Re Monkeypox, part of the solution was gay people pushing for access to the vaccines (far more than public health pushing the vaccines) AND the people engaging in the riskiest behavior generally being the people who were trying to get the vaccines first. If the most effected population is also the most interested in getting vaccines, good things happen
I agree with your take on the likely outcome of the FDA's warning about Nyquil Chicken, but haven't we spent a whole lot of the pandemic wishing that the FDA and CDC wouldn't tailor their messages based on how they _think_ people will respond?
I mean, my critique would probably be "this is completely trivial, why are you wasting time on it", but _if_ they decide it's worth their time, then I'd rather they completely avoid any consideration of the public's response to what they say. They have proven to be totally inept in that regard.
There are near-closed-loop CGM/Insulin pump systems, but they're still not nearly a panacea, because they don't take into account things like activity level and lots of factors that can affect how quickly food is absorbed and causes blood glucose to rise. (Type of carbohydrates, amount of fat, how quickly the person ate, stress levels, hormonal factors., etc.)
A short spike in bG can cause an overcorrection of insulin, creating a dangerous low. The dangerous low can trigger a release of glucagon, causing a long-lasting high hours later. And the amount of insulin it takes for an individual to process a given amount of glucose can vary significantly over long time periods and shorter-term. (For example, women who are menstruating often run high, to the point that there is a setting on insulin pumps to temporarily raise the basal rate in response.
Now feels like a good time to thank you again for this whole series. Every week for quite some time you've been keeping us up-to-date with all the stuff happening with COVID, and a lot of stuff besides. Rationalist bloggers have mostly dropped COVID awhile back, and it's been nice to check in every Thursday for all this data and good rationalist ways of looking at it.
"Trying to dig out from minus a million points" - I think it might be time to add a point to your score ;)
It's 600 pages, but a skim reveals a whole lot of government officials informing Facebook of specific pieces of content that they say are misinformation, and Facebook thanking them and presumably removing the content.
I'm somewhat conflicted on how first-amendment-violationy this behavior is, but mostly leaning towards it being a violation.
On one hand, notwithstanding current efforts, websites are free to establish their own rules on what its users are allowed to post. It's possible that the government played no part in establishing those rules, and was merely informing the website operator about violations of its own rules. If so, the government is not literally ordering any content to be taken down.
On the other hand, I think if we modified the scenario slightly it'd become obviously violating. Imagine instead of a general purpose social media site, the website was a Christian theology forum with a ban on heresy. I can't imagine that the Supreme Court would be cool with the government informing the website operator every time it saw someone posting that Jesus was not the messiah. The government is still not literally censoring, it's relying on the website operator to actually take down the content. But the government knows full well what's going to happen.
I'm sneaking the establishment clause in with that example which invites stricter scrutiny, sure, but I think the core question is unchanged. Is this viewpoint-based censorship or not?
The government tattling on speech to a private party that it can reasonably anticipate will censor the speech is functionally equivalent to the government censoring speech itself but with a random number generator that will sometimes deny the attempt. We definitely wouldn't allow that.
If the government actually helped craft the rules, that's gotta be a slam dunk of a violation, no?
I'm not sold that Texas' internet law (or similar laws) are ultimately workable as a matter of policy and technology. Questions such as filters for spam, profanity, pornography, etc. and extraterritoriality questions are particular challenges.
That said, I take issue with both (1) Zvi's statement that this law is "definitely patently unconstitutional" and (2) Mike Masnick's analogy to requiring news organizations to cover events.
Regarding point 1, the Fifth Circuit opinion references the U.S. Supreme Court's unanimous 1980 decision in Pruneyard ( https://en.wikipedia.org/wiki/Pruneyard_Shopping_Center_v._Robins ). In that case, a group argued that a provision of the California (not U.S.) Constitution allowed them to set up a table at a *private* shopping center, in an area generally open to the public, to collect signatures for a petition. The California Supreme Court agreed.
So there's established U.S. Supreme Court precedent that a state has the ability to require a private owner of a generally public area to allow the public to exercise free speech rights. In short, such a requirement - per SCOTUS - doesn't violate the private owners' First or Fifth Amendment rights. So the reasoning in Pruneyard provides an analogy for what states may require from online platforms as privately owned spaces that are generally open to the public.
As for (2), the Fifth Circuit opinion also references the Supreme Court's 1974 decision in Miami Herald ( https://en.wikipedia.org/wiki/Miami_Herald_Publishing_Co._v._Tornillo ). In that decision, SCOTUS struck down a Florida law that required newspapers to offer equal space to political candidates in the event of political editorials or endorsements. One key factor in that decision was that it penalized certain speech based on its content: a newspaper has finite resources to compose and print content, so requiring it to print an equally prominent reply could dissuade it from writing about the issue at all. A second key factor is - quoting the Fifth Circuit opinion - that "because a newspaper prints a curated set of material selected by its editors, everything it publishes is, in a sense, the newspaper’s own speech."
The distinction that the Fifth Circuit draws is that large social media platforms, unlike news organizations, exercise virtually no editorial control or judgement. So it's abundantly clear that requiring a news organization to devote its resources to covering certain politicians or other content is very different: that imposes a requirement that the news organization devote scarce resources to particular topics and would compel speech by the news organization.
Mandated publication of opinion pieces by politicians is a closer call, though I still see an argument that news organizations apply dramatically more scrutiny and editorial judgement to opinion pieces than social media companies apply to their content. Opinion pieces published by a news organization are therefore speech of the organization in a way that social media posts simply aren't.
This last point is admittedly where I see validity to counter-arguments because of the lack of a bright line. Clearly the opinion pieces - and even letters to the editor - of publications such as the New York Times or Wall Street Journal are curated and reviewed in a way that posts on large social media platforms simply are not. But there are many models in between those two, so where's the bright line between a publisher with editorial discretion and a platform that can (at least arguably) be viewed as a common carrier of communications, more akin to a telecom operator than to an enterprise exercising editorial judgement over its content? So that's a reason that I'm hesitant about whether these laws are ultimately workable, but I see it as a much closer call than Zvi or the people who he quotes here.
Thanks for making your Omicron bet and sharing your thoughts on it looking back. I commented on the Metaculus question to explain my disagreement with your analysis of your mistake.
Copying it here (without hyperlinks):
I disagree with Zvi on:
> In hindsight, the question of ‘what counts as Omicron’ does have a strong bearing on who had the right side of this wager, and also is a key insight into the mistake that I made here.
Even if subvariants counted as different variants that would only increase the chance that the bet resolves ambiguously. There was (IMO) never a >70% chance (or even >50%) chance that Zvi would win (conditional on someone winning), even if the language of the bet considered subvariants to be different variant. So failure to examine "what counts as Omicron" closely does not explain why a 70% forecast would seem otherwise-reasonable--it still wouldn't be reasonable.
I think Zvi's mistake was what I wrote last December:
> (1) For Zvi to win, Covid cases (of all variants, including any future ones) need to average ~18 times higher over the first two months of 2022 than over the following 12 months (math: 18 = (0.75/2) / (0.25/12)).
> 18 is such a high ratio given Covid's track record so far. The ratio of the 7-day-average of US cases from its high (1/11/2021 = ~256,000/day) to its low (6/21/2021 = ~12,000/day) is ~21, barely higher than ~18. Plus, those two weeks were months apart, giving time for cases to drop off.
I think this simple base rate argument was enough to see that Zvi was very unlikely to win. Absent some strong reason to know that Covid would not become endemic, there was just very little chance Zvi would win, because even with a big Omicron wave like happened, 18:1 ratio is just too high for the following year period to not either cause Holden to win or cause the bet to resolve ambiguously (in the best case scenario for Zvi in which a new variant comes along quickly before Holden's win condition is met).
I think my original explanation from December of Zvi's poor bet as being a poor operationalization of a reasonable view still seems like a good explanation in retrospect:
> My thought was that Zvi's view in the conceptual disagreement they were betting on seems much more plausible to me than Zvi's position in the bet operationalization. That is, there are many scenarios that would make it look like Zvi was basically right that might technically cache out as a Holden win according to the exact betting terms described here. For example, there might be a huge Omicron wave--the largest Covid wave yet--and cases might drop quickly afterwards and it might be the last wave of the pandemic, and yet despite all of that, perhaps only 50% of the cases after January 1, 2022 happen before the end of February rather than the 75% necessary for Zvi to win.
Between the people who 'don't eat out' which aren't included, and the eating out that's not indoor dining, I'd be kind of surprised if the median American went out for indoor dining more than once a week. It's EXPENSIVE.
I am a retired submariner, twenty years operating submarine engine rooms on both operational submarines and land-based submarine training platforms (Nuclear Power Training Units or NPTU).
Your intuitions on this are unfortunately way off. A vast majority of a submarines steam power is used for propulsion, we called them the Main Engines, and they use steam to mechanically turn the screw, or in the newer boats, the propulsor. There isn't any way to convert that to electricity and keep the submarine operational. Think of it like a car engine, you use it to produce electricity with the alternator, but if you want to use the mechanical side to generate electricity then those modifications would make the car undrivable.
Additionally, Naval Nuclear power plants aren't going to be able to be used as land-based generators for a few reasons.
1. they are built to have a full lifetime fuel load, extremely difficult to refuel. They do this by using uranium enriched to amounts way above the civilian limits of ~5%. No one outside of the government can use the enrichment levels in a submarine reactor.
2. Submarines don't have an equivalent of the Emergency Core Cooling Systems (ECCS) used in civilian Reactors. The land-based training reactors the Navy uses have this added as a separate seismically rated building at NPTU Ballston Spa, or as a specially built section added to the Moored Training Ships in Charleston. Naval Nuclear considers this acceptable since they are rarely operated near land and when they are it's a low power levels, so there is very little risk of an accident happening that would cause an increased dose to the public.
A common refrain in the Navy is that rust never sleeps. Putting anything in water (especially salt water) for long periods of time means a lot of maintenance needed for preservation. Submarines (and Aircraft Carriers) are designed for this by necessity, and their cooling systems are based on just taking a suction on the ocean. You wouldn't intentionally do this for an electrical generation plant, so simply contracting with Electric Boat or Norfolk Naval to build just extra power plants and use them on land wouldn't work without significant modifications.
All of the above is to explain why you can't just piggyback on the efficiencies gained by building two submarines a year to feed nuclear electricity generation cheaper. Once you accept that, there is no reason to accept the lower efficiency of the Navy Nuclear design decisions. The Naval plants are made to change power very quickly (slow and quiet to run for your life in less than a handful of munities). Example: Rapidly changing power means significant swings in temperature of your turbines, and different parts of the turbine heat up or cool down at different rates. The Naval turbines have (relatively) a lot of space between the spinning part (blades or buckets) and the stationary part because those two can never touch. This means more tip leakage reducing efficiency, an acceptable tradeoff if you're a submarine, but not if your only purpose is to generate electricity.
I'm a big fan of Nuclear Power, and I currently work at a civilian nuclear generating station, but there is no hope of using anything on the Navy side to bootstrap civilian nuclear electricity generation. Please update accordingly.
Re Monkeypox, part of the solution was gay people pushing for access to the vaccines (far more than public health pushing the vaccines) AND the people engaging in the riskiest behavior generally being the people who were trying to get the vaccines first. If the most effected population is also the most interested in getting vaccines, good things happen
I agree with your take on the likely outcome of the FDA's warning about Nyquil Chicken, but haven't we spent a whole lot of the pandemic wishing that the FDA and CDC wouldn't tailor their messages based on how they _think_ people will respond?
I mean, my critique would probably be "this is completely trivial, why are you wasting time on it", but _if_ they decide it's worth their time, then I'd rather they completely avoid any consideration of the public's response to what they say. They have proven to be totally inept in that regard.
There are near-closed-loop CGM/Insulin pump systems, but they're still not nearly a panacea, because they don't take into account things like activity level and lots of factors that can affect how quickly food is absorbed and causes blood glucose to rise. (Type of carbohydrates, amount of fat, how quickly the person ate, stress levels, hormonal factors., etc.)
A short spike in bG can cause an overcorrection of insulin, creating a dangerous low. The dangerous low can trigger a release of glucagon, causing a long-lasting high hours later. And the amount of insulin it takes for an individual to process a given amount of glucose can vary significantly over long time periods and shorter-term. (For example, women who are menstruating often run high, to the point that there is a setting on insulin pumps to temporarily raise the basal rate in response.
Now feels like a good time to thank you again for this whole series. Every week for quite some time you've been keeping us up-to-date with all the stuff happening with COVID, and a lot of stuff besides. Rationalist bloggers have mostly dropped COVID awhile back, and it's been nice to check in every Thursday for all this data and good rationalist ways of looking at it.
"Trying to dig out from minus a million points" - I think it might be time to add a point to your score ;)
I'm curious about your take on Missouri v. Biden. I first read about it today in this Tablet article: https://www.tabletmag.com/sections/arts-letters/articles/government-privatized-censorship-regime
On the Hacker News thread where I came across the article I also came across a commenter that posted some of the discovery in the case, here: https://nclalegal.org/wp-content/uploads/2022/09/Joint-Statement-on-Discovery-Disputes-Combined.pdf
It's 600 pages, but a skim reveals a whole lot of government officials informing Facebook of specific pieces of content that they say are misinformation, and Facebook thanking them and presumably removing the content.
I'm somewhat conflicted on how first-amendment-violationy this behavior is, but mostly leaning towards it being a violation.
On one hand, notwithstanding current efforts, websites are free to establish their own rules on what its users are allowed to post. It's possible that the government played no part in establishing those rules, and was merely informing the website operator about violations of its own rules. If so, the government is not literally ordering any content to be taken down.
On the other hand, I think if we modified the scenario slightly it'd become obviously violating. Imagine instead of a general purpose social media site, the website was a Christian theology forum with a ban on heresy. I can't imagine that the Supreme Court would be cool with the government informing the website operator every time it saw someone posting that Jesus was not the messiah. The government is still not literally censoring, it's relying on the website operator to actually take down the content. But the government knows full well what's going to happen.
I'm sneaking the establishment clause in with that example which invites stricter scrutiny, sure, but I think the core question is unchanged. Is this viewpoint-based censorship or not?
The government tattling on speech to a private party that it can reasonably anticipate will censor the speech is functionally equivalent to the government censoring speech itself but with a random number generator that will sometimes deny the attempt. We definitely wouldn't allow that.
If the government actually helped craft the rules, that's gotta be a slam dunk of a violation, no?
T2 diabetes is easier in one way, harder in another - but doable. https://www.virtahealth.com/
I'm not sold that Texas' internet law (or similar laws) are ultimately workable as a matter of policy and technology. Questions such as filters for spam, profanity, pornography, etc. and extraterritoriality questions are particular challenges.
That said, I take issue with both (1) Zvi's statement that this law is "definitely patently unconstitutional" and (2) Mike Masnick's analogy to requiring news organizations to cover events.
For reference, the Fifth Circuit opinion is here - https://www.ca5.uscourts.gov/opinions/pub/21/21-51178-CV1.pdf
Regarding point 1, the Fifth Circuit opinion references the U.S. Supreme Court's unanimous 1980 decision in Pruneyard ( https://en.wikipedia.org/wiki/Pruneyard_Shopping_Center_v._Robins ). In that case, a group argued that a provision of the California (not U.S.) Constitution allowed them to set up a table at a *private* shopping center, in an area generally open to the public, to collect signatures for a petition. The California Supreme Court agreed.
So there's established U.S. Supreme Court precedent that a state has the ability to require a private owner of a generally public area to allow the public to exercise free speech rights. In short, such a requirement - per SCOTUS - doesn't violate the private owners' First or Fifth Amendment rights. So the reasoning in Pruneyard provides an analogy for what states may require from online platforms as privately owned spaces that are generally open to the public.
As for (2), the Fifth Circuit opinion also references the Supreme Court's 1974 decision in Miami Herald ( https://en.wikipedia.org/wiki/Miami_Herald_Publishing_Co._v._Tornillo ). In that decision, SCOTUS struck down a Florida law that required newspapers to offer equal space to political candidates in the event of political editorials or endorsements. One key factor in that decision was that it penalized certain speech based on its content: a newspaper has finite resources to compose and print content, so requiring it to print an equally prominent reply could dissuade it from writing about the issue at all. A second key factor is - quoting the Fifth Circuit opinion - that "because a newspaper prints a curated set of material selected by its editors, everything it publishes is, in a sense, the newspaper’s own speech."
The distinction that the Fifth Circuit draws is that large social media platforms, unlike news organizations, exercise virtually no editorial control or judgement. So it's abundantly clear that requiring a news organization to devote its resources to covering certain politicians or other content is very different: that imposes a requirement that the news organization devote scarce resources to particular topics and would compel speech by the news organization.
Mandated publication of opinion pieces by politicians is a closer call, though I still see an argument that news organizations apply dramatically more scrutiny and editorial judgement to opinion pieces than social media companies apply to their content. Opinion pieces published by a news organization are therefore speech of the organization in a way that social media posts simply aren't.
This last point is admittedly where I see validity to counter-arguments because of the lack of a bright line. Clearly the opinion pieces - and even letters to the editor - of publications such as the New York Times or Wall Street Journal are curated and reviewed in a way that posts on large social media platforms simply are not. But there are many models in between those two, so where's the bright line between a publisher with editorial discretion and a platform that can (at least arguably) be viewed as a common carrier of communications, more akin to a telecom operator than to an enterprise exercising editorial judgement over its content? So that's a reason that I'm hesitant about whether these laws are ultimately workable, but I see it as a much closer call than Zvi or the people who he quotes here.
Thanks for making your Omicron bet and sharing your thoughts on it looking back. I commented on the Metaculus question to explain my disagreement with your analysis of your mistake.
Copying it here (without hyperlinks):
I disagree with Zvi on:
> In hindsight, the question of ‘what counts as Omicron’ does have a strong bearing on who had the right side of this wager, and also is a key insight into the mistake that I made here.
Even if subvariants counted as different variants that would only increase the chance that the bet resolves ambiguously. There was (IMO) never a >70% chance (or even >50%) chance that Zvi would win (conditional on someone winning), even if the language of the bet considered subvariants to be different variant. So failure to examine "what counts as Omicron" closely does not explain why a 70% forecast would seem otherwise-reasonable--it still wouldn't be reasonable.
I think Zvi's mistake was what I wrote last December:
> (1) For Zvi to win, Covid cases (of all variants, including any future ones) need to average ~18 times higher over the first two months of 2022 than over the following 12 months (math: 18 = (0.75/2) / (0.25/12)).
> 18 is such a high ratio given Covid's track record so far. The ratio of the 7-day-average of US cases from its high (1/11/2021 = ~256,000/day) to its low (6/21/2021 = ~12,000/day) is ~21, barely higher than ~18. Plus, those two weeks were months apart, giving time for cases to drop off.
I think this simple base rate argument was enough to see that Zvi was very unlikely to win. Absent some strong reason to know that Covid would not become endemic, there was just very little chance Zvi would win, because even with a big Omicron wave like happened, 18:1 ratio is just too high for the following year period to not either cause Holden to win or cause the bet to resolve ambiguously (in the best case scenario for Zvi in which a new variant comes along quickly before Holden's win condition is met).
I think my original explanation from December of Zvi's poor bet as being a poor operationalization of a reasonable view still seems like a good explanation in retrospect:
> My thought was that Zvi's view in the conceptual disagreement they were betting on seems much more plausible to me than Zvi's position in the bet operationalization. That is, there are many scenarios that would make it look like Zvi was basically right that might technically cache out as a Holden win according to the exact betting terms described here. For example, there might be a huge Omicron wave--the largest Covid wave yet--and cases might drop quickly afterwards and it might be the last wave of the pandemic, and yet despite all of that, perhaps only 50% of the cases after January 1, 2022 happen before the end of February rather than the 75% necessary for Zvi to win.
> Given Americans indoor dine something like five times a week
Wait what?! They do? And here I feel wasteful if it's TWICE per week.
I don't think your fact is right here. According to this: https://www.cbsnews.com/minnesota/news/survey-shows-how-often-americans-dine-out/ the median American who eats out does so 2-3 times per week. And indoor dining is a subset of eating out; you can also eat outside or use a drive thru.
Between the people who 'don't eat out' which aren't included, and the eating out that's not indoor dining, I'd be kind of surprised if the median American went out for indoor dining more than once a week. It's EXPENSIVE.
Nuclear Submarines:
I am a retired submariner, twenty years operating submarine engine rooms on both operational submarines and land-based submarine training platforms (Nuclear Power Training Units or NPTU).
Your intuitions on this are unfortunately way off. A vast majority of a submarines steam power is used for propulsion, we called them the Main Engines, and they use steam to mechanically turn the screw, or in the newer boats, the propulsor. There isn't any way to convert that to electricity and keep the submarine operational. Think of it like a car engine, you use it to produce electricity with the alternator, but if you want to use the mechanical side to generate electricity then those modifications would make the car undrivable.
Additionally, Naval Nuclear power plants aren't going to be able to be used as land-based generators for a few reasons.
1. they are built to have a full lifetime fuel load, extremely difficult to refuel. They do this by using uranium enriched to amounts way above the civilian limits of ~5%. No one outside of the government can use the enrichment levels in a submarine reactor.
2. Submarines don't have an equivalent of the Emergency Core Cooling Systems (ECCS) used in civilian Reactors. The land-based training reactors the Navy uses have this added as a separate seismically rated building at NPTU Ballston Spa, or as a specially built section added to the Moored Training Ships in Charleston. Naval Nuclear considers this acceptable since they are rarely operated near land and when they are it's a low power levels, so there is very little risk of an accident happening that would cause an increased dose to the public.
A common refrain in the Navy is that rust never sleeps. Putting anything in water (especially salt water) for long periods of time means a lot of maintenance needed for preservation. Submarines (and Aircraft Carriers) are designed for this by necessity, and their cooling systems are based on just taking a suction on the ocean. You wouldn't intentionally do this for an electrical generation plant, so simply contracting with Electric Boat or Norfolk Naval to build just extra power plants and use them on land wouldn't work without significant modifications.
All of the above is to explain why you can't just piggyback on the efficiencies gained by building two submarines a year to feed nuclear electricity generation cheaper. Once you accept that, there is no reason to accept the lower efficiency of the Navy Nuclear design decisions. The Naval plants are made to change power very quickly (slow and quiet to run for your life in less than a handful of munities). Example: Rapidly changing power means significant swings in temperature of your turbines, and different parts of the turbine heat up or cool down at different rates. The Naval turbines have (relatively) a lot of space between the spinning part (blades or buckets) and the stationary part because those two can never touch. This means more tip leakage reducing efficiency, an acceptable tradeoff if you're a submarine, but not if your only purpose is to generate electricity.
I'm a big fan of Nuclear Power, and I currently work at a civilian nuclear generating station, but there is no hope of using anything on the Navy side to bootstrap civilian nuclear electricity generation. Please update accordingly.