5 ms·

Has the modeling been wrong? Or have all the measures you wrote about above worked and prevented the "worst case scenarios" of the models?

by edanm 5y ago

zpeti 5y ago
https://www.telegraph.co.uk/opinion/2021/12/19/tackled-sage-covid-modeller-twitter-quite-revelation/ https://www.telegraph.co.uk/opinion/2021/12/19/tackled-sage-...
alvah 5y ago
The modelling has been wrong. Early on someone ran Sweden's numbers through Neil Ferguson's model's business-as-usual scenario, and the model's estimates of cases and deaths were much higher than observed reality (by around an order of magnitude IIRC).
- noahtallen 5y ago
  I’m a bit confused by this example. Are you saying that the model showed higher cases in a non-lockdown analysis compared to fewer cases in the real world with lockdowns? If so, wouldn’t that be expected? If not, what does “business-as-usual” mean?
  - thinkingemote 5y ago
    In this example, Sweden had no lockdown.
    - noahtallen 5y ago
      Cool, thanks for elaborating.
- Gibbon1 5y ago
  I've been finding it somewhat telling that people that make the argument you are making presume that no one else has any agency. And won't take precautions without the government forcing them to.
- _djo_ 5y ago
  That's false, Imperial College's modelling of Sweden actually matched real-world deaths fairly closely. For instance, it assessed Swedish deaths to 31 March 2020 as being 240 [140-440] with zero interventions, or 160 [110-240] with slight interventions. It estimated deaths up to 28 March 2020 as 89 [61-120]. Actual deaths to 28 March 2020 were 92. What someone else did was to take the Imperial College model and run it against a bunch of populations without properly adjusting the parameters to account for factors like population density, age, etc. They had no real idea how the model worked, but then claimed that it was wrong. I think it's clear the most of the COVID-19 epidemiology models created around the world were accurate enough in their goal, which was to test the relative effectiveness of interventions and give broad predictions of relative impact. None were perfect, but then no model like this whether epidemiological, economic, or similar is ever going to be 100% accurate.
benjaminwootton 5y ago
Well we obviously have no control group for this question, but the predictions have been off the charts wrong in the past whilst the impact of the restrictions appear to have been overstated. Of course this is a subjective view, but this is almost becoming the accepted narrative in the UK. As of today, we are not getting further restrictions because SAGE and the modelling has been discredited. They are screaming for more restrictions and escalating to talk about millions of cases per day, and our politicians are (rightly, in my opinion) putting them back in their boxes. The more interesting point is that this time we may actually have data before restrictions. When cases begin to fall without further restrictions even with the more transmissible omicron, I am hoping it will be the final nail in the coffin for credibility of the modellers.
- maxehmookau 5y ago
  > this is almost becoming the accepted narrative in the UK "Accepted narrative" != "Anything resembling the truth". > restrictions because SAGE and the modelling has been discredited. They are screaming for more restrictions and escalating to talk about millions of cases per day, and our politicians are (rightly, in my opinion) putting them back in their boxes. This narrative of "policians vs. scientists" is gaining traction here and I hate it. It's unhelpful and incorrect. It's reductive, but who would you rather have making decisions about our country in a pandemic? Chris Whitty, or Boris Johnson? I know who I'd pick. > I am hoping it will be the final nail in the coffin for credibility of the modellers. Then what? Who do we put our faith in? Politicians with no scientific background who make decisions based on political convienience?
  - WickyNilliams 5y ago
    Furthermore, I would not say it is an accepted narrative among people I know. Anecdotal of course, but it is definitely not universal. Though the UK media has a particular talent for unifying to push a message that suits their ideological leanings, which can often give a distorted view of public mood
brigandish 5y ago
The Spectator has a page[1] showing the actual outcomes vs the models, and a worrying exchange[2] with the chairman of the Sage Covid modelling committee. You can judge for yourself. [1] https://data.spectator.co.uk/category/sage-scenarios https://data.spectator.co.uk/category/sage-scenarios [2] https://www.spectator.co.uk/article/my-twitter-conversation-with-the-chairman-of-the-sage-covid-modelling-committee https://www.spectator.co.uk/article/my-twitter-conversation-...
- maxehmookau 5y ago
  The spectator has, however, had a strong anti-lockdown skew (I avoid the word bias.) since the start which is worth bearing in mind when viewing any statistics that they put out. Not saying that they're wrong, but their general libertarian-right stance is worth noting.
  - brigandish 5y ago
    > Not saying that they're wrong Then I fail to see how the rest of what you wrote is relevant to what I posted. How is it relevant?
    - benjaminwootton 5y ago
      Especially when they simply post official data without commentary. The chart on the linked page "Covid hospitalisation vs Sage scenarios published 14 Sep 2021" is particularly damming. The modellers were calling our government all the names under the sun at the time of reopening, predicting almost certain doom. We then removed all restrictions and absolutely nothing happened. It seems like a UK phenomenon, but bad modelling has been at the heart of the crisis here. I'm glad we are finally asking the right questions.
      - maxehmookau 5y ago
        > bad modelling has been at the heart of the crisis here I would suggest that its probably the novel respiritory disease at the heart of the crisis here. Jokes aside, I don't see what the alternative is? If better modelling exists, we would've used it. The alternative was much worse.

Has the modeling been wrong? Or have all the measures you wrote about above worked and prevented the "worst case scenarios" of the models?

by edanm 5y ago

Has the modeling been wrong? Or have all the measures you wrote about above worked and prevented the "worst case scenarios" of the models?

zpeti 5y ago

https://www.telegraph.co.uk/opinion/2021/12/19/tackled-sage-covid-modeller-twitter-quite-revelation/ https://www.telegraph.co.uk/opinion/2021/12/19/tackled-sage-...

alvah 5y ago

The modelling has been wrong. Early on someone ran Sweden's numbers through Neil Ferguson's model's business-as-usual scenario, and the model's estimates of cases and deaths were much higher than observed reality (by around an order of magnitude IIRC).

noahtallen 5y ago
I’m a bit confused by this example. Are you saying that the model showed higher cases in a non-lockdown analysis compared to fewer cases in the real world with lockdowns? If so, wouldn’t that be expected? If not, what does “business-as-usual” mean?
- thinkingemote 5y ago
  In this example, Sweden had no lockdown.
  - noahtallen 5y ago
    Cool, thanks for elaborating.
Gibbon1 5y ago
I've been finding it somewhat telling that people that make the argument you are making presume that no one else has any agency. And won't take precautions without the government forcing them to.
_djo_ 5y ago
That's false, Imperial College's modelling of Sweden actually matched real-world deaths fairly closely. For instance, it assessed Swedish deaths to 31 March 2020 as being 240 [140-440] with zero interventions, or 160 [110-240] with slight interventions. It estimated deaths up to 28 March 2020 as 89 [61-120]. Actual deaths to 28 March 2020 were 92. What someone else did was to take the Imperial College model and run it against a bunch of populations without properly adjusting the parameters to account for factors like population density, age, etc. They had no real idea how the model worked, but then claimed that it was wrong. I think it's clear the most of the COVID-19 epidemiology models created around the world were accurate enough in their goal, which was to test the relative effectiveness of interventions and give broad predictions of relative impact. None were perfect, but then no model like this whether epidemiological, economic, or similar is ever going to be 100% accurate.

benjaminwootton 5y ago

Well we obviously have no control group for this question, but the predictions have been off the charts wrong in the past whilst the impact of the restrictions appear to have been overstated. Of course this is a subjective view, but this is almost becoming the accepted narrative in the UK. As of today, we are not getting further restrictions because SAGE and the modelling has been discredited. They are screaming for more restrictions and escalating to talk about millions of cases per day, and our politicians are (rightly, in my opinion) putting them back in their boxes. The more interesting point is that this time we may actually have data before restrictions. When cases begin to fall without further restrictions even with the more transmissible omicron, I am hoping it will be the final nail in the coffin for credibility of the modellers.

maxehmookau 5y ago
> this is almost becoming the accepted narrative in the UK "Accepted narrative" != "Anything resembling the truth". > restrictions because SAGE and the modelling has been discredited. They are screaming for more restrictions and escalating to talk about millions of cases per day, and our politicians are (rightly, in my opinion) putting them back in their boxes. This narrative of "policians vs. scientists" is gaining traction here and I hate it. It's unhelpful and incorrect. It's reductive, but who would you rather have making decisions about our country in a pandemic? Chris Whitty, or Boris Johnson? I know who I'd pick. > I am hoping it will be the final nail in the coffin for credibility of the modellers. Then what? Who do we put our faith in? Politicians with no scientific background who make decisions based on political convienience?
- WickyNilliams 5y ago
  Furthermore, I would not say it is an accepted narrative among people I know. Anecdotal of course, but it is definitely not universal. Though the UK media has a particular talent for unifying to push a message that suits their ideological leanings, which can often give a distorted view of public mood

brigandish 5y ago

The Spectator has a page[1] showing the actual outcomes vs the models, and a worrying exchange[2] with the chairman of the Sage Covid modelling committee. You can judge for yourself. [1] https://data.spectator.co.uk/category/sage-scenarios https://data.spectator.co.uk/category/sage-scenarios [2] https://www.spectator.co.uk/article/my-twitter-conversation-with-the-chairman-of-the-sage-covid-modelling-committee https://www.spectator.co.uk/article/my-twitter-conversation-...

maxehmookau 5y ago
The spectator has, however, had a strong anti-lockdown skew (I avoid the word bias.) since the start which is worth bearing in mind when viewing any statistics that they put out. Not saying that they're wrong, but their general libertarian-right stance is worth noting.
- brigandish 5y ago
  > Not saying that they're wrong Then I fail to see how the rest of what you wrote is relevant to what I posted. How is it relevant?
  - benjaminwootton 5y ago
    Especially when they simply post official data without commentary. The chart on the linked page "Covid hospitalisation vs Sage scenarios published 14 Sep 2021" is particularly damming. The modellers were calling our government all the names under the sun at the time of reopening, predicting almost certain doom. We then removed all restrictions and absolutely nothing happened. It seems like a UK phenomenon, but bad modelling has been at the heart of the crisis here. I'm glad we are finally asking the right questions.
    - maxehmookau 5y ago
      > bad modelling has been at the heart of the crisis here I would suggest that its probably the novel respiritory disease at the heart of the crisis here. Jokes aside, I don't see what the alternative is? If better modelling exists, we would've used it. The alternative was much worse.