https://lbry.tv/Mike-Yeadon-Unlocked:0
For reference, this is the former head of research for Pfizer. The video (and him) have been banned off all the social media platforms due to “misinformation” (really a disagreement about the efficacy of the PCR protocol and how it has lead to a larger false positive rate).
First of all, the number of tests isn't growing at the same rate as the number of cases, hospitalizations and deaths: https://imgur.com/2cVrGz7
You can notice the very low positive rate in mid August (150 000 tests and only 500 cases - even if all false positives, thats 0.3% FP rate)
But okay, lets assume that they did something to ramp up the false positive rate, If we look at the daily rate of growth (% increase) for cases, hospitalizations and deaths:
We can see that there is a perfect exponential 7% rate of growth starting in September and slowly decaying after. But more than that, the rate of growth in hospitalizations and deaths also follows the same pattern almost perfectly but with a 5 day and 16 day delay, respectively!
If you were accidentally getting false positives, you would either not expect hospitalizations and deaths to grow at the same rate as false positives do. Even if they did, you would expect them to grow at exactly the same time without any delays. If there is no causal relationship they should all grow at the same rate, at the same time as any other random subgroup of people
To get the delay to happen, you would have to somehow perfectly fine tune the false positives for hospitalizations and deaths as well. In which case you're better off just making numbers up.
But more importantly, there is a much simpler explanation, which we get if we super-impose google mobility trends % increase in amount spent staying at home, and we shift cases by 6 days (avg time to symptoms), hospitalizations by 11 and deaths by 22 days:
We can see that the rate of growth is almost perfectly inversely correlated with the % increase in staying at home (with perhaps some light seasonal / immunity / other measures effects included).
Additionally we have non-PCR data like the Zoe symptom tracker app largely corraborating more or less the same rate of growth as detected by PCR tests: https://covid.joinzoe.com/data#levels-over-time
First, it gets thousands of people to generate various conspiracy theories and plausable looking ideas. Many of these will be interesting and have some real merit, but most of them wont. (By itself, this doesn't do much - we're people, we generate ideas and theories, many of them wrong - nothing special about this process).
Then, give people the retweet button, and have them act as a filter for the most convincing looking bullshit. This is where things go terribly wrong. Retweets are a powerful amplifier. What gets retweeted? Is it the most throughly researched content? Highly unlikely, most people aren't experts in the subject (heck, I don't even dare go into the mechanics of the theory, only the math and stats) but they will still retweet if they find it interesting, plausible and aligned with what they strongly believe or wish was true.
Side note: having some authority (Nobel prize, etc) also helps a lot with this process. Still, remember that social networks won't differentiate whether 1 scientist is actually having this idea or many, and worse, it won't necessarily inform you if many scientists are actually demonstrating how the idea is wrong (unless they make it equally interesting for a retweet)
Finally, trick people by giving them dopamine system rewards (notifications for retweets and likes). They inevitably end up surrounded by a group of supporters that praise and retweet what they say. This whole process makes it so much harder for people (even scientists) to go back on what they said if it turns out they were wrong.
So what do you end up with? You end up with a system that delivers the most perfect bullshit to the largest number of people. The scientists that tweet the same old boring conservative stuff (words like "likely" often appear) are largely voiceless.
I don't think any single person is necessarily at fault in this system. But I do think that the system is a total disaster, and we have to invent something better than this. These simplistic social network designs are failing humanity.
At the same time mainstream media has become pure propaganda. Interesting times.
To his point and yours, how does this compare to normal flu? I don’t know, I don’t see that example in your data. The problem as this highly qualified individual states it, is that they’re looking for a diagnosis and are changing the number of times amplification of the dna occurs over time.
Am I sure of his claims, no.
But either way, I appreciate hearing both view points. Having minored in bioengineering in school I can understand both arguments and make the determination myself... provided I have the info.
Bans don't solve the issue. We have to invent better social networks, where scientists can be free to theorize and be connected with their peers, without getting false awards (retweets, likes).
In general as humanity we also need to be more aware of the nature of social networks and how they don't always elevate the best content.
As for credentials, you have to realize that the mechanics of social networks are very good at finding and amplifying the few individuals with credentials that have something wrong (but interesting) to say. What you don't get to hear about are the dozens of reviews of that theory, also by experts, that show how its wrong. Because thats just not interesting for a retweet, and its kind of a downer, too.
It’s called InsiderOpinion (insideropinion.com).
That being said, the reality is that social media has over stepped as late. I’m literally being forced to “thedonald.win” or some other sh to find broadened perspectives.
Perhaps some insight as to why people have trouble believing scientists at the moment... they told everyone “masks and gloves don’t work” (to keep supply for essential workers) then told everyone “make your own masks” (btw studies going back to the early 1900s says masks will work if it’s droplet based) we also now have evidence masks don’t work. We knew this would be the case because back in January and February they confirmed it was airborne. Multiple studies since confirmed this. Basically the virus can survive in the air and our cloth masks aren’t good enough. Part of Biden’s transition team said that on Joe rogan’s podcast in March.
Then we had the whole “don’t go into crowds” followed by “oh protests are for a good cause, so it’s okay” and stuff like that.
The reality is that the public shouldn’t trust what they are being told. Literally everything has been a lie thus far.
* WHO recommended that sick people wear (surgical) masks since February at home to reduce the chances of spreading the virus in the household - if they can't fully isolate. (You can check this using archive.org - I don't have the link readily at hand, unfortunately)
* Yes, PPE was a concern but also it was thought that there is no spread without symptoms initially. Which was the main reason why masks were not recommended for the wider population:
* non-respirators don't really *protect* the wearer
* respirators like N95 do but PPE is in short suply
* if you have symptoms, wearing a mask helps, sure, but best to fully isolate as much as possible.
I'm sure there were some people and groups that connected the PPE and "masks don't work and thats why they are not recommended" but it was certainly not a blatant lie - it was really thought they wouldn't help and isolation for the infected is much better. With presymptomatic (and therefore possibly asymptomatic) spread and the realization of millions infected everywhere, everyone wearing masks became a viable option.
Aerosols are a whole other topic I won't get into (and I don't really have the expertise for it either!) but suffice to say from my understanding its not really as simple as "works" / "doesn't work".
Yes there is a problem with science failing us here as well, as the bar for evidence is simply set too high and the processes are too slow to deal with a pandemic. This is true for medications too...
Flu is really serious, which is why we have nationally co-ordinated campaigns of surveillance and monitoring, and then rapid vaccination development, and then programmes to immunise as many vulnerable people as possible.
So far, we think covid-19 is about 10 times worse than flu.
On an average year in the UK flu kills about 10,000 people. So far Covid has killed many more - at least 50,000.
But this is using an inclusive definition for "killed by flu" that will include many people, and it's using a restrictive definition for "killed by covid" that will undercount deaths.
To give an idea of how flu in the UK is counted you can look at this report: https://assets.publishing.service.gov.uk/government/uploads/...
What does this mean?
https://bitesizebio.com/24581/what-is-a-ct-value/
If you have to cycle 40 times for example, that is like finding a needle in a haystack, i.e. not understood how relevant such high precision is in terms of weather someone is actively spreading, or had it a month ago, etc.
> We can see that there is a perfect exponential 7% rate of growth starting in September and slowly decaying after. But more than that, the rate of growth in hospitalizations and deaths also follows the same pattern almost perfectly but with a 5 day and 16 day delay, respectively!
What locale is this graph from? Because it does not at all match.. let's say, the US, where there's now been 3 peaks and the number of daily cases doubled each peak, but the daily deaths was highest in the first peak. They are not growing at remotely the same rate over here, even with a 5 or 16 day delay.
If one tries comparing the US data with the data from other developed countries, what can be seen is that only the third peak in the US has the "realistic" ratio of cases to deaths, that is, one which better corresponds to the ratio from the countries which were able to handle the outbreaks better, as in, being able to detect enough cases early enough, and for that appropriate test capability, adequate to the incidence, seems to be necessary. The approximate expected ratio, worldwide, seems currently (with obviously better testing infrastructure compared to Spring, and as long as the hospitals aren't too stressed) to be 1 case of death per 100 registered "cases", after a delay, and when using some averages across a moving window of data for both both cases and deaths, e.g. 7 day average.
In the first peak in the US it was as high as almost 1 death per 14 cases(!!) -- around 32K detected daily cases with 2.2K deaths were the peaks of the averages! In the second peak in the US it was 1 death per 60 cases. Both show that there weren't enough cases detected, compared to some other countries, and that more appropriate level of testing was achieved in the US only during the third wave. But if there are some systematic delays in reporting deaths during the third wave, it can still turn out that what we see at the moment will look worse, once the wave is over. Even that effect is known to exist in the reporting of one European country: Sweden, where the systematic reporting delay of deaths during the wave gives very wrong impression of the ratio as long as the wave is raising, compared to other countries: https://news.ycombinator.com/item?id=25122196 There are some signs that such an effect can be in play in the US as well.
Additionally, what one has to be aware of is that any statistics over a big population will at different points of time not always proportionally reflect the contribution of the events in different areas -- e.g. New York. That way, some big problems for the local population will appear "less bad" when averaged, which shouldn't be comforting.
Its similar to Rt but its much easier to calculate. To get Rt, you need to estimate how many people are infected on average by one person at any point in time. This is usually done by simulating the process in various ways.
To get the daily rate of growth on the other hand, all you need to do is look at the actual numbers. Assuming you had 110 cases today and 100 cases yesterday, the current rate of growth is 10%. Since the rate calculated this way will be a bit erratic, I use some averaging and exponential smoothing to get the general trend.
The code to generate these graphs is available here: https://colab.research.google.com/drive/1QDZgdGkOzs6dP74-iED... - it uses the UK coronavirus API and google mobility trends reports directly from their respective URLs, so it should be easy to make a copy of the notebook, review/modify it and re-run it.
Why is the ratio of positive cases to hospitalizations different? I'd say this is mostly due to testing capacities being very different between now and March. In March in the UK for example it was impossible to get a test unless things are so bad that you need to get hospitalized. This is very different from now where anyone with symptoms (or just saying they have symptoms) can get tested.
Based on the graph, another reason for the different ratio between cases and deaths seems to be the period July to mid August. This was a period where the growth rate of cases didn't match the growth rate of hospitalizations and deaths at all. This likely means that parts of the population that wasn't that succeptable to severe outcomes were getting infected (which makes sense given that bars were opened up again at the beginning of July - although that is speculation)
Wrong on a spectacular level.
We empircally know he is wrong because when schools went back there was a MASSSSSSIVE spike it requested tests (as school children got their start of season colds and worried parents got them tested in the first 2 weeks of school returning).
But there was not a MASSSSSIVE spike in Covid cases.
If Covid numbers were being driven by false positives then there HAD to be a massive spike in cases as testing numbers spiked. There was not. So false positives cannot be the reason.
Johns Hopkins Newsletter deletes article re excess Covid-19 deaths --
https://twitter.com/JHUNewsLetter/status/1332100155986882562
It's written by an economist, which maybe explains why they didn't check. You won't get far in economics if you insist on checking whether your work matches reality. Ask Arthur Laffer.
(wrap in a joke tag)
Given the average age of deaths by COVID-19 being above average life expectancy, this totally seems possible to me - lots of excess deaths early in the year, but the opposite later in the year, resulting in net "deaths as usual" over the whole year.
Life expectancy at birth, and life expectancy given a particular age are very different.
Most of those Covid deaths were above life expectancy are using life expectancy at birth figures, which is very misleading (as if you make it 82 you're almost certainly a fitter than average 82 year old).
But that's not a hypothesis about the existing data, that's a guess about the future, and probably not a very good one.
This presentation masqueraded as an analysis of existing data, and the existing data shows a big rise in excess deaths. A competent analysis ought to at the very least point out this discrepancy and try to address it.
The alternatives are the author wasn't aware that excess deaths statistics are available or worse they were aware those statistics contradicted their thesis and decided to hope the audience didn't know.
Somebody else has linked one of the sites that charts US death figures. They are not subtle. In some countries you have to squint to see any impact from COVID-19, because the data is noisy and their population is small. (And maybe their reaction was effective, we can say this is the case for New Zealand but less clearly for say, Finland). But for the US it's an obvious anomaly, a prolonged excess over months.
A weirdly large number of economists seem to have taken it upon themselves to do dubious COVID work lately.
https://www.ons.gov.uk/peoplepopulationandcommunity/birthsde...
So compared to 2018 and 2019, there is no change in deaths for those aged 45 and younger.
However there is an increase in death apparent for those aged above that, such that overall deaths when comparing up to Week 46 are 9.5% higher in 2020 than 2019. However, 2019 was a very weak flu season, with no excess death periods in England:
https://www.euromomo.eu/graphs-and-maps#z-scores-by-country
If you reduce 2020 deaths by the amount that didn't occur in 2019 due to the weak flu season but did occur in 2018, then 2020's deaths are only 6.8% higher than 2018. Consider also that in those two years the population of the UK aged 45 and above increased by 1.7%.
The question is - is COVID just shuffling fundamentally unhealthy people off their mortal coil a bit faster than would have occurred naturally? How many COVID deaths are really due to ongoing societal issues like Vitamin D deficiency, air pollution, poor diets and obesity? Or simply just old age?
Is shutting down and completely disrupting society for something which is about 5% worse than a regular flu season in terms of deaths really justified? Would we have noticed without the media attention?
Why would someone who deals in facts do such a thing? Surely the correct position would be to correct and remove.
Personally anyone who refers to his material is tarnished.
On a general point, there are many people now growing in fame (and monetising that) due to the pandemic. Are their interests aligned with a pandemic ending, what do they pivot to?
Considering the OP is calling into question why this person was booted from some social media, posting lies about public hospitals and their staff during a pandemic seems a fair reason.
Shame the posts on Twitter remain and a shame Twitter themselves have not taken action.
"A source just gave me the following information. A hospital in Wales, Nevill Hall in Abergavenny, has 250 beds. 200 beds are empty. The surgeons are bored & pass much time on the nearby golf course. I think this might be one answer to why UK doctors are not investigating to find out why hospitals have no patients in them! Apparently all covid cases are currently being sent to this hospital. This story tallies with the original evidence which drew me into this ghastly mess we’ve made of our country. I’ve not mentioned this..."
What in it is a known lie?
It might be a sixth form debating point, but you have also have previously posted "The problem is that it has become acceptable to treat ideas as dangerous. For example, if I want to deny the Holocaust, that is my right."
That is where I bow out, with a link to show how such 'ideas' and 'rights' have real world impact during this pandemic https://twitter.com/dashaburns/status/1332798207911399424?s=...
Have a great day / life etc.
Here's a pretty good explainer: https://twitter.com/andrew_croxford/status/13308829944716533...
And another: https://twitter.com/andrew_croxford/status/13317220787777290...
I'm not sure of the absolute veracity of your 'regardless of symptoms' claim. As I understand it, the R number is calculated based on all the NHS tests done (and private, with some lag iirc). This even explicitly says "You cannot get a free NHS test unless you have symptoms, have been asked to by your local council, live in England and have been told to by your hospital, or are taking part in a government pilot project."
Obviously if this is the only / main input to the R number then my previous comment is crap and I retract it. Otherwise, I think I gave reasonable examples of how the sampling has been skewed, in good faith.
That's true for people seeking tests, and workplaces etc.
But the statistical sampling was conducted separately by the Office of National Statistics (not the NHS) across the whole country in order to understand the pandemic, and it did have access to tests for this purpose.
The ONS having tests doesn’t change the fact that for the vast majority of the population, being an essential worker was the only way to get a test.
The ONS tests are not the only input to the announced R number are they? Maybe I’m misunderstanding horribly, but it seems like you’re saying there’s absolutely no correlation whatsoever between availability of tests to the general public and the accuracy of the announced R number for the general public.
If I’m misunderstanding I do genuinely want to understand!
Depends what you mean by accuracy, whether that's bias or uncertainty. If you're talking about bias, then I agree with the above statement. My estimate of the mean bias in published R estimates is zero. (Possibly on a logarithmic scale :-)
But if you're talking about uncertainty and not bias, then in general more data is better provided its biases are known, but it's hard to say that focusing tests on a subset of the population reduces certainty. In a mathematical sense, to minimise uncertainty from sampling estimates if you have a fixed number of samples but a choice about which situations to assign them to, you want to focus more testing on the situations which provide the highest information content. That is not necessarily the same as spreading them evenly through the population in an unbiased manner.
I think what you may be misunderstanding, and therefore misrepresenting, is the idea that a combination of statistically sampled tests (ONS) plus biased targeted tests (NHS, key workers, Test & Trace etc) results in "skewed" or more misleading R estimates than just the statistically sampled tests (ONS) by themselves. I think that's unlikely.
I'm assuming the data is combined by competent professional statisticians.
Assuming they are competent, the likelihood of any biases due to NHS sampling that you or I might think of not having already been evaluated by the statisticians is negligible.
So, provided the bias can be estimated, combining data from multiple sources tends to reduce uncertainty rather than introducing bias in a particular direction.
That's why I say my personal "estimate of mean bias" is zero. Published R may by higher or lower than true R, and we can take it for granted it will be off by some amount and constantly (and retroactively) revised with new data, which is fair enough for estimations, but I have no basis on which to assume corrected results are more likely to have bias in one direction or the other.
A nice feature of having two or more kinds of sampling is that intentionally-randomly-sourced data (ONS) acts as an "anchor" on the interpretation of non-ONS data, allowing raw biases from various sources to be estimated and adjusted for, uncertainties to be estimated too, while at the same time trends (such as over time) remain trackable with the higher statistical power that comes from larger numbers of samples.
In other words, a bit of the best of both worlds. And since r and R are trend parameters, that's quite helpful.
>I think what you may be misunderstanding, and therefore misrepresenting, is the idea that a combination of statistically sampled tests (ONS) plus biased targeted tests (NHS, key workers, Test & Trace etc) results in "skewed" or more misleading R estimates than just the statistically sampled tests (ONS) by themselves.
Not quite, and I apologise if I've communicated poorly. Totally agree that multiple sampling methods lead to more convincing results. I avoid optimising until I've profiled with both a sampling AND an instrumenting profiler, for example.
What I was trying (failing!) to get across is that I think that if the distribution of 'generally available' tests is sufficiently non-representative, then the statistically sampled test model potentially couldn't correct for it enough for the derived estimate to be reliable.
Obviously the responsible subgroup at SAGE will have tried to weight accordingly, and the people running the individual studies will as you say, be much more skilled at accounting for potential bias than I am. I'm absolutely not saying that I've stumbled on some insight that they missed!
Here's where I struggle - the pool of generally available tests at the time fall into two relatively narrow groups: patients exhibiting symptoms bad enough to require hospitalisation, and people at high risk of transmission. The random longitudinal studies don't have that problem of course and should be used to correct for the bias. I realise there are methods for quantifying bias and uncertainty, but isn't this more at risk of inaccuracy than simply having a higher N? N matters, surely? The higher confidence we have that any one set of data is representative, the higher confidence we can have in the final derived estimate.
I just can't see any angle from which having tested more people we wouldn't have improved our ability to estimate R.
You’re not saying true things. Why are you spreading this disinformation despite everyone correcting you? They’ve been randomly sampling. R values are randomly sampled.
So. Are you saying that the R number given by the government is derived from data only from random sampling, and that other test statistics do not feed into the estimate at all?
I’m not a statistician or epidemiologist and I find that very hard to believe. But I will honestly be happy to have learned something if you can teach me otherwise.
My understanding was that longitudinal studies using random sampling were used to check and correct the numbers derived from the actual test numbers for the real population. That meant the longitudinal studies would have been even more important early on when access to testing was poor, because the sample distribution was obviously skewed wrt the whole population.
Again, I really would appreciate a substantial rebuttal, link or otherwise.
Thanks
https://www.uchealth.org/today/the-truth-about-asymptomatic-...
Lack of it? But isn’t that exactly how R is estimated?
Hard to tell this page doesn't really give details on how large the longitudinal and survey is conducted.
https://www.gov.uk/guidance/the-r-number-in-the-uk#about-r-n...