Public Release of Newest Imperial College Report for the UK Delayed By 'Politicized Nature' of Lockdown Debate (ft.com) 37
"The publication of a long-awaited report from Imperial College London that models the impact of coming out of lockdown has been delayed for several weeks, following criticism of the team's methods, as the debate around the UK's coronavirus restrictions has become increasingly politicised," reports the Financial Times:
The report has yet to be released, although its findings have been shared with government, according to two people associated with the Imperial team. The delay comes as the rightwing press and some Conservative politicians question the need for such stringent lockdown measures in the UK. A number of Tory figures, including former minister David Davis and Eurosceptic MP Steve Baker, have cast doubt on the Imperial team. They accuse the scientists of using an outdated computer code in an influential March report that predicted the UK could suffer 500,000 deaths during the pandemic if the government failed to take action. The Telegraph newspaper suggested last week that Imperial's modelling could be "the most devastating software mistake of all time..."
A senior member of the team told the Financial Times, "Given the increasingly politicised nature of debate around the science of Covid-19, we have decided to prioritise submitting this research for publication in a peer-reviewed scientific journal and will release it publicly at that time...."
Sir Venki Ramakrishnan, president of the Royal Society, Britain's senior scientific body, added that the public had a false impression that the Imperial model dominated government decision-making in mid-March, when ministers decided to impose a lockdown. "Theirs was not the only model considered," he said, "and we didn't need a model to know what would happen when this highly infectious virus arrived. Italy's healthcare system had already reached its limits and they were begging us to act."
A senior member of the team told the Financial Times, "Given the increasingly politicised nature of debate around the science of Covid-19, we have decided to prioritise submitting this research for publication in a peer-reviewed scientific journal and will release it publicly at that time...."
Sir Venki Ramakrishnan, president of the Royal Society, Britain's senior scientific body, added that the public had a false impression that the Imperial model dominated government decision-making in mid-March, when ministers decided to impose a lockdown. "Theirs was not the only model considered," he said, "and we didn't need a model to know what would happen when this highly infectious virus arrived. Italy's healthcare system had already reached its limits and they were begging us to act."
Proper science = politicised? (Score:3, Insightful)
A senior member of the team told the Financial Times, "Given the increasingly politicised nature of debate around the science of Covid-19, we have decided to prioritise submitting this research for publication in a peer-reviewed scientific journal and will release it publicly at that time...."
Their junk code and methodology were only reviewed after they massively impacted the world. At best that was scientific malpractice.
If a "politicized debate" leads to them finally doing peer reviews that sounds like a good thing, doesn't it?
Re:Proper science = politicised? (Score:5, Informative)
As a reminder, the original report (which was panned by other researchers even when it came out [necsi.edu]) predicted that the US would have 2,2 million deaths (half that if the US employs an aggressive quarantine, social distance, testing and isolation programme), "not accounting for the potential negative effects of health systems being overwhelmed on mortality", and that with no curve flattening, the US wouldn't peak until the start of July, while with control measures to flatten the curve the peak wouldn't be until the end of the year.
This hilariously wrong, widely shared piece of comedy was brought to you by Neil Ferguson, the guy who predicted up to 50k UK deaths from BSE (150k if it spread to sheep) (actual = ~200 deaths), 200 million people would die globally from bird flu in 2005 (actual = a couple hundred globally), and that swine flu had a mortality rate of 0,3-1,5% and would kill 65k people in the UK (actual = 457 deaths in the UK, 0,026% mortality).
He was also recently forced to resign for breaking the lockdown rules that he himself designed in order to sleep with a married woman [telegraph.co.uk].
The guy has to be the worst epidemiologist of our generation.
Re:Proper science = politicised? (Score:5, Insightful)
This is exactly what they are getting at. The original report didn't say there would be 2.2 million deaths, it said that if certain assumptions were true that their model suggested that could happen.
This not so subtle distinction was ignored and the report instantly politicised. Seems they learnt their lesson and are trying to rewrite this one too be less vulnerable to this type of attack.
Re: (Score:2)
Part of the problem there was that these predictions were essentially "worst-case". Why did he not also provide best-case and intermediate predictions as well? As a result, the "politicisation" is mostly his fault. These "scientific" predictions are little more than scaremongering if they aren't counterbalanced by additional scenarios to give a complete picture of the situation.
Re: (Score:2)
The same reason the Treasury, which always releases 3 reports at a time - expected case, along with worst and best cases, suddenly started releasing only 2 reports during the Brexit referendum debate. You can guess which two, one obviously was their expected case.
Like polls, nobody wants a report that doesn't support their already-conceived views.
Re: (Score:2)
It's the Y2K bug all over again - people think that because nothing happened, we wasted all the money trying to ensure things would continue as normal.
The truth is everyone accepted those worst case figures and made restrictions so it wouldn't happen. The fact that the worst didn't happen is because we took acti
Re: (Score:2)
This hilariously wrong
How do you know? You're claiming a scenario that was actively mitigated against with none of the assumptions achieved had estimated the number incorrectly. Now if you actually go into the report and compared the CI_HQ_SD figures, they actually trend against the UK results quite well. So given that part of the report is true, what evidence or basis do you have that the scenario which the USA actively prevent wouldn't have occurred?
Re: (Score:1)
These models don't make "predictions". That's what you science-deniers just don't get. The models make "projections". Assuming x and y, then z will happen. That's not a prediction. They're not saying "this will definitely happen". There are a huge number of assumptions baked in and they run the models under multiple different scenarios exploring how changes to the assumptions affect changes to projected outcomes.
Further, epidemiological models are fundamentally extremely difficult to build because at
Re:Proper science = politicised? (Score:4, Interesting)
Their junk code and methodology were only reviewed after they massively impacted the world. At best that was scientific malpractice.
If Western countries (and the WHO) had been paying attention to the reports coming out of Imperial College early this year then they would have prevented covid-19 from spreading to their countries. Getting the exact numbers right when modelling exponential growth is difficult to impossible. The message from Imperial College was clear and accurate: without taking measures to control it, this virus spreads and kills rapidly. TAKE THIS SERIOUSLY! ACT NOW!
The problem is not that it's hard to impossible to make exact numerical predictions of exponential growth, the problem is that the predictions of exponential growth were ignored by most countries in the West and by the WHO. Those countries that did take the virus, and its exponential growth, seriously were able to contain it: South Korea, Taiwan, Chine (eventually), Hong Kong, New Zealand, Australia. This one graph is the best way I've found to track how countries (and now US states) are doing in their fight against covid-19: Covid Trends [aatishb.com].
This pandemic in Western countries is due to our leaders either ignoring or not understanding the science and math including that coming out of Imperial College. If you want to keep abreast of the covid-19 situation I highly recommend these two YouTube channels: Dr. John Campbell [youtube.com] (he is a professor of Nursing, not an MD) and MedCram [youtube.com]. If our leaders or their advisors had been paying attention to these channels then they would have averted this disaster. Much of their early prediction were based on the studies coming out of Imperial College. These channels are still mostly ignored. If our leaders or news media were paying attention to them then everyone would be taking Vitamin D and the death rate due to covid-19 would almost certainly go down drastically.
There are no peer reviewed double blind studies of the efficacy of Vitamin D in treating covid-19 but the amount of circumstantial evidence is massive. In one study of covid-19 patients in Malaysia, the death rate among people who had enough Vitamin D was 8 to 10 times lower than the death rate of people who were low on Vitamin D.
Since Vitamin D was already known to be extremely effective in preventing and minimizing most respiratory infections, Dr. John Campbell was recommending it before any of the papers about Vitamin D and covid-19 came out. Here is a recent clip of Dr. John Campbell explaining the importance of Vitamin D: Vitamin D on the news [youtube.com].
Waiting for a peer reviewed double blind study before telling people to take Vitamin D is nuts! Likewise, waiting for these reports coming from Imperial College to be peer reviewed is nuts. The covid-19 pandemic is very serious. All 50 states of the US are starting to open up. For states that have an R naught greater than one, opening up could be disastrous. We should be able to see the paper before it is peer reviewed. Trying to silence or delay the information coming out of Imperial College is going full speed ahead in the wrong direction. We should have been listening to their warnings back in January, February, March, and April. We should also listen to their warnings in May.
Re: (Score:3)
Re:Proper science = politicised? (Score:5, Informative)
Their junk code and methodology were only reviewed after they massively impacted the world. At best that was scientific malpractice.
There code is fairly typical academic code. Better than lots of such code I've seen. From a quick review of their bug tracker and patch tracker, the bugs weren't such that they would have significant impact on the results. The complaints about bugs and code quality were false distractions.
As one person who has reviewed the code put it
To be honest I've seen much less maintainable code at blue chips.
Feel free to review the code yourself.
https://github.com/mrc-ide/cov... [github.com]
I also asked in the bug tracker if anyone is aware of any bugs that could have significantly impacted their results, and the answer is no. The output after various refactorings and bug fixes is stochastically the same (previous versions were non deterministic, so could give slightly different results, etc.)
https://github.com/mrc-ide/cov... [github.com]
Re: (Score:2)
The issue log [github.com] is also full of the same sort of unfocused B.S. idiots being idiots, who when challenged about cannot come up with specifics, other than claiming to be experts.
I've taken a look at the code from the perspective of 30 years of developer and while the code is not exemplar, it is certainly not junk code.
https://github.com/mrc-ide/cov... [github.com]
Re: (Score:2)
IIRC it wasn't about maintenance, it was about repeatability.
Nobody coudl runt he program and get a similar answer, so they couldn't tweak the assumptions or variables in the model to get any kind of answer.
eg, if you rtun it twice and it returns 10,000 deaths, you might think "lets see what happens if we change the variable to get people staying indoors", and run it again with that change and it says 8,000 deaths. You'd say that was a valid result.
But if you ran the original and it says 5,000 deaths. You h
Bullshit (Score:2)
You are the one you are the one trying to politicised the science.
You claim the code is junk, provide some specific examples, then put up or shut up, raise specific concrete issue on github and link it here.
Be part of the solution, not part of the problem.
Re: (Score:2)
Well, the summary shows the problem:
there is no "right wing press" in the UK, even the right-leaning media is left/liberal these days. Even the much maligned Daily Mail has new leadership who appears to worship at the alter of woke liberalism.
Everything is now politicised and by that I mean tribal. Nobody cares about the message anymore, just who delivered it and whether it supports their side or the other one. Jesus himself could make his second coming today and he'd
Re: (Score:2)
"All models are wrong but some are useful" – George Box.
Re: (Score:2)
Relying on Z-scores is problematic in its own way, because it includes "collateral damage", such as suicide, people not seeking care for other conditions (a more serious issue), etc (though offset by, say, reduced traffic accidents). It's "a" useful tool, but not a substitute for actual data on the disease. Official UK numbers (without collateral damage) are 36675, 27432 thereof in England (annual England flu deaths [fullfact.org] = 1692 ('18-'19) to 28330 ('14-'15)).
England is an interesting case in terms of excessive
Re: (Score:2)
The reason I particularly worry that the UK's excess deaths are lockdown collateral damage is the age distribution. COVID-19 deaths are highly biased toward the elderly. England's excess deaths among the 65+ age group is high, but the really striking, abnormal difference in excess deaths is in the 15-64 age group. Z-scores are actually as high in the 15-64 age group as in the 65-74 age group, which is really unusual.
Here's a graph between England vs. the rest of the UK for age 65+: link [voxeu.org]
Now here's one for
Re: (Score:1)
Sorry, what the fuck are those pictures meant to be showing? 30 million people in England having sex each week?
Link to the actual fucking source, not some shitty fucking graph that could be interpreted as almost anything.
Re: (Score:2)
The source was linked in the original post [voxeu.org], in the part talking about England's abnormal Z-scores (2nd paragraph). They're graphs of Z-scores for different age cohorts, by week, for the UK's different regions.
If one were dealing with simply a higher infection rate in England, one would expect the ratios between Z scores "England vs. Others" to be roughly the same for each cohort. But they're not - the ratio vs. other areas is dramatically higher for the 20-64 cohort than for other cohorts - during their l
Re: (Score:2)
Fuck me, you actually posted graphs that plotted Z Scores and didn't bother to tell anybody that's what they were, or indeed what the fuck a Z score even is?
https://www.euromomo.eu/how-it... [euromomo.eu] needs a fucking A level in maths just to read.
In English it's purporting to show how much the current figures vary relative to the standard amount by which figures vary. This is actually not useful at all as it fails to take into account any variance in volatility between the different nations being compared; does Engla
Re: (Score:2)
There's a possibility deatsh are under reported for political purposes in some places. England is usually very good for accurate reporting, but I'm not sure I'd trust the Scottish to release anything properly these days, and the Welsh NHS to tie its shoelaces.
However, those graphs don't show the axes - are they deaths or percentages or what? If absolute figures, then sure - England has ten times the population of the other regions, you'd expect higher.
For actual deaths, the numbers are, according to Public [england.nhs.uk]
Re: (Score:2)
England's excess deaths among the 65+ age group is high, but the really striking, abnormal difference in excess deaths is in the 15-64 age group. Z-scores are actually as high in the 15-64 age group as in the 65-74 age group, which is really unusual.
This is because those 65 and above tended to die in care homes without getting counted in the official statistics at all. Those under 65 tended to get a chance in the hospital before dying, and so they had a better chance of being counted.
Re: (Score:2)
There's no real reason to think that England is so massively undercounting COVID deaths (everyone tested on admission with COVID-like symptoms)
The official message is that everyone was tested on admission with COVID-like symptoms. This was not the case until very recently, and there is doubt whether it is even the case now. The UK is massively undercounting COVID-19 deaths.
Re: (Score:2)
So far we have had around 50,000 excess deaths over normal in the UK.
Given that many of those will be due to the lockdown and other factors, and not to the pandemic itself, that's an incorrect number to use.
For example fewer than normal people died from the flu last year, resulting in abnormally high numbers of vulnerable people this year. People aren't attending hospital at the rates they usually would. People dying from non-Coronavirus causes are being counted as a Coronavirus death, even if they caught it only after being admitted to hospital with a life-ending issue.
Unde
Re: (Score:2)
The control is that in those places with very low COVID-19 deaths (and reliable statistics) due to effective lockdowns, mortality is significantly below the average year.
This shows that people do not die in vast numbers due to lockdown alone.
So many lessons.. (Score:3)
The real key lesson this also relates to is the use of predictions.. I majored in both math and comp sci, and was floored by all the use of "exponential" functions applied to a physical system. Math works like that, real life doesn't. South Park actually did a perfect parody of this situation years ago (Two Days Before the Day After Tomorrow [wikipedia.org]), and it was a news reporter saying something like "our model predicts Chicago will have over 6 trillion deaths by next month".
Those are often too complex of subjects for people to truly understand, so what I've been trying to impress is for people to look back at all the predictions made, and understand they were all wrong. It doesn't matter the reasons, causes, how we affected it, etc., bottom line is many people got very scared over a future that did not materialize, which as a lesson almost always holds true.
Re: (Score:2)
I've lost a lot of respect for epidemiology as a field the more I've looked into the sort of models that pass for "scientific". A lot of fitting things to gaussians. Zero feedback mechanisms between infection levels and human behavior adjustment. All people considered equally likely to become infected and equally likely to infect others, which is obviously incredibly wrong. Numerous models treating the ratio of cases to infections as constant, and only adding mere footnotes that this ratio may change (a
Re: (Score:2)
The science has been awfully wrong in a lot of cases on COVID-19 unfortunately. The fact that many politicians have been even worse does not absolve the scientific community from guilt.
Most of the scientific advice focused on "flattening the curve". We are still hearing that from various experts -- saying that the same number of people will get infected, just spread out over more time, and that it is impossible to do anything about that.
Yet we know it is untrue, we have known it is untrue at least since Sou
Who to believe... politicians or scientists? (Score:2)
A hard question?
Then again...
Flat earthers
Anti-vaxxers
5G
Fake moon landings
CIA hit of Kennedy
Ok, that last one is probably right...
Re: (Score:2)
A hard question?
Then again...
Flat earthers Anti-vaxxers 5G Fake moon landings CIA hit of Kennedy
Ok, that last one is probably right...
Yeah - and you believe the others are wrong. Because.
Re: (Score:2)
The actuaries, they are the experts. Epidemiologists have an opinion, but they aren't betting money on it.
Re: (Score:2)
Tricky question really.... too many scientists are now political. Not just Ferguson and his crappy code, trying to keep his cushy government-funding going, but how about Michael "hockey stick" Mann who refused to show his data to a law court.
Or how about Andrew Wakefield, the now ex-doctor, who told the world that MMR vaccine would turn children into brain-dead zombies and that, coincidentally, had a better vaccine that you should use. Or the Lancet a "respected "publiciation that only kept his original, bu