Chris Martenson is one of the people who got it right back in January and his latest video is quite interesting
I watched the video and I think it hard to convey just how bad it is. The guy is also annoyingly smug (the first ten minutes of the video are just him reading headlines and saying the word “dangerous” in a mocking voice) while being wrong about his important claims.
The tl;dr is that while Delta has a significantly higher r data from Israel suggests that this doesn’t translate into increased deaths. This suggest that the virus is less harmful
Vaccines work great at preventing severe disease and deaths. Obviously one of the highest-vaccinated countries will have fewer deaths now – that says nothing about how deadly the Delta variant is in an unvaccinated population.
And then he goes on to talk about how there are more vaccinated people aged >50 in the UK hospitals than unvaccinated people, and he thinks that’s strange or even implies that it could mean something bad about vaccine efficiency. He doesn’t understand base rates (9/10 people over 50 are vaccinated in the UK!).
Doing the analysis right, we see that Delta is probably significantly more deadly than Alpha, which itself was more deadly than the original variant. We also see that Delta affects children more.
(Edit: I initially thought he’s also wrong about not factoring in that the majority of Delta variant cases in the UK haven’t had enough time to run their course, but I think the chart he used actually factored that in and was considering cases with a known outcome. So I deleted a paragraph above – one fewer mistake than I initially thought!)
Obviously one of the highest-vaccinated countries will have fewer deaths now – that says nothing about how deadly the Delta variant is in an unvaccinated population.
I don’t think vaccinations resulting in fewer deaths explain how the death rate in Israel stays relatively constant while the case rate goes up.
One thing worth noting here is also that just because both the strains in the UK and Israel are classified as Delta doesn’t mean that there can’t be differences that are clinically relevant.
The weekly average of deaths went from roughly zero, to one, to two. That’s going up. Soon it will be at three or four if things continue like that. That looks like normal growth, deaths always lag behind surprisingly much when case numbers are growing exponentially. It also took the UK a really long time to from 7 weekly deaths to even just 10, but now the weekly average is above 50.
From my perspective it seems relatively hard to know what’s happening. I’d love to have some prediction for the Israeli numbers in 2-4 weeks, so that I can then update to believe your thesis or Martenson. Do you feel confident to make some prediction in confidence intervals of where you believe the numbers will be?
Scientific consensus opinion is that Delta is more deadly, not less. The guy in the video says otherwise, but his specific arguments are flawed. This isn’t difficult to check/verify: He didn’t factor in that the death rate is automatically lower in a vaccinated population.
Without the evidence he thought he had from the UK data, all that’s left in support of his position, that Delta is less deadly, is this: “Intuitively, deaths are climbing really slowly in Israel compared to case numbers.”
By itself, intuition about deaths climbing slowly doesn’t seem like anywhere close to a good reason to question expert consensus. (It also wouldn’t make me think that Australia may have its own less deadly subtype of Delta.)
I feel like if your plan is to wait 2-4 weeks to check my predictions against reality, you’re losing time unnecessarily. There’s enough info here to update sooner.
Maybe we have different intuitions about how weird the slowly climbing death rate is. I think it’s not weird at all – it’s always gone this way. You can compare the situation in Israel now to how things were in the UK throughout June, see the charts here, specifically comparing how the rise in case numbers came early than the rise in death numbers.
Or, to give another example, back in early 2020, people kept pointing out that the case fatality rate in South Korea seems really low, and that this means the virus isn’t that bad. But that was just the lag from cases to deaths, and after South Korea’s outbreak went beyond its peak and people had time to die, the case fatality rate went up by multiples!
Regarding predictions: Israel only has 1k cases and they do extensive testing, so there’s not a lot of underreporting. This makes predictions a bit difficult because the variance is high. Probably we’ll see at least 4 average daily deaths in three weeks. But most importantly, and most confidently, I’ll say that as long as cases keep rising, deaths will eventually go up as well – that’s the prediction of the model that Delta isn’t weaker.
I have a good track record on Covid predictions. I won the first big forecasting tournament on Covid on Metaculus and got 3rd in the second installment of the tournament. I live in the UK and therefore have a headstart (except vs. people in India) on following Delta developments closely. I’m also a bit addicted to virus news and spend 2h per day on that.
Scientific consensus opinion is that Delta is more deadly, not less
The phrase scientific consensus seems to me quite ill-placed in a state of evidence like this. Scientific consensus is usually something that takes years to build for a hypothesis. At the moment it usually gets used to manipulate people to treat positions that come out of a very political process has having to do something with science.
I think the phrase should only be used when there’s at least a meta-study that looked at the literature and found consensus.
That said, what makes you believe that this is a “Scientific consensus opinion”?
I feel like if your plan is to wait 2-4 weeks to check my predictions against reality, you’re losing time unnecessarily. There’s enough info here to update sooner.
Losing time suggests that there’s an action I’m currently not taking that I should be taking. It’s likely that the same action I take now is twice as expensive in terms of COVID-19 risk in a month (given COVID numbers where I live)
It’s really not that difficult to ballpark IFR estimates (especially now that everyone knows that there’s no giant iceberg of asymptomatic cases). The Delta variant has been around for a while. You can play word games but it’s fairly obvious what I mean by “scientific consensus.” I mean that >100 media articles I’ve read in the last couple of months, from various mainstream UK sources, I’ve never seen anyone seriously entertain the hypothesis that the Delta variant is less deadly than previous variants.
I’m not saying to blanket endorse the media’s perspective on what’s the scientific consensus. That would be strawmanning me. I’m saying that when you claim that the consensus is wrong (edit: or that it very well might be), at least have more of a reason than “Intuitively, those numbers look too low on the chart.” At least have an argument for what, specifically, the mainstream experts are getting wrong. The video doesn’t have that since it fails to pass the Ideological Turing test for people who are concerned about the Delta variant.
Losing time suggests that there’s an action I’m currently not taking that I should be taking. It’s likely that the same action I take now is twice as expensive in terms of COVID-19 risk in a month (given COVID numbers where I live)
Having accurate beliefs can be beneficial in unanticipated ways. There’s no point in needlessly delaying epistemic updates. I’m saying that there’s something off about how much credibility you give to confident-sounding contrarians on youtube (with some credentials, admittedly, though not that it matters). (Esp. once some of their core arguments get debunked.)
Again, I’m not making a blanket argument about contrarians always being wrong. I’m making the specific argument that contrarians who are right don’t tend to make easily visible mistakes.
I mean that >100 media articles I’ve read in the last couple of months, from various mainstream UK sources, I’ve never seen anyone seriously entertain the hypothesis that the Delta variant is less deadly than previous variants.
That’s journalistic consensus and calling it scientific consensus is an insult to the scientific project. For society it’s very costly to confuse academic discourse with journalistic discourse. There’s a huge cost for the public trust in science from confusing the two. Making the public case that while there are a bunch of media articles suggesting that climate change doesn’t exist there’s a scientific consensus that it clearly exists wasn’t an easy sell. There’s huge collateral damage from equating the two for propaganda purposes.
The decision of what mainstream journalistic sources print depends on editorial policy and is not based on asking a representative sample of scientists in a field.
Chris Martenson was telling people to get masks in January 2020. That might make him 4-5 months ahead of the journalistic consensus on getting the mask question right. In general the journalistic discourse on COVID-19 is so bad that people who open windows in public transportation get angry looks for it because the journalist failed to tell people that opening windows and getting airflow is very useful for reducing COVID-19 risk.
You’re right, I think I phrased my point poorly. What I should have said is “If there’s no reputable or intelligent-seeming person making some claim for a long time, and then comes along some lone contrarian, that contrarian better make a good impression – otherwise it probably doesn’t make sense to invest a lot of time steelmanning their claims and digging into a hypothesis that wouldn’t even have occured to us without this person.”
Maybe you’ve seen more people talk about Delta deadliness concerns being overblown. But based on my media diet as well as based on all the conversations I’ve had with EAs and rationalists about this, the view that Delta might be a less infectious (but more contagious) variant has never come up.
It’s true that media consensus isn’t great to go by, as we’ve seen with all the instances Zvi documents in his post (mask wearing, possibility of lab escape, etc.).
The UK had a political fight about whether or not to end lockdown and what you read in UK mainstream sources likely reflects how the outlets editorial position stands in the fight.
I do think that having a good track record is what makes people credible and rather listen to people who got COVID-19 right at the start then people who don’t and who published crap about masks don’t working because they were to lazy to do real journalism and critically report.
If we take the Guardian as an example, what possible reason can there be to not publish a single article about patient zero despite Huang Yanling being recognized as important in the NIH letter to the EcoHealth Alliance? Outlets that act that strange should not be trusted.
I do think that having a good track record is what makes people credible and rather listen to people who got COVID-19 right at the start
I got Covid right at the start (among others) and I posted above that my track record is winning the largest prediction tournament at the time.
So let me repeat what I cared about conveying in this discussion, one last time:
The person in the youtube video you linked to may have gotten Covid right in early 2020, but so did hundreds of people (but maybe not thousands). Out of the set of people with a good track record on Covid, this guy is now pushing an extreme minority position. In theory, he could have been right with that. But he’s wrong because his arguments are bad in an easily verifiable way. Once someone’s core arguments for a fringe position (fringe in reference to the best sources we’d want to listen to here, not fringe with respect to the media) get discredited, there’s no reason to continue treating the fringe position as though it still has a high chance of being right. By that point, we must be prepared to say “This guy went off the rails.”
All I wanted to convey is that it makes no sense to continue holding a person’s specific opinion in high regards (good track record or not) when the opinion is highly contrarian* and just had its core arguments refuted. By continuing to argue as though the guy might still be right, you were employing a type of epistemology which, to me, seems doomed. I get the impression that you not only distrust the media consensus, but any consensus seems worthless to you when you see a single confident-sounding expert who stands out as having gotten something right when others had gotten it wrong. I think that’s too strong of an update, because lots of people got things right, and some of them may still be completely nuts and bad at reasoning, and we can spot that by checking things against a mental reference class of “consensus among the people we hold in high esteem.”
*Again, I’m talking about contrarian with respect to the sources we’d want to listen to. That’s a subjective reference class, but since we’re both on this site and value track records, our takes on this may not be crazily different. It is my highly confident impression that <3% of Lesswrongers with high karma, and <3% of people who got Covid right in the early days, think that the Delta variant is less deadly in unvaccinated population than the original variant.
BTW, it’s am minor point, but I feel like the media is biased to fail to identify new variants as more deadly, because of racism concerns. Every single media article about the South Africa variant said “there’s no evidence it’s more deadly.” They were saying this before there was enough time to know with confidence (and “no evidence” was technically false because there were anecdotal reports of children being more affected).
I watched the video and I think it hard to convey just how bad it is. The guy is also annoyingly smug (the first ten minutes of the video are just him reading headlines and saying the word “dangerous” in a mocking voice) while being wrong about his important claims.
Vaccines work great at preventing severe disease and deaths. Obviously one of the highest-vaccinated countries will have fewer deaths now – that says nothing about how deadly the Delta variant is in an unvaccinated population.
And then he goes on to talk about how there are more vaccinated people aged >50 in the UK hospitals than unvaccinated people, and he thinks that’s strange or even implies that it could mean something bad about vaccine efficiency. He doesn’t understand base rates (9/10 people over 50 are vaccinated in the UK!).
Doing the analysis right, we see that Delta is probably significantly more deadly than Alpha, which itself was more deadly than the original variant. We also see that Delta affects children more.
(Edit: I initially thought he’s also wrong about not factoring in that the majority of Delta variant cases in the UK haven’t had enough time to run their course, but I think the chart he used actually factored that in and was considering cases with a known outcome. So I deleted a paragraph above – one fewer mistake than I initially thought!)
I don’t think vaccinations resulting in fewer deaths explain how the death rate in Israel stays relatively constant while the case rate goes up.
One thing worth noting here is also that just because both the strains in the UK and Israel are classified as Delta doesn’t mean that there can’t be differences that are clinically relevant.
The weekly average of deaths went from roughly zero, to one, to two. That’s going up. Soon it will be at three or four if things continue like that. That looks like normal growth, deaths always lag behind surprisingly much when case numbers are growing exponentially. It also took the UK a really long time to from 7 weekly deaths to even just 10, but now the weekly average is above 50.
From my perspective it seems relatively hard to know what’s happening. I’d love to have some prediction for the Israeli numbers in 2-4 weeks, so that I can then update to believe your thesis or Martenson. Do you feel confident to make some prediction in confidence intervals of where you believe the numbers will be?
Scientific consensus opinion is that Delta is more deadly, not less. The guy in the video says otherwise, but his specific arguments are flawed. This isn’t difficult to check/verify: He didn’t factor in that the death rate is automatically lower in a vaccinated population.
Without the evidence he thought he had from the UK data, all that’s left in support of his position, that Delta is less deadly, is this: “Intuitively, deaths are climbing really slowly in Israel compared to case numbers.”
By itself, intuition about deaths climbing slowly doesn’t seem like anywhere close to a good reason to question expert consensus. (It also wouldn’t make me think that Australia may have its own less deadly subtype of Delta.)
I feel like if your plan is to wait 2-4 weeks to check my predictions against reality, you’re losing time unnecessarily. There’s enough info here to update sooner.
Maybe we have different intuitions about how weird the slowly climbing death rate is. I think it’s not weird at all – it’s always gone this way. You can compare the situation in Israel now to how things were in the UK throughout June, see the charts here, specifically comparing how the rise in case numbers came early than the rise in death numbers.
Or, to give another example, back in early 2020, people kept pointing out that the case fatality rate in South Korea seems really low, and that this means the virus isn’t that bad. But that was just the lag from cases to deaths, and after South Korea’s outbreak went beyond its peak and people had time to die, the case fatality rate went up by multiples!
Regarding predictions: Israel only has 1k cases and they do extensive testing, so there’s not a lot of underreporting. This makes predictions a bit difficult because the variance is high. Probably we’ll see at least 4 average daily deaths in three weeks. But most importantly, and most confidently, I’ll say that as long as cases keep rising, deaths will eventually go up as well – that’s the prediction of the model that Delta isn’t weaker.
I have a good track record on Covid predictions. I won the first big forecasting tournament on Covid on Metaculus and got 3rd in the second installment of the tournament. I live in the UK and therefore have a headstart (except vs. people in India) on following Delta developments closely. I’m also a bit addicted to virus news and spend 2h per day on that.
The phrase scientific consensus seems to me quite ill-placed in a state of evidence like this. Scientific consensus is usually something that takes years to build for a hypothesis. At the moment it usually gets used to manipulate people to treat positions that come out of a very political process has having to do something with science.
I think the phrase should only be used when there’s at least a meta-study that looked at the literature and found consensus.
That said, what makes you believe that this is a “Scientific consensus opinion”?
Losing time suggests that there’s an action I’m currently not taking that I should be taking. It’s likely that the same action I take now is twice as expensive in terms of COVID-19 risk in a month (given COVID numbers where I live)
It’s really not that difficult to ballpark IFR estimates (especially now that everyone knows that there’s no giant iceberg of asymptomatic cases). The Delta variant has been around for a while. You can play word games but it’s fairly obvious what I mean by “scientific consensus.” I mean that >100 media articles I’ve read in the last couple of months, from various mainstream UK sources, I’ve never seen anyone seriously entertain the hypothesis that the Delta variant is less deadly than previous variants.
I’m not saying to blanket endorse the media’s perspective on what’s the scientific consensus. That would be strawmanning me. I’m saying that when you claim that the consensus is wrong (edit: or that it very well might be), at least have more of a reason than “Intuitively, those numbers look too low on the chart.” At least have an argument for what, specifically, the mainstream experts are getting wrong. The video doesn’t have that since it fails to pass the Ideological Turing test for people who are concerned about the Delta variant.
Having accurate beliefs can be beneficial in unanticipated ways. There’s no point in needlessly delaying epistemic updates. I’m saying that there’s something off about how much credibility you give to confident-sounding contrarians on youtube (with some credentials, admittedly, though not that it matters). (Esp. once some of their core arguments get debunked.)
Again, I’m not making a blanket argument about contrarians always being wrong. I’m making the specific argument that contrarians who are right don’t tend to make easily visible mistakes.
That’s journalistic consensus and calling it scientific consensus is an insult to the scientific project. For society it’s very costly to confuse academic discourse with journalistic discourse. There’s a huge cost for the public trust in science from confusing the two. Making the public case that while there are a bunch of media articles suggesting that climate change doesn’t exist there’s a scientific consensus that it clearly exists wasn’t an easy sell. There’s huge collateral damage from equating the two for propaganda purposes.
The decision of what mainstream journalistic sources print depends on editorial policy and is not based on asking a representative sample of scientists in a field.
Chris Martenson was telling people to get masks in January 2020. That might make him 4-5 months ahead of the journalistic consensus on getting the mask question right. In general the journalistic discourse on COVID-19 is so bad that people who open windows in public transportation get angry looks for it because the journalist failed to tell people that opening windows and getting airflow is very useful for reducing COVID-19 risk.
You’re right, I think I phrased my point poorly. What I should have said is “If there’s no reputable or intelligent-seeming person making some claim for a long time, and then comes along some lone contrarian, that contrarian better make a good impression – otherwise it probably doesn’t make sense to invest a lot of time steelmanning their claims and digging into a hypothesis that wouldn’t even have occured to us without this person.”
Maybe you’ve seen more people talk about Delta deadliness concerns being overblown. But based on my media diet as well as based on all the conversations I’ve had with EAs and rationalists about this, the view that Delta might be a less infectious (but more contagious) variant has never come up.
It’s true that media consensus isn’t great to go by, as we’ve seen with all the instances Zvi documents in his post (mask wearing, possibility of lab escape, etc.).
The UK had a political fight about whether or not to end lockdown and what you read in UK mainstream sources likely reflects how the outlets editorial position stands in the fight.
I do think that having a good track record is what makes people credible and rather listen to people who got COVID-19 right at the start then people who don’t and who published crap about masks don’t working because they were to lazy to do real journalism and critically report.
If we take the Guardian as an example, what possible reason can there be to not publish a single article about patient zero despite Huang Yanling being recognized as important in the NIH letter to the EcoHealth Alliance? Outlets that act that strange should not be trusted.
I feel like we got sidetracked.
I got Covid right at the start (among others) and I posted above that my track record is winning the largest prediction tournament at the time.
So let me repeat what I cared about conveying in this discussion, one last time:
The person in the youtube video you linked to may have gotten Covid right in early 2020, but so did hundreds of people (but maybe not thousands). Out of the set of people with a good track record on Covid, this guy is now pushing an extreme minority position. In theory, he could have been right with that. But he’s wrong because his arguments are bad in an easily verifiable way. Once someone’s core arguments for a fringe position (fringe in reference to the best sources we’d want to listen to here, not fringe with respect to the media) get discredited, there’s no reason to continue treating the fringe position as though it still has a high chance of being right. By that point, we must be prepared to say “This guy went off the rails.”
All I wanted to convey is that it makes no sense to continue holding a person’s specific opinion in high regards (good track record or not) when the opinion is highly contrarian* and just had its core arguments refuted. By continuing to argue as though the guy might still be right, you were employing a type of epistemology which, to me, seems doomed. I get the impression that you not only distrust the media consensus, but any consensus seems worthless to you when you see a single confident-sounding expert who stands out as having gotten something right when others had gotten it wrong. I think that’s too strong of an update, because lots of people got things right, and some of them may still be completely nuts and bad at reasoning, and we can spot that by checking things against a mental reference class of “consensus among the people we hold in high esteem.”
*Again, I’m talking about contrarian with respect to the sources we’d want to listen to. That’s a subjective reference class, but since we’re both on this site and value track records, our takes on this may not be crazily different. It is my highly confident impression that <3% of Lesswrongers with high karma, and <3% of people who got Covid right in the early days, think that the Delta variant is less deadly in unvaccinated population than the original variant.
BTW, it’s am minor point, but I feel like the media is biased to fail to identify new variants as more deadly, because of racism concerns. Every single media article about the South Africa variant said “there’s no evidence it’s more deadly.” They were saying this before there was enough time to know with confidence (and “no evidence” was technically false because there were anecdotal reports of children being more affected).