Discussions about serious topics, for serious people
-
Woodchopper
- Princess POW
- Posts: 7317
- Joined: Sat Oct 12, 2019 9:05 am
Post
by Woodchopper » Thu Apr 09, 2020 12:31 pm
PeteB wrote: ↑Thu Apr 09, 2020 11:51 am
That seems bonkers - 13,900 (8,380 - 16,200) UK deaths week begining 5/4/20 - just had a quick look through their methodology - surely there is a step change starting with the lockdown and this will start feeding through into the death figures probably this week/next week (if we'd not had the lockdown then people that would have been infected at the begining of the lockdown would be dying now) - is this just a 'curve fitting' exercise ?
The analysis is pretty sophisticated in that the curves are based upon data up to 4 April, and they have very different curves for each country. So they are taking control measures into account in other countries.
It looks like they are assuming that the UK curve will continue on a similar exponential track to what it has been for the last few weeks.
The UK lockdown was two and a half weeks ago. Given that the incubation period is up to two weeks, someone who was infected on the day the lockdown was announced could started showing symptoms on Monday this week. Obviously some will be earlier.
Elsewhere it seemed that there was a 3-4 week gap between the lockdown and it having a noticeable effect on mortality. If that's the case we wouldn't expect to see the UK lockdown reflected in the mortality figures until next week.
Anyway, I don't know whether its correct.
We'll see over the next few days.
-
Woodchopper
- Princess POW
- Posts: 7317
- Joined: Sat Oct 12, 2019 9:05 am
Post
by Woodchopper » Thu Apr 09, 2020 12:37 pm
lpm wrote: ↑Thu Apr 09, 2020 12:05 pm
Up to 4 April there were 4,934 UK fatalities.
This report forecasts 13,900 in the week commencing 5 April (range 8,380 to 16,200).
So his own report says 18,000 by the end of this Easter weekend (range 13,400 to 21,100).
Yet he's on TV and radio and newspapers with 10,000 to 20,000 figures
for the total first wave. Something has been f.cked up - must be our understanding, surely? He made headlines across the country with his forecasts, the public understanding is to expect something like 20,000 due to him.
What are we missing?
Possible explanations:
a) He is listed as an author along with many others. Perhaps these aren't the estimates that he believes in.
b) His rational mind accepts the estimates but at a certain deep level he can't believe that its true.
c) He believes its true but has chosen not to repeat the estimates as he is afraid doing so would result in even worse consequences (eg public panic).
-
lpm
- Junior Mod
- Posts: 6361
- Joined: Mon Nov 11, 2019 1:05 pm
Post
by lpm » Thu Apr 09, 2020 12:40 pm
PeteB wrote: ↑Thu Apr 09, 2020 11:51 am
That seems bonkers - 13,900 (8,380 - 16,200) UK deaths week begining 5/4/20
First 3 days
5 Apr 439
6 Apr 786
7 Apr 938
leaves 11,737 in four days to achieve the
target forecast.
Say:
8 Apr 1,500
9 Apr 2,200
10 Apr 3,200
11 Apr 4,800
It'll be well over 1,000 today - but can't possibly reach 4,800 a day by this weekend. Surely?
It must be explanation (a) - He is listed as an author along with many others. Perhaps these aren't the estimates that he believes in. He thinks his colleagues have f.cked up.
Awarded gold star 4 November 2021
-
Bird on a Fire
- Princess POW
- Posts: 10142
- Joined: Fri Oct 11, 2019 5:05 pm
- Location: Portugal
Post
by Bird on a Fire » Thu Apr 09, 2020 12:48 pm
Some of these guys are going to see exponential growth in their citation rates over the next year or so, which is great news for their careers and funding prospects.
I'd probably stick my name on that covid report if they asked me, even without reading it, let alone believing it.
We have the right to a clean, healthy, sustainable environment.
-
shpalman
- Princess POW
- Posts: 8428
- Joined: Mon Nov 11, 2019 12:53 pm
- Location: One step beyond
-
Contact:
Post
by shpalman » Thu Apr 09, 2020 1:00 pm
shpalman wrote: ↑Thu Apr 09, 2020 12:25 pm
Seems to me anyway that the UK could follow the same curve as Germany and Italy because why not? Although the UK is already falling off the curve slightly (maybe because testing can't keep up) and may flatten a bit but cross the other curves as they come down and the UK's one doesn't.
Covid Trends
... so the
deaths should be peaking and then coming down soon if you're not totally f.cked.
Assuming the same sorts of reporting problems going on in both Italy and the UK.
having that swing is a necessary but not sufficient condition for it meaning a thing
@shpalman@mastodon.me.uk
-
Woodchopper
- Princess POW
- Posts: 7317
- Joined: Sat Oct 12, 2019 9:05 am
Post
by Woodchopper » Thu Apr 09, 2020 1:09 pm
lpm wrote: ↑Thu Apr 09, 2020 12:40 pm
PeteB wrote: ↑Thu Apr 09, 2020 11:51 am
That seems bonkers - 13,900 (8,380 - 16,200) UK deaths week begining 5/4/20
First 3 days
5 Apr 439
6 Apr 786
7 Apr 938
leaves 11,737 in four days to achieve the
target forecast.
Say:
8 Apr 1,500
9 Apr 2,200
10 Apr 3,200
11 Apr 4,800
It'll be well over 1,000 today - but can't possibly reach 4,800 a day by this weekend. Surely?
It must be explanation (a) - He is listed as an author along with many others. Perhaps these aren't the estimates that he believes in. He thinks his colleagues have f.cked up.
The forecast is from the 6th, which allows an extra day.
-
PeteB
- Clardic Fug
- Posts: 205
- Joined: Wed Nov 13, 2019 1:02 pm
Post
by PeteB » Thu Apr 09, 2020 2:42 pm
-
lpm
- Junior Mod
- Posts: 6361
- Joined: Mon Nov 11, 2019 1:05 pm
Post
by lpm » Thu Apr 09, 2020 3:24 pm
Lol
This is the [Imperial College] forecast for the UK for this week again (pink plume again, below). The data were already outside their range by yesterday. What on earth were they thinking?
Why didn't Imperial College sense-check their results? How come amateurs can take a five minute look and think their results are bonkers?
Awarded gold star 4 November 2021
-
shpalman
- Princess POW
- Posts: 8428
- Joined: Mon Nov 11, 2019 12:53 pm
- Location: One step beyond
-
Contact:
Post
by shpalman » Thu Apr 09, 2020 3:29 pm
lpm wrote: ↑Thu Apr 09, 2020 3:24 pm
Lol
This is the [Imperial College] forecast for the UK for this week again (pink plume again, below). The data were already outside their range by yesterday. What on earth were they thinking?
Why didn't Imperial College sense-check their results? How come amateurs can take a five minute look and think their results are bonkers?
Because their model is too complicated and nobody can remember how it works?
having that swing is a necessary but not sufficient condition for it meaning a thing
@shpalman@mastodon.me.uk
-
EACLucifer
- Stummy Beige
- Posts: 4177
- Joined: Fri Dec 13, 2019 7:49 am
- Location: In Sumerian Haze
Post
by EACLucifer » Thu Apr 09, 2020 3:30 pm
PeteB wrote: ↑Thu Apr 09, 2020 12:22 pm
Interesting
twitter thread
Our failure, on the Helmand campaign, was not (as the media always assumed) that we were ignoring or overruling military advice. It was that we failed to challenge it, to interrogate it enough, to expose the differences within the expert community and have a proper debate.
This seems pretty astute. There seems to have been no understanding that one does not simply listen to one or two experts and get a complete picture.
One does not need to be an expert to ask the following questions - "how up to date is this model?" and "how have you verified the accuracy of this model?"
Nor does one have to be an expert to ask a range of experts to ask those questions of every simulation presented to politicians
-
Sciolus
- Dorkwood
- Posts: 1380
- Joined: Mon Nov 11, 2019 6:42 pm
Post
by Sciolus » Thu Apr 09, 2020 3:41 pm
lpm wrote: ↑Thu Apr 09, 2020 7:20 am
Also, the virus becomes less lethal and transmittable with time. A wave next year is less damaging than the current wave. Not because of any change to the RNA but because of change to the human hosts - far more hospital beds, thousands of ventilators, nurses with PPE. And better techniques, experience of which drug combinations work and millions of tests. The virus becomes less infectious because 2 metres becomes second nature.
What's infuriating about the govt approach is they got obsessed with 2nd waves so deliberately ran the first one hotter, while failing to prepare the equipment etc. They missed the fundamental fact that the virus becomes less deadly with time (in a rich country with vast resources to ramp up healthcare).
I'm not too worried about future waves, even if they reach higher than this one - we will have mass testing, PPE and healthcare, plus a proven method of crushing them down again with bursts of full lockdowns.
Up to a point. It's easy to make thousands of ventilators and put beds in convention centres; the bottleneck is getting the staff to operate them. We can't continue using cardiologists and kidneyologists as interim ICUologists, or have people working 80-hour weeks for more than a couple of weeks. Getting more trained staff takes a long time.
-
EACLucifer
- Stummy Beige
- Posts: 4177
- Joined: Fri Dec 13, 2019 7:49 am
- Location: In Sumerian Haze
Post
by EACLucifer » Thu Apr 09, 2020 4:10 pm
Sciolus wrote: ↑Thu Apr 09, 2020 3:41 pm
lpm wrote: ↑Thu Apr 09, 2020 7:20 am
Also, the virus becomes less lethal and transmittable with time. A wave next year is less damaging than the current wave. Not because of any change to the RNA but because of change to the human hosts - far more hospital beds, thousands of ventilators, nurses with PPE. And better techniques, experience of which drug combinations work and millions of tests. The virus becomes less infectious because 2 metres becomes second nature.
What's infuriating about the govt approach is they got obsessed with 2nd waves so deliberately ran the first one hotter, while failing to prepare the equipment etc. They missed the fundamental fact that the virus becomes less deadly with time (in a rich country with vast resources to ramp up healthcare).
I'm not too worried about future waves, even if they reach higher than this one - we will have mass testing, PPE and healthcare, plus a proven method of crushing them down again with bursts of full lockdowns.
Up to a point. It's easy to make thousands of ventilators and put beds in convention centres; the bottleneck is getting the staff to operate them. We can't continue using cardiologists and kidneyologists as interim ICUologists, or have people working 80-hour weeks for more than a couple of weeks. Getting more trained staff takes a long time.
While this is true, there are other advantages bought by time, too; there's dozens of drug and vaccine candidates in trials right now. If a second wave hit after better treatment, prophylaxis, or even better a vaccine were developed, it would cost a lot less human life.
-
shpalman
- Princess POW
- Posts: 8428
- Joined: Mon Nov 11, 2019 12:53 pm
- Location: One step beyond
-
Contact:
Post
by shpalman » Thu Apr 09, 2020 6:35 pm
shpalman wrote: ↑Thu Apr 09, 2020 3:29 pm
lpm wrote: ↑Thu Apr 09, 2020 3:24 pm
Lol
This is the [Imperial College] forecast for the UK for this week again (pink plume again, below). The data were already outside their range by yesterday. What on earth were they thinking?
Why didn't Imperial College sense-check their results? How come amateurs can take a five minute look and think their results are bonkers?
Because their model is too complicated and nobody can remember how it works?
Alternatively they like their model (or their prior) and if it deviates from the UK data they assume the data are wrong, just like they assumed the Italian data were wrong a month ago.
having that swing is a necessary but not sufficient condition for it meaning a thing
@shpalman@mastodon.me.uk
-
Millennie Al
- After Pie
- Posts: 1621
- Joined: Mon Mar 16, 2020 4:02 am
Post
by Millennie Al » Fri Apr 10, 2020 2:43 am
lpm wrote: ↑Thu Apr 09, 2020 7:20 am
What's the problem with allowing exponential growth to restart if we have a method to halt it? Lockdowns are now a proven success.
The problem is that lockdown is an extreme measure which is unacceptable in a democratic society,except in extreme circumstances, not something to be adopted as the new normal to be used routinely. We need to eliminate this disease as much as possible on the first attempt so that the only people alive today who ever need to experience another lockdown will do so in th 22nd century an be in the news as the centenarians who survived the last great plague - just like we now regard those who survived the 1918 Spanish flu.
There are actually two approaches to eliminating the disease - the Wuhan strategy of extreme measures to stamp it out rapidly, and the "herd immunity" strategy of letting it rip through the population. The latter is unavoidable in the poorest areas of the world where many people would starve if they tried to stay at home doing nothing for months and their homes are far too small to isolate them anyway. Rich countries need to get to a point where we have locally driven the disease to the brink of extinction when we can retrurn soceity to normal. This will then leave a large surplus of ventilators which we can ship to poor countries that wouldn't have been able to afford them themselves. This can be seen as part of the Chinese strategy - China has suppressed the disease to the extent that factories there can safely operate and churn out hugely increased stocks of medical equipment to send around the world to countries (like us) still in need. Rich countries can express their gratitude by paying for this, while poor ones can accept charity.
-
lpm
- Junior Mod
- Posts: 6361
- Joined: Mon Nov 11, 2019 1:05 pm
Post
by lpm » Fri Apr 10, 2020 7:35 am
Sadly that's not achievable. The disease cannot be eradicated to such an extent. It's embedded in every European country and US state.
Even millions of tests wouldn't be enough, due to asymptomatic cases not being tested. Flare ups will happen.
Awarded gold star 4 November 2021
-
Pucksoppet
- Snowbonk
- Posts: 599
- Joined: Mon Nov 11, 2019 8:13 pm
- Location: Girdling the Earth
Post
by Pucksoppet » Fri Apr 10, 2020 8:50 am
lpm wrote: ↑Fri Apr 10, 2020 7:35 am
Sadly that's not achievable. The disease cannot be eradicated to such an extent. It's embedded in every European country and US state.
Even millions of tests wouldn't be enough, due to asymptomatic cases not being tested. Flare ups will happen.
Accurate and timely detection of flare-ups and effective containment will be needed, and is a reasonable strategy to practise, ready for the next pandemic. Pandemics don't come once-every-hundred-years; occurrences can probably be modelled over time by a Poisson distribution, and it is entirely possible to have another one occur next year; or for two to occur simultaneously.
-
Woodchopper
- Princess POW
- Posts: 7317
- Joined: Sat Oct 12, 2019 9:05 am
Post
by Woodchopper » Fri Apr 10, 2020 9:13 am
Millennie Al wrote: ↑Fri Apr 10, 2020 2:43 am
lpm wrote: ↑Thu Apr 09, 2020 7:20 am
What's the problem with allowing exponential growth to restart if we have a method to halt it? Lockdowns are now a proven success.
There are actually two approaches to eliminating the disease - the Wuhan strategy of extreme measures to stamp it out rapidly, and the "herd immunity" strategy of letting it rip through the population. The latter is unavoidable in the poorest areas of the world where many people would starve if they tried to stay at home doing nothing for months and their homes are far too small to isolate them anyway. Rich countries need to get to a point where we have locally driven the disease to the brink of extinction when we can retrurn soceity to normal. This will then leave a large surplus of ventilators which we can ship to poor countries that wouldn't have been able to afford them themselves. This can be seen as part of the Chinese strategy - China has suppressed the disease to the extent that factories there can safely operate and churn out hugely increased stocks of medical equipment to send around the world to countries (like us) still in need. Rich countries can express their gratitude by paying for this, while poor ones can accept charity.
There will hopefully be a vaccine within a year. If we assume so there is a third strategy which is to keep social controls until that occurs but vary them and perform a difficult balancing act between minimizing infection and also avoiding excessive poverty (which also has negative public health consequences).
-
Pucksoppet
- Snowbonk
- Posts: 599
- Joined: Mon Nov 11, 2019 8:13 pm
- Location: Girdling the Earth
Post
by Pucksoppet » Fri Apr 10, 2020 9:51 am
Once a country has the outbreak under control and can lift the lockdown, I assume international travel restrictions will continue. It could be interesting to live in a world where every time you cross a border, you go into a mandatory 14-day quarantine. A 'simple' business meeting could well take 28 days, and holidays outside your home quarantine zone would become expensive.
I imagine some countries might well group together to have a common quarantine zone -there would be intrinsic economic incentives to do so.
-
PeteB
- Clardic Fug
- Posts: 205
- Joined: Wed Nov 13, 2019 1:02 pm
Post
by PeteB » Fri Apr 10, 2020 10:53 am
My two worries are both around how infectious Covid-19 is
1) The turning on and off of controls (hammer and dance) was modelled on R0 being 2.2 when in reality it is more like 3.2. When you relax controls it is going to explode very quickly. Also, business can't be turned on and off like a light switch
2) I think even with the current UK controls in place, the daily infections and death toll won't fall quickly, it will take months to fall to a level where we can relax controls
-
bob sterman
- Dorkwood
- Posts: 1200
- Joined: Mon Nov 11, 2019 10:25 pm
- Location: Location Location
Post
by bob sterman » Fri Apr 10, 2020 11:24 am
PeteB wrote: ↑Fri Apr 10, 2020 10:53 am
My two worries are both around how infectious Covid-19 is
1) The turning on and off of controls (hammer and dance) was modelled on R
0 being 2.2 when in reality it is more like 3.2. When you relax controls it is going to explode very quickly. Also, business can't be turned on and off like a light switch
Once a portion of the population has been infected it won't be R
0 anymore. It'll just be R which should have a value less than the initial R
0 value.
-
Pucksoppet
- Snowbonk
- Posts: 599
- Joined: Mon Nov 11, 2019 8:13 pm
- Location: Girdling the Earth
Post
by Pucksoppet » Fri Apr 10, 2020 11:50 am
bob sterman wrote: ↑Fri Apr 10, 2020 11:24 am
PeteB wrote: ↑Fri Apr 10, 2020 10:53 am
My two worries are both around how infectious Covid-19 is
1) The turning on and off of controls (hammer and dance) was modelled on R
0 being 2.2 when in reality it is more like 3.2. When you relax controls it is going to explode very quickly. Also, business can't be turned on and off like a light switch
Once a portion of the population has been infected it won't be R
0 anymore. It'll just be R which should have a value less than the initial R
0 value.
That's the hope, but there are reports that some people are not gaining long-term immunity from their first infection.
TIME Magazine: Can You Be Re-Infected After Recovering From Coronavirus? Here's What We Know About COVID-19 Immunity: By Hillary Leung April 3, 2020
Troubling headlines have been cropping up across Asia: Some patients in China, Japan and South Korea who were diagnosed with COVID-19 and seemingly recovered have been readmitted to the hospital after testing positive for the virus again.
...
With other coronavirus strains, experts say the antibodies that patients produce during infection give them immunity to the specific virus for months or even years, but researchers are still figuring out if and how that works with COVID-19.
-
lpm
- Junior Mod
- Posts: 6361
- Joined: Mon Nov 11, 2019 1:05 pm
Post
by lpm » Fri Apr 10, 2020 12:07 pm
bob sterman wrote: ↑Fri Apr 10, 2020 11:24 am
Once a portion of the population has been infected it won't be R
0 anymore. It'll just be R which should have a value less than the initial R
0 value.
That's months off. The reciprocal calc barely notices an increase from 5% to 10% immune.
Gets relevant once 20 million have had it - till then artificial changes to R0 via behaviour is all that matters.
Awarded gold star 4 November 2021
-
lpm
- Junior Mod
- Posts: 6361
- Joined: Mon Nov 11, 2019 1:05 pm
Post
by lpm » Fri Apr 10, 2020 12:13 pm
PeteB wrote: ↑Fri Apr 10, 2020 10:53 am
My two worries are both around how infectious Covid-19 is
1) The turning on and off of controls (hammer and dance) was modelled on R0 being 2.2 when in reality it is more like 3.2. When you relax controls it is going to explode very quickly. Also, business can't be turned on and off like a light switch
2) I think even with the current UK controls in place, the daily infections and death toll won't fall quickly, it will take months to fall to a level where we can relax controls
To be brutal, the NHS can cope at the 1,000 deaths per day level long term, assuming not regionally clustered. And for short spells can cope with 2,000 a day - ie now.
If you could stabilise it then it would be more straightforward. But it's a non balancy equilibrium thingy. That inevitably means you have to target very low rates - say 100 deaths a day - in order to have space to respond when it spikes up to 1,000.
Awarded gold star 4 November 2021
-
lpm
- Junior Mod
- Posts: 6361
- Joined: Mon Nov 11, 2019 1:05 pm
Post
by lpm » Fri Apr 10, 2020 1:44 pm
The way to do this is to list every normal activity we used to do in The Before.
Then give each 3 scores:
1) infectiousness
E.g gym 10 points, cinema 20 points, office work 5 points, pop concert 30 points, walks in countryside 1 point, aeroplanes 50 points. Get a panel of experts to do the estimates.
2) Importance Ranking
How much we value it. List in order of priority E.g. meeting family top, health near top, meeting friends higher than meeting work colleagues, getting nails manicured low down, etc
3) Velocity
How fast it can be lockdowned on and off. E.g meeting friends fast, opening restaurants medium, arranging pop festivals slow.
Then give your country a certain number of Infectiousness Points to allocate. Like choosing between spending priorities. Accept you can't do everything and must ration.
So you wouldn't unlock "expensive" low ranked stuff, cinemas say. But you would unlock seeing friends, even though it costs a lot of points. And you'd avoid slow velocity things like a sports event - better to spend your points on something that could be effortlessly switched off.
Aim is to get most of your top ranked things, sacrifice some medium ranked (particularly if slow velocity), and keep a ban on low ranked (unless very cheap and fast velocity).
Awarded gold star 4 November 2021