Please log in or register. Registered visitors get fewer ads.
Forum index | Previous Thread | Next thread
Interesting thread on modelling to say the least 21:03 - Dec 18 with 1356 viewshype313


Poll: Simpson - Keep, Sell or Loan

0
Interesting thread on modelling to say the least on 09:08 - Dec 19 with 1068 viewsGlasgowBlue

I’m not sure why you got a downvote for that. I think it’s a very interesting conversation.

I didn’t realise that they were only ever asked to produce worst case scenarios rather than best and sheds some light on why that real figures have never matched the modelling.

If I put a business plan together I produce best, worst and various scenarios in between. Sage appear to be ignoring any positive data and only going with the negative. Then again, that seems to be their remit.

Edit. I’d also add that we, as the general public, shouldn’t really have access to this modelling. It’s commissioned by the government in order to take action in the worst case scenario. They get to make an informed decision, or would hope they do, whereas we see the worst case scenario and start panicking.

I’d dread to think what the public reaction would have been during the war if they’d have known everything that was going in.

Then again, I suppose the government have to justify their actions to the public. It’s a hard one.
[Post edited 19 Dec 2021 9:53]

Iron Lion Zion
Poll: Our best central defensive partnership?
Blog: [Blog] For the Sake of My Football Club, Please Go

0
Interesting thread on modelling to say the least on 09:30 - Dec 19 with 1018 viewsBanksterDebtSlave

Interesting thread on modelling to say the least on 09:08 - Dec 19 by GlasgowBlue

I’m not sure why you got a downvote for that. I think it’s a very interesting conversation.

I didn’t realise that they were only ever asked to produce worst case scenarios rather than best and sheds some light on why that real figures have never matched the modelling.

If I put a business plan together I produce best, worst and various scenarios in between. Sage appear to be ignoring any positive data and only going with the negative. Then again, that seems to be their remit.

Edit. I’d also add that we, as the general public, shouldn’t really have access to this modelling. It’s commissioned by the government in order to take action in the worst case scenario. They get to make an informed decision, or would hope they do, whereas we see the worst case scenario and start panicking.

I’d dread to think what the public reaction would have been during the war if they’d have known everything that was going in.

Then again, I suppose the government have to justify their actions to the public. It’s a hard one.
[Post edited 19 Dec 2021 9:53]


I found the negative vote strange, as is the lack of replies! That often says the most in these situations!

Edit.... I added this link to the "ummm" thread too and no response there either.
[Post edited 19 Dec 2021 9:37]

"They break our legs and tell us to be grateful when they offer us crutches."
Poll: If the choice is Moore or no more.

-1
Interesting thread on modelling to say the least on 09:33 - Dec 19 with 1000 viewseireblue

Interesting thread on modelling to say the least on 09:30 - Dec 19 by BanksterDebtSlave

I found the negative vote strange, as is the lack of replies! That often says the most in these situations!

Edit.... I added this link to the "ummm" thread too and no response there either.
[Post edited 19 Dec 2021 9:37]


You can only get a few free Spectator articles a day. Others were posted yesterday.
0
Interesting thread on modelling to say the least on 09:48 - Dec 19 with 960 viewsStokieBlue

Interesting thread on modelling to say the least on 09:08 - Dec 19 by GlasgowBlue

I’m not sure why you got a downvote for that. I think it’s a very interesting conversation.

I didn’t realise that they were only ever asked to produce worst case scenarios rather than best and sheds some light on why that real figures have never matched the modelling.

If I put a business plan together I produce best, worst and various scenarios in between. Sage appear to be ignoring any positive data and only going with the negative. Then again, that seems to be their remit.

Edit. I’d also add that we, as the general public, shouldn’t really have access to this modelling. It’s commissioned by the government in order to take action in the worst case scenario. They get to make an informed decision, or would hope they do, whereas we see the worst case scenario and start panicking.

I’d dread to think what the public reaction would have been during the war if they’d have known everything that was going in.

Then again, I suppose the government have to justify their actions to the public. It’s a hard one.
[Post edited 19 Dec 2021 9:53]


I think he's been rattled on Twitter and not responded as eloquently as he maybe should have.

Essentially he's saying that they don't model the "steady-state" because it's known, it's just the existing Delta line that we've had for 6 months now. Any modelling which requires a decision to be made is a deviation from that baseline and thus that is what they are focused on.

Many would perhaps see this as irresponsible but I would argue it's just as irresponsible to create a scenario where less virulence is assumed before it is confirmed because there will be a large number of people (and Tories) seize on that line and never waiver from it no matter what the subsequent data on virulence shows. Given that an assumption of the same virulence as Delta is certainly defendable.

I don't think there is some conspiracy here (although the Spectator are very "anti-covid" reporting at the moment so are likely to spin it that way), it's worth noting that other countries are modelling similar numbers hence Holland is going into full-lockdown today. There is even a caveat there though in that they only have 10% of their population boosted.

Also worth noting that it's not just about the modelled scenarios. Staff absences are factored in separately. As it stands 40 of the 142 fire engines in London are out of service because the staff are off with covid. That could be incredibly dangerous at some point if it continues and the same is likely to happen with AE staff and ambulance staff.

I certainly think you're right that the modelling is essentially too complex for the "feral public" to understand (and caveats and assumptions are ignored along with probabilities when the numbers are reported in the press) and thus it causes issues when it's released but that is the way of the world now where everyone expects access to every piece of information even if they don't understand it. There is also a very valid argument which can be made that everyone should have access to all the data.

Here is an article by the same SAGE scientist also in the Spectator:

https://www.spectator.co.uk/article/what-the-media-gets-wrong-about-sage-s-model

Some interesting quotes:

"SPI-M, the subcommittee of Sage that I chair, produces scenarios to inform government decisions, not to inform public understanding directly. They are not meant to be predictions of what will actually happen and a range of scenarios are presented to policymakers with numerous caveats and uncertainties emphasised. These can be misinterpreted in the media, where headlines often use the ‘worst case scenario’ alone and then later suggest we are ‘doomsters’ if that didn’t come to pass. But the media are not the primary audience of these models, though of course we want to do what we can to communicate with the media and the public."

SB
[Post edited 19 Dec 2021 11:25]

Avatar - IC410 - Tadpoles Nebula

3
Interesting thread on modelling to say the least on 09:54 - Dec 19 with 932 viewsgordon

I think the horror over this belies a misunderstanding of the purpose and limitations of mathematical epidemiological modelling in this context.

Epidemiological modelling isn't a tool that is able to produce probabilities of different outcomes - while you could produce loads of models parameterised with different assumptions on transmissibility, pre-existing immunity and virulence where knowledge of these factors is poor, this would effectively just produce a range of outcomes (with no indication of which was most likely) from very very few deaths/hospitalisations etc., to a really, really high level of deaths/hospitalisations.

So it isn't that useful to model when knowledge of parameters is poor, except for producing a worst case scenario - but everything less bad than a worst case scenario should also be considered possible, and modelling wouldn't produce evidence on how likely the outcome is to be significantly less than that worst case scenario.

But the idea that seems to be prevalent in twitter etc that Johnson / Javid are itching to do a lockdown and have commissioned some evidence to support that from Sage is pretty bizarre.
2
Interesting thread on modelling to say the least on 10:02 - Dec 19 with 901 viewseireblue

Interesting thread on modelling to say the least on 09:08 - Dec 19 by GlasgowBlue

I’m not sure why you got a downvote for that. I think it’s a very interesting conversation.

I didn’t realise that they were only ever asked to produce worst case scenarios rather than best and sheds some light on why that real figures have never matched the modelling.

If I put a business plan together I produce best, worst and various scenarios in between. Sage appear to be ignoring any positive data and only going with the negative. Then again, that seems to be their remit.

Edit. I’d also add that we, as the general public, shouldn’t really have access to this modelling. It’s commissioned by the government in order to take action in the worst case scenario. They get to make an informed decision, or would hope they do, whereas we see the worst case scenario and start panicking.

I’d dread to think what the public reaction would have been during the war if they’d have known everything that was going in.

Then again, I suppose the government have to justify their actions to the public. It’s a hard one.
[Post edited 19 Dec 2021 9:53]


A business plan, if wrong, doesn’t lead to excess deaths. A business plan is not written to anticipate what will happen in three weeks, with exponential changes that occur in days.

Did you ever have in any of your business plans, a section on what to do when forced to shut by government action with 3 days notice.

We are still in a pandemic, with variants that grow exponentially and even faster than previously, such that if something really bad is going to happen, you will start to know three weeks later.

That Twitter conversation, shows the difference between a mathematics scientist, and journalist.

It is entirely logical to not model a scenario where everything is fine.
In a model, you state the assumptions, and show the possible outcomes.

The decision making is then the same, what is the difference in outcome between different possible worst case scenarios and an assumption that everything is fine, and crucially is that gap big enough to do anything.

Asking someone to create a scenario then a range of models where everything is fine, doesn’t add anything to that decision process.

On the models, you will notice some of the graphs are fairly accurate, in terms of initial growth.

Therefore you can compare real-time data to a model, and try and gauge if you are on one of the bad paths. And of course, if you decide too late that you are on a trajectory that will lead to overwhelming the NHS and excess deaths….
1
Interesting thread on modelling to say the least on 10:06 - Dec 19 with 865 viewsGlasgowBlue

Interesting thread on modelling to say the least on 10:02 - Dec 19 by eireblue

A business plan, if wrong, doesn’t lead to excess deaths. A business plan is not written to anticipate what will happen in three weeks, with exponential changes that occur in days.

Did you ever have in any of your business plans, a section on what to do when forced to shut by government action with 3 days notice.

We are still in a pandemic, with variants that grow exponentially and even faster than previously, such that if something really bad is going to happen, you will start to know three weeks later.

That Twitter conversation, shows the difference between a mathematics scientist, and journalist.

It is entirely logical to not model a scenario where everything is fine.
In a model, you state the assumptions, and show the possible outcomes.

The decision making is then the same, what is the difference in outcome between different possible worst case scenarios and an assumption that everything is fine, and crucially is that gap big enough to do anything.

Asking someone to create a scenario then a range of models where everything is fine, doesn’t add anything to that decision process.

On the models, you will notice some of the graphs are fairly accurate, in terms of initial growth.

Therefore you can compare real-time data to a model, and try and gauge if you are on one of the bad paths. And of course, if you decide too late that you are on a trajectory that will lead to overwhelming the NHS and excess deaths….


I haven’t read your whole post as I’ve got to this part….” Did you ever have in any of your business plans, a section on what to do when forced to shut by government action with 3 days notice. ”

I’ve been doing exactly that for the last year and a half. Creating scenarios months ago for what May happen next week if we are closed again. And plans for what do do if we don’t.

It would be irresponsible and foolish not to do so in such a fluid environment.
[Post edited 19 Dec 2021 10:07]

Iron Lion Zion
Poll: Our best central defensive partnership?
Blog: [Blog] For the Sake of My Football Club, Please Go

0
Interesting thread on modelling to say the least on 10:29 - Dec 19 with 802 viewsBanksterDebtSlave

Interesting thread on modelling to say the least on 09:48 - Dec 19 by StokieBlue

I think he's been rattled on Twitter and not responded as eloquently as he maybe should have.

Essentially he's saying that they don't model the "steady-state" because it's known, it's just the existing Delta line that we've had for 6 months now. Any modelling which requires a decision to be made is a deviation from that baseline and thus that is what they are focused on.

Many would perhaps see this as irresponsible but I would argue it's just as irresponsible to create a scenario where less virulence is assumed before it is confirmed because there will be a large number of people (and Tories) seize on that line and never waiver from it no matter what the subsequent data on virulence shows. Given that an assumption of the same virulence as Delta is certainly defendable.

I don't think there is some conspiracy here (although the Spectator are very "anti-covid" reporting at the moment so are likely to spin it that way), it's worth noting that other countries are modelling similar numbers hence Holland is going into full-lockdown today. There is even a caveat there though in that they only have 10% of their population boosted.

Also worth noting that it's not just about the modelled scenarios. Staff absences are factored in separately. As it stands 40 of the 142 fire engines in London are out of service because the staff are off with covid. That could be incredibly dangerous at some point if it continues and the same is likely to happen with AE staff and ambulance staff.

I certainly think you're right that the modelling is essentially too complex for the "feral public" to understand (and caveats and assumptions are ignored along with probabilities when the numbers are reported in the press) and thus it causes issues when it's released but that is the way of the world now where everyone expects access to every piece of information even if they don't understand it. There is also a very valid argument which can be made that everyone should have access to all the data.

Here is an article by the same SAGE scientist also in the Spectator:

https://www.spectator.co.uk/article/what-the-media-gets-wrong-about-sage-s-model

Some interesting quotes:

"SPI-M, the subcommittee of Sage that I chair, produces scenarios to inform government decisions, not to inform public understanding directly. They are not meant to be predictions of what will actually happen and a range of scenarios are presented to policymakers with numerous caveats and uncertainties emphasised. These can be misinterpreted in the media, where headlines often use the ‘worst case scenario’ alone and then later suggest we are ‘doomsters’ if that didn’t come to pass. But the media are not the primary audience of these models, though of course we want to do what we can to communicate with the media and the public."

SB
[Post edited 19 Dec 2021 11:25]


Just to confirm. Does this mean they are modelling based on delta?

"They break our legs and tell us to be grateful when they offer us crutches."
Poll: If the choice is Moore or no more.

0
Login to get fewer ads

Interesting thread on modelling to say the least on 10:31 - Dec 19 with 794 viewseireblue

Interesting thread on modelling to say the least on 10:06 - Dec 19 by GlasgowBlue

I haven’t read your whole post as I’ve got to this part….” Did you ever have in any of your business plans, a section on what to do when forced to shut by government action with 3 days notice. ”

I’ve been doing exactly that for the last year and a half. Creating scenarios months ago for what May happen next week if we are closed again. And plans for what do do if we don’t.

It would be irresponsible and foolish not to do so in such a fluid environment.
[Post edited 19 Dec 2021 10:07]


That is why I wrote in the past tense, “Did you…” not “Do you..”

Sorry, I appreciate you have been through a lot, it wasn’t meant to be a personal dig.

I was just riffing on the difference between a business plan, since you introduced the concept, and pandemic planning using models.
They are not analogous.
0
Interesting thread on modelling to say the least on 10:44 - Dec 19 with 767 viewseireblue

Interesting thread on modelling to say the least on 10:29 - Dec 19 by BanksterDebtSlave

Just to confirm. Does this mean they are modelling based on delta?


You seem to have done a lot of reading on this topic.

What known data sets of significant size would you use to estimate how bad Omicron is?

And how would you describe it?
0
Interesting thread on modelling to say the least on 10:54 - Dec 19 with 711 viewsBanksterDebtSlave

Interesting thread on modelling to say the least on 10:44 - Dec 19 by eireblue

You seem to have done a lot of reading on this topic.

What known data sets of significant size would you use to estimate how bad Omicron is?

And how would you describe it?


Is that a yes?

Edit...nope, hardly any as pretty busy with work and family but considering their low vaccination rates this looks promising.
https://www.google.com/search?ie=UTF-8&client=tablet-android-samsung-rev2&am
[Post edited 19 Dec 2021 10:57]

"They break our legs and tell us to be grateful when they offer us crutches."
Poll: If the choice is Moore or no more.

0
Interesting thread on modelling to say the least on 11:00 - Dec 19 with 672 viewsPinewoodblue

Interesting thread on modelling to say the least on 10:29 - Dec 19 by BanksterDebtSlave

Just to confirm. Does this mean they are modelling based on delta?


When a new variant hits the first thing you find out is how transmissible it is. You don’t find out about the numbers likely to be hospitalised until much later. Presumably you therefore use the data you have on other variants.

The biggest problem is when this information gets into the hands of mainstream media.

2023 year of destiny
Poll: Dickhead "Noun" a stupid, irritating, or ridiculous man.

0
Interesting thread on modelling to say the least on 11:04 - Dec 19 with 658 viewsgordon

Interesting thread on modelling to say the least on 10:54 - Dec 19 by BanksterDebtSlave

Is that a yes?

Edit...nope, hardly any as pretty busy with work and family but considering their low vaccination rates this looks promising.
https://www.google.com/search?ie=UTF-8&client=tablet-android-samsung-rev2&am
[Post edited 19 Dec 2021 10:57]


I think one of the ways to think about is that if sage produce a modelled outcome based on R = 1.3, and then as every day/week passes in which ministers are getting in more information that says R = 1.3 and they want to avoid the modelled outcome, they have to act.

But if the real-world data is saying R = 1.1 then they know that the actual outcome is probably going to be better than the modelled outcome, and intervention is likely not needed.

Equally if the model is based on 1% of reported cases resulting in hospitalisation, and the actual world data is showing that 0.5% is more accurate, then that modelled outcome is likely to be avoided.

In these cases it isn't that 'the model is wrong' - the model(s) has/have helped to inform the decision-making process. The problem is when people think of models as being predictions of what's actually going to happen.
[Post edited 19 Dec 2021 11:04]
2
Interesting thread on modelling to say the least on 11:14 - Dec 19 with 610 viewsFreddies_Ears

Interesting thread on modelling to say the least on 11:04 - Dec 19 by gordon

I think one of the ways to think about is that if sage produce a modelled outcome based on R = 1.3, and then as every day/week passes in which ministers are getting in more information that says R = 1.3 and they want to avoid the modelled outcome, they have to act.

But if the real-world data is saying R = 1.1 then they know that the actual outcome is probably going to be better than the modelled outcome, and intervention is likely not needed.

Equally if the model is based on 1% of reported cases resulting in hospitalisation, and the actual world data is showing that 0.5% is more accurate, then that modelled outcome is likely to be avoided.

In these cases it isn't that 'the model is wrong' - the model(s) has/have helped to inform the decision-making process. The problem is when people think of models as being predictions of what's actually going to happen.
[Post edited 19 Dec 2021 11:04]


Exactly. To quote George Box: "All models are wrong but some are useful".
0
Interesting thread on modelling to say the least on 11:19 - Dec 19 with 594 viewsGlasgowBlue

Interesting thread on modelling to say the least on 10:31 - Dec 19 by eireblue

That is why I wrote in the past tense, “Did you…” not “Do you..”

Sorry, I appreciate you have been through a lot, it wasn’t meant to be a personal dig.

I was just riffing on the difference between a business plan, since you introduced the concept, and pandemic planning using models.
They are not analogous.


No problem. I didn’t take it as a personal dig.

The modellers, like most people in business, are now two years into this so I would have expected them to adapt as business’s have and produced models based on good and bad outcomes.

As I said at the start, they are only working to the brief given them by government.

Iron Lion Zion
Poll: Our best central defensive partnership?
Blog: [Blog] For the Sake of My Football Club, Please Go

0
Interesting thread on modelling to say the least on 11:21 - Dec 19 with 588 viewsWD19

Interesting thread on modelling to say the least on 10:29 - Dec 19 by BanksterDebtSlave

Just to confirm. Does this mean they are modelling based on delta?


When they say they are modelling based on a range of scenarios I imagine it means they are probably modelling on a range of scenarios.

I love these threads almost as much as I love the threads complaining about the fact the authorities haven’t listened to the science and moved far and fast enough.
0
Interesting thread on modelling to say the least on 11:21 - Dec 19 with 588 viewsgiant_stow

Interesting thread on modelling to say the least on 11:04 - Dec 19 by gordon

I think one of the ways to think about is that if sage produce a modelled outcome based on R = 1.3, and then as every day/week passes in which ministers are getting in more information that says R = 1.3 and they want to avoid the modelled outcome, they have to act.

But if the real-world data is saying R = 1.1 then they know that the actual outcome is probably going to be better than the modelled outcome, and intervention is likely not needed.

Equally if the model is based on 1% of reported cases resulting in hospitalisation, and the actual world data is showing that 0.5% is more accurate, then that modelled outcome is likely to be avoided.

In these cases it isn't that 'the model is wrong' - the model(s) has/have helped to inform the decision-making process. The problem is when people think of models as being predictions of what's actually going to happen.
[Post edited 19 Dec 2021 11:04]


Nice explanation - thanks. You'd better post it to Dan hodges before he explodes in a piff of self righteousness outrage.

Has anyone ever looked at their own postings for last day or so? Oh my... so sorry. Was Ullaa
Poll: A clasmate tells your son their going to beat him up in the playground after sch

2
Interesting thread on modelling to say the least on 11:44 - Dec 19 with 524 viewsgordon

Interesting thread on modelling to say the least on 11:19 - Dec 19 by GlasgowBlue

No problem. I didn’t take it as a personal dig.

The modellers, like most people in business, are now two years into this so I would have expected them to adapt as business’s have and produced models based on good and bad outcomes.

As I said at the start, they are only working to the brief given them by government.


Because the value of the models is in informing the decision-making process, it only really makes sense to produce models parameterised to produce outcomes at the margins of where intervention would / wouldn't be necessary.

If the real world data coming in (on transmissability, hospitalisation rates etc) is better than the modelled outcome by a large margin (which is probably very much possible in this case), then this means the decision-makers can be really confident that no further restrictions is the correct decision - and knowing that the real outcome will be much better than the modelled outcome has allowed them to be confident in that decision.
2
About Us Contact Us Terms & Conditions Privacy Cookies Advertising
© TWTD 1995-2024