hodcroft, a molecular epidemiologist at the University of Basel, Switzerland.
Other genome data have emphasized this stability — more than 90,000 isolates have been sequenced and made public (see www.gisaid.org). Two SARS-CoV-2 viruses collected from anywhere in the world differ by an average of just 10 RNA letters out of 29,903, says Lucy Van Dorp, a computational geneticist at University College London, who is tracking the differences for signs that they confer an evolutionary advantage.
This seems to be a close analogue of something I’ve seen in business communications settings: customer segmentation. In that context, I had the same reaction to it as you’re expressing on an individual interpersonal basis: it seems better to make predictions about the individual directly, rather than binning people into segments and then making predictions about the segments.
If you have a bunch of data about a bunch of (prospective?) customers, can your algorithms perform better (say in terms of identifying people’s preferred means of contact) by predicting for each individual customer, rather than going via some pencil-sketched segment and then declaring that for such-and-such a segment you’re best off communicating with them via email because that’s what that segment as a whole prefers?
In the numbered list, the third item (地 I think) is translated as “earth”. In the item descriptions below, I believe the same item is translated as “geography”. These seem like they are intended to be the same things, so consistent translation seems like it might be an improvement.
Have you thought about treating ‘no confidence’ as a candidate? How would it play out if there were a variant of the approach detailed under ‘Assume Confidence + STAR’ where instead of assuming confidence you have an extra n.c. ‘candidate’ who gets scored the same as the others, and if it wins then the election is rerun?
You mention that planting a seed of Anki usage in 6th grade is a good thing. Do you have any thoughts about how to help a child get started with an Anki habit?
I am interested in a couple of points around this:
What makes this worse is that the region of 0.05% to 1% of my net worth is full of long tails. The wagers I’m skipping could easily repay themselves a thousandfold. If I take a wager like this every day for 2 years and just a single one of them repays itself a thousandfold then I win bigtime.
If I take b = 1000 and f* at 0.0005 I get p ~= 0.0015
If I take b = 1000 and f* at 0.01 I get p ~= 0.011
You draw from this the lesson that you need to bet more. What do you see as a source of potential wagers, coming along daily, where there’s a potential 1000x payout and your probability of a win is between around 0.15% and 1.1%?
Also, I’d urge caution over the “[if] just a single one of them repays itself a thousandfold then I win bigtime” framing.
At the p=1.1% / bet=1% of wealth situation, one win in 730 (daily for 2 years) would take a starting balance of $100,000 down to $718. Now, treating this as binomial distribution, that’ll only happen 0.25% of the time, but 2 wins leaving you with $7,990 will happen 1% of the time and 3 wins, leaving $88,871 will happen 2.8% of the time (assuming I’ve not mangled my calculations).
(Yes, in ~96% of cases you’d be left with $988,451 or greater, typically much greater, in the $millions, $billions or $trillions. But this is a caution against thinking ‘only one of these 1% chances in 2 years has to come off to be winning bigtime’.)
At the other end of the spectrum you mention, i.e. the bets are all at the 0.05% of wealth (p=0.15%) situation, the results are much less dramatic.
wins=0, closing balance = $69,337 (prob = 0.33)
wins=1, closing balance = $104,162 (prob = 0.37)
wins=2, closing balance = $156,478 (prob = 0.20)
wins=3, closing balance = $235,070 (prob = 0.07)
wins=4, closing balance = $353,134 (prob = 0.02)
Maybe ‘Private Eye’ in the UK might have some parallels.
This is not intended as a defence of the quantity of advertising that you’re describing, or even to say that there’s not better ways that these goals might be achieved, but this in particular got me thinking of how it might fall down:
“Advertising is a zero-sum or even negative-sum game.”
I think there are at least 2 things going on here.
One is that it might be helpful to draw a distinction between well-known product categories with minimum differentiation and more innovative products. In the case of laundry powder, the brands are just jostling for relative position, you’re going to be buying some form of laundry powder either way, and it’s just about who gets that money. That seems to fit your zero sum situation. But sometimes there are entirely new products and (at least in theory) those consumers who choose to buy them presumably do so because they think that the value they’ll get from them is greater than or equal to the cost that the supplier is charging. “Ooh, that movie looks like it’ll give me £20 of joy and they’re only charging £10 for the cinema ticket, so I’ll go.” In that case the informational value of an ad for the new product is telling the consumer how they can buy something that’s a net benefit.
The other thing is that even in the undifferentiated product case, the advertising is paying for branding. And (again, in theory) the branding potentially has some value to the consumers. It gives suppliers more to lose. Say that there’s a way to adulterate laundry powder with something that fills the box cheaply but has the unfortunate side effect of dissolving clothes. If you’re ‘Anonymous White Box Laundry Inc.” then you might be tempted to use this adulterant. If you get caught, you’ll just start again next week as ‘Anonymous Beige Box Laundry Inc.’. But if you’re a big brand name laundry powder, you’ll steer clear of that since the loss of future sales is far more costly than the savings from using the adulterant. To the extent that advertising builds brand awareness, it can in theory have value to consumers who can pick suppliers who have more to lose by not living up to a reputation for reasonably high quality products.
One caveat here is that the larger the sample size the less a low p-value implies a large effect size, and I don’t feel like I have a good intuition about how exactly the two are connected.
An alternative to p-value NHST that’s still frequentist is to produce the point estimate and confidence intervals for the effect size of interest. Assuming they are defined the same way (one-side vs two-sided, etc.), they will give the same conclusion if you choose to interpret them as a dichotomous ‘significant or not’ status. (I.e., the 95% CI will exclude the null in the same cases as where the p-value will be below 0.05.) But with the CIs you’re getting that indication of the size of effect that you were craving.
I suspect that the answer to this question may depend on the situation where you are and your behaviour intentions after one and two doses.
I believe that the original clinical trials of the vaccines did not test different dose intervals. The idea that a longer gap might be helpful came, I believe, from other vaccines, where it’s been established that a longer gap sometimes improves long-term protection.
I think a reasonable current estimate is that one dose provides somewhat less protection than two doses.
If you are in a place where there’s lots of current infections, the short-term issue of getting more protection may dominate the decision, so pointing towards getting your second dose as soon as possible.
If you are somewhere with very low prevalence of the disease then it might be more worthwhile to try to optimise for longer-term protection, so could be worth digging into whether you believe that a longer gap is helpful.
You might also want to factor in your enjoyment of the next several weeks. If you’re going to hold off on social contact whilst in the gap between doses, but will feel happy meeting people once you’ve had both doses, then it’s reasonable for you to think how much extra enjoyment you’ll get by pulling the socialisation time sooner.
Another factor that potentially pulls in the direction of earlier second doses is the likelihood of booster vaccinations. I suspect it’s reasonably likely that we’ll be getting boosters or vaccines for new variants each year, like with flu. That tends to make me think that the longer-term effectiveness issue is less important than short term, because you’ll probably be getting more doses in the future that will help you with the long term.
What about Dobble / Spot-It? They are cards designed so each pair of cards has exactly one shared symbol between them.
I think the switch to 8 weeks was at least in part driven by the different characteristics of the Delta variant. Whereas for the previous variants the two doses of the vaccine did something like 70% and 90% protection, for the Delta variant it’s more like 35% and 80% (vague hand wavey numbers). So now there is similar marginal benefit to getting second doses into people, and because the roll out has been by age category, getting (say) 60-year olds from 1 dose to 2 might well be more beneficial than getting 20-year olds from 0 to 1.
I think that this is far more parsimoniously explained by the existing mechanisms working as intended. This decision was made by the JVCI, which was established (according to Wikipedia) in 1963, and having existed for some time before that as a polio advisory board. So there’s a well-established group of experts who have been looking at immunisation schedules for all sorts of diseases for decades.
The last scare is underway. Delta is an increasing share of Covid cases around the world, causing cases in many places to rise.
Describing Delta as the last scare seems like it could be premature. There are lots of people around the world who are not yet vaccinated, and who are catching the disease, and each of those transmissions is effectively a chance for a new variant. Unless there’s a solid reason for thinking that Delta is as effective as this virus can get, surely we need to admit the possibility of future variants that are stronger still, and hence that there may be more scares to come?
And the biggie of those potential future scares would be if we get variants that are more effective at escaping vaccine immunity, right?
You mention near the start of the piece:
Do you have a source / calculation for that? (Not doubting it at all—I am interested in digging into how those values vary for other wind strengths etc., and the search results I’m getting are dominated by calculations on internal air changes.)