This isn't exactly a new argument. Prof John Robison of Edinburgh in the early 1700s argued that an equation's solution set that approximated a data set could be both useful and incorrect. Prof Laplace of Paris insisted contrarily that the equation was always perfect and the measurements were flawed. Euler took the iterative solutions route between the two: For series calculations that converged, we could attain adequate accuracy with enough calculations. Digital computing has made Euler's approach practical for validating models, but three centuries later, those who insist on unknowable levels of certainty, use their imagined certainty to ignore all contradictory facts and impose their beliefs upon the world, by force.
Excellent analysis thank you. I am reminded of an economics lesson during my studies when we were discussing the alignment of Stone Henge. Our professor suggested we go out into the city and log the location of as many Zebra Crossings we could in the time allowed and analyse the data for astronomical alignment. Voila, there appeared to be a correlation of these crossings with the stars!! Must be some kind of weird hidden hand at work - like markets?
I believe that complex adaptive systems are not possible to model. Even with their super-computers, the Met Office rarely get a few days right let alone the intricacies of climate change. Confirmation Bias reigns supreme.
I'm a simple guy. If you've developed a 95% effective vaccine that's a safe preventative for the 3rd leading cause of death in your society, then total deaths CANNOT increase. Any attempt to attribute that increase in death to something else is farcical because, sure, it's possible, but it's blatantly unreasonable. I'm an attorney, and I remind people that we don't execute criminals based on proof; we execute them based on proof BEYOND a reasonable doubt. When you start vaccinating & total deaths increase, you have no reasonable doubt - the vaccines must end. Over-intellectualizing, over-thinking are extremely dangerous trends, especially for the intelligent. It's very easy to persuade yourself that there's some universe in which these vaccines work, or there are 73 genders, or that socialism just hasn't been tried, but none of these beliefs are reasonable - they're all nuts.
It's really very simple: if you "cure" the third leading cause of death, then total deaths MUST decline. Anyone claiming there's an exception to that rule must present extraordinary evidence, not speculation. Indeed, the exception would be obvious (like CA fell into the ocean & drowned millions). It's not some BS about masks make people sad, which makes them sick...such subtle things are not plausible in light of the obvious: you've "cured" the 3rd leading cause of death but deaths increased.
Please stop taking the world "reasonable" out of your analysis/discussions. It's great that you have theories about how the vaccines might not be to blame, but are those theories reasonable? That's the threshold question we must demand our adversaries answer. (You reason from what you know to what you don't, not from what you don't know. We know deaths are increasing despite the cure; what do we infer from that? Granted, we don't know why deaths are increasing, but so what? Uncertainty drives INACTION, not action. That is, you don't vaccinate until you know it's unsafe; you STOP vaccinating until you KNOW it's safe.
Correlation may not prove causation but proving causation in the absence of correlation is surely even less likely?! So, I like to start with easier hypotheses to rule out, like "the vaccine saved millions of lives" kind of hypothesis in the situation where there is actually more COVID and excess death after vaccination than before. How does your model manage in that situation? Are models that prove lack of causation less wrong?
I have your book and can confirm that it is excellent, always clear examples. I knew a lot and learned a lot more.
I will modify the George Box statement, courtesy of George Orwell: All models are wrong, but some are wronger than others. That is what you are getting at. Even causality has been a contested issue. There has been a big battle in physics for a hundred years on the interpretation of quantum mechanics with some just focused on treating it as a computational machine (the results are certainly spectacular like the fine structure constant accuracy of 10^-10 theory/experiment) and others seeking some meaning beyond, as did David Bohm. Everything we work out depends on a theory that may be limited but ultimately is set in a global weltanschauung. Even the experiments we do depend on a weltanschauung. A fascinating book illustrating that 'science' is never settled is Bruno Latour's Science in Action. I have a book by Julian Schwinger that has on the dedication page: If you can't join them, beat them. Thank you for all you are doing to cut through the muddle and deceit going on.
Jan 23, 2023·edited Jan 23, 2023Liked by Martin Neil
I'll need some time to digest this, but on "all models are wrong", it's a bit of a meaningless philosophical point isn't it? Even well-established models of physical nature are not right for all time. I would rather say that all models are incomplete. Some have a degree of predictive ability. In general for models of social processes it is hard to say how long they will maintain any predictive ability that they might seem to have based on past data since these processes are very elastic. So there is not only omitted variable basis but the very processes which they seek to model are not necessarily of a kind that can be reliably modeled at all. We can account for omitted variables, measurement problems etc in how the model is set up but we cannot account for the endogeneity which comes from policy responses to the model and human responses to the policy, certainly not unless we observe a counterfactual. Even if the model does well for past data, we cannot be sure it will continue to do so. Models themselves change the world.
This brings us to the role of heuristics, nicely summarized yesterday by Scott Adams.
"Never trust the government" (putatively) gives better results than any model.
Such heuristics integrate relevant information which *cannot* be in the model, such as the government's known propensity for regulatory capture and the precautionary principle. The decision not to vax today is only for today and retains the real option value of doing so tomorrow. This is a major reason why most people thinking correctly about this, and without impertinent constraints, would not do so ever (even if it may be a prisoner's dilemma).
The ultimate heuristic based on the observed behavior of governments is the constitution, the Bill of Rights, and the rule of law. The entire system of restraints on governments is predicated on their observed propensity to deliver, on occasion, socially suboptimal outcomes. The beauty of this system is that it works for everything and for ever. If it constitutes an obstacle to the implementation of a certain course of action presently thought to be desirable, it is doing exactly what it was set up to do and we have no choice other than to take it on the chin, analyze it properly after the event, and propose, for next time, any relevant adjustments which do not throw out the baby with the bathwater.
This is because the system of one nation under the Constitution delivers societal welfare across the board. Even if it were to turn out that on any specfic issue it impeded a response which would have been welfare enhancing.
I really think it is high time we remembered this. We all have to live our lives making decisions under uncertainty and using heuristics to guide us. The government cannot simply override this. There is too much it doesn't, cannot and even should not know, about the problem and about our personal situations. A case needs to be made for any government intervention, and outside of the context of crisis management. It really isn't hard to grasp this. We have rights inter alia because the government has consistently shown a dreadful propensity to get things horribly wrong. Their defense is rationally worth internalizing some social cost.
I found an extremely strong correlation between non-COVID excess deaths and new vaccinations in the German data (Pearson correlation coefficient is 0.98 nationally, between 0.85 and 0.97 for each federal state). Writing an article about it now.
Non-COVID excess deaths correlate very strongly with newly administered vaccine doses in all German federal states during the time between calendar week 8/2021 and calendar week 27/2021 (Pearson correlation coefficient on the national level: 0.980).
95.8% of the variance in weekly, nationwide non-COVID excess deaths is explained by the number of administered vaccine doses in Germany per week (Standard Error: 78.7 deaths per week).
The correlation is strong enough to suggest a causal link. The insights we have gained into the pharmacodynamics of the nucleotide-based COVID-19 vaccines throughout the past years support the notion of a causal link between non-COVID excess deaths and administered vaccine doses.
One excess death occurred for every 1642 administered vaccine doses (95% CI: 1477-1808 doses).
From the time of their introduction to the market until the end of 2022 a total of 191,029,491 COVID-19 vaccine doses were administered in Germany according to RKI figures. This translates to 116,309 vaccinees who died as a direct consequence of receiving a dose of the nucleotide-based vaccines against COVID-19 in Germany (95% CI: 105,669 - 129,331).
The rationale looks so obvious when it's so clearly explained like this. Critical thinking can be hard and in a world where attention spans seem short (and are often celebrated for being so) it's not surprising that students and others are shocked to learn that models are imperfect. I follow a bit of stuff on climate change and it amazes me how much trust people put in those models. Anyway, another lovely article that helps us all think more critically. Well done chaps!
Thankyou for a great piece clearly investigating the system view of excess deaths. As an engineer who has spent my career on expermental model validation, and so have experienced the effects of the overall system and context on measured data. It is good to know that this analysis has been done and in the public domain even if the media ignore it.
Very interesting article. My first reaction is to provide an alternative aphorism to your consideration:
"Some models are right but are not useful".
What I mean is that a model like yours, that takes into consideration much more variables that have an influence on an aggregate figure cannot be constructed, because many of the relationships between the variables are not quantifiable (for example, they might be not constant, varying in time or space).
For that reason it is only possible to use simplified models, that are necessarily wrong, but at least can be used to test hypothesis and, possibly, reject them. Unfortunately in some cases, simplified models are not useful because they cannot capture a complex phenomenon.
Trying to determine if the vaccines are the main underlying cause in an observed increase in all-cause death using only these two variables, proportion of vaccinated people and observed deaths in many different places and different times is not useful because the required mathematical model to capture the phenomenon is unfeasible as you brilliantly show in the article.
Therefore, in those cases we have to proceed in a different way, forget about modelling complex issues that are not appropriate to that kind of analysis and stop fooling ourselves.
It depends on the country. In the UK everyone was deprived of decent healthcare regardless in order to protect the NHS (when it's supposed to be there to protect us). I appreciate that in other countries things were very different with vaccine mandates etc. and discrimination was one way.
Disentangling these issues is the proverbial Gordian knot.
"In November 2022, excess mortality varied across the EU. Bulgaria, Italy, Romania and Slovakia recorded little or no excess deaths. The most affected countries, Cyprus and Finland recorded excess mortality rates of 23.8 % and 20.5 % respectively, followed by Germany (15.6 %)."
As the UK is no longer a part of the EU (so sad!) the Eurostat is mum on those numbers. But those numbers just baffle me. I'm trying to make sense of them in terms of the causal model of excess death, but I clearly just don't know enough about what the nitty-gritty details of each particular country to make sense of why Finland should have such a high number of excess deaths, but why Italy should have so little. Some of these countries are highly vaccinated (like Italy) but others not so much (Romania for example). It's as if whatever causal story might work for one country is just falsified by another. At any rate, their excess mortality map is quite nice - nicely color-coded, except for sad little UK just in grey, excluded from "our" club!
Understanding a bi-variate model is a stretch for most policy makers. Even then they prefer to have someone identified as an 'expert' to summarise a conclusion. It is surprising just how often that summarised conclusion corresponds to the projected result desired by the policy maker!
So, I disagree. Not all models are 'wrong'. Many models give precisely the 'right' answer.
Thanks for this great explanation. Just by going through the exercise of enunciating the various nodes it really highlights the uncertainty. I'm sure no-one in any government health department has gone through such an exercise. As you point out you can't assign probabilities to the nodes because we don't have data. If doing experiments, in say manufacturing, you could adjust a single variable and see what happens. But say we stopped vaccination now, and it was true there were adverse effects, we wouldn't know for how long that was occurring. There could be a long lag time in the system. It's the ultimate "wicked problem".
Another thought I had was, taking it a level higher, if one were really doing risk engineering, you would look at the outcome, and at some point you would have to say things aren't getting better. We better stop whatever we are doing. No-one is prepared to set a limit. Eg if excess mortality gets to this level we know we have done something wrong. Its like in a clinical trial where you are supposed to set the endpoints in advance of the trial. When excess non-COVID mortaility got to some level like 10% there should have been a pause. Instead a new explanation is pulled out like everyone needs statins. Seems hopeless when noone in authority has any concept of systems thinking.
Whether and how the "vaccines" trigger cancer is covered here by Marc Giradot. "...some critical cells are naturally protected from immune attacks; stem cells and progenitor cells. At the core of the cellular foundation, these “royal” cells are too valuable to the body to be scuttled: They are indispensable for blood, for immunity and for regeneration. Moreover, they can replicate considerably given they are the closest to our original foetal cells.
This is a major loop-hole vaccinologists haven’t apparently considered: What happens if such a cell survives vaccine transfection? Inevitably, given their high replicability, they could trigger cancer - in a form or another".
And Sasha Latypova says: 'Nobody Knows What is in the Vials. Covid-19 injections are dangerous, non-compliant biological materials. Their production must be stopped until a full investigation can be done'. So that is another factor to take in to account. They may be varying wildly. https://sashalatypova.substack.com/p/nobody-knows-what-is-in-the-vials
I remember that initially the Pfizer vaccine had to be administered within 5 days of being removed from ultra-low temperature freezers. Then that went out the window. As no proper trials were carried out or even legally required, the whole thing is a charade of safety and science.
This isn't exactly a new argument. Prof John Robison of Edinburgh in the early 1700s argued that an equation's solution set that approximated a data set could be both useful and incorrect. Prof Laplace of Paris insisted contrarily that the equation was always perfect and the measurements were flawed. Euler took the iterative solutions route between the two: For series calculations that converged, we could attain adequate accuracy with enough calculations. Digital computing has made Euler's approach practical for validating models, but three centuries later, those who insist on unknowable levels of certainty, use their imagined certainty to ignore all contradictory facts and impose their beliefs upon the world, by force.
Amazing article. Truly extraordinary. This is what real science and modelling looks like.
Excellent analysis thank you. I am reminded of an economics lesson during my studies when we were discussing the alignment of Stone Henge. Our professor suggested we go out into the city and log the location of as many Zebra Crossings we could in the time allowed and analyse the data for astronomical alignment. Voila, there appeared to be a correlation of these crossings with the stars!! Must be some kind of weird hidden hand at work - like markets?
I believe that complex adaptive systems are not possible to model. Even with their super-computers, the Met Office rarely get a few days right let alone the intricacies of climate change. Confirmation Bias reigns supreme.
I'm a simple guy. If you've developed a 95% effective vaccine that's a safe preventative for the 3rd leading cause of death in your society, then total deaths CANNOT increase. Any attempt to attribute that increase in death to something else is farcical because, sure, it's possible, but it's blatantly unreasonable. I'm an attorney, and I remind people that we don't execute criminals based on proof; we execute them based on proof BEYOND a reasonable doubt. When you start vaccinating & total deaths increase, you have no reasonable doubt - the vaccines must end. Over-intellectualizing, over-thinking are extremely dangerous trends, especially for the intelligent. It's very easy to persuade yourself that there's some universe in which these vaccines work, or there are 73 genders, or that socialism just hasn't been tried, but none of these beliefs are reasonable - they're all nuts.
It's really very simple: if you "cure" the third leading cause of death, then total deaths MUST decline. Anyone claiming there's an exception to that rule must present extraordinary evidence, not speculation. Indeed, the exception would be obvious (like CA fell into the ocean & drowned millions). It's not some BS about masks make people sad, which makes them sick...such subtle things are not plausible in light of the obvious: you've "cured" the 3rd leading cause of death but deaths increased.
Please stop taking the world "reasonable" out of your analysis/discussions. It's great that you have theories about how the vaccines might not be to blame, but are those theories reasonable? That's the threshold question we must demand our adversaries answer. (You reason from what you know to what you don't, not from what you don't know. We know deaths are increasing despite the cure; what do we infer from that? Granted, we don't know why deaths are increasing, but so what? Uncertainty drives INACTION, not action. That is, you don't vaccinate until you know it's unsafe; you STOP vaccinating until you KNOW it's safe.
Correlation may not prove causation but proving causation in the absence of correlation is surely even less likely?! So, I like to start with easier hypotheses to rule out, like "the vaccine saved millions of lives" kind of hypothesis in the situation where there is actually more COVID and excess death after vaccination than before. How does your model manage in that situation? Are models that prove lack of causation less wrong?
I have your book and can confirm that it is excellent, always clear examples. I knew a lot and learned a lot more.
I will modify the George Box statement, courtesy of George Orwell: All models are wrong, but some are wronger than others. That is what you are getting at. Even causality has been a contested issue. There has been a big battle in physics for a hundred years on the interpretation of quantum mechanics with some just focused on treating it as a computational machine (the results are certainly spectacular like the fine structure constant accuracy of 10^-10 theory/experiment) and others seeking some meaning beyond, as did David Bohm. Everything we work out depends on a theory that may be limited but ultimately is set in a global weltanschauung. Even the experiments we do depend on a weltanschauung. A fascinating book illustrating that 'science' is never settled is Bruno Latour's Science in Action. I have a book by Julian Schwinger that has on the dedication page: If you can't join them, beat them. Thank you for all you are doing to cut through the muddle and deceit going on.
I'll need some time to digest this, but on "all models are wrong", it's a bit of a meaningless philosophical point isn't it? Even well-established models of physical nature are not right for all time. I would rather say that all models are incomplete. Some have a degree of predictive ability. In general for models of social processes it is hard to say how long they will maintain any predictive ability that they might seem to have based on past data since these processes are very elastic. So there is not only omitted variable basis but the very processes which they seek to model are not necessarily of a kind that can be reliably modeled at all. We can account for omitted variables, measurement problems etc in how the model is set up but we cannot account for the endogeneity which comes from policy responses to the model and human responses to the policy, certainly not unless we observe a counterfactual. Even if the model does well for past data, we cannot be sure it will continue to do so. Models themselves change the world.
This brings us to the role of heuristics, nicely summarized yesterday by Scott Adams.
https://twitter.com/ScottAdamsSays/status/1616904705195794432
"Never trust the government" (putatively) gives better results than any model.
Such heuristics integrate relevant information which *cannot* be in the model, such as the government's known propensity for regulatory capture and the precautionary principle. The decision not to vax today is only for today and retains the real option value of doing so tomorrow. This is a major reason why most people thinking correctly about this, and without impertinent constraints, would not do so ever (even if it may be a prisoner's dilemma).
The ultimate heuristic based on the observed behavior of governments is the constitution, the Bill of Rights, and the rule of law. The entire system of restraints on governments is predicated on their observed propensity to deliver, on occasion, socially suboptimal outcomes. The beauty of this system is that it works for everything and for ever. If it constitutes an obstacle to the implementation of a certain course of action presently thought to be desirable, it is doing exactly what it was set up to do and we have no choice other than to take it on the chin, analyze it properly after the event, and propose, for next time, any relevant adjustments which do not throw out the baby with the bathwater.
This is because the system of one nation under the Constitution delivers societal welfare across the board. Even if it were to turn out that on any specfic issue it impeded a response which would have been welfare enhancing.
I really think it is high time we remembered this. We all have to live our lives making decisions under uncertainty and using heuristics to guide us. The government cannot simply override this. There is too much it doesn't, cannot and even should not know, about the problem and about our personal situations. A case needs to be made for any government intervention, and outside of the context of crisis management. It really isn't hard to grasp this. We have rights inter alia because the government has consistently shown a dreadful propensity to get things horribly wrong. Their defense is rationally worth internalizing some social cost.
Excellent, thank you.
I found an extremely strong correlation between non-COVID excess deaths and new vaccinations in the German data (Pearson correlation coefficient is 0.98 nationally, between 0.85 and 0.97 for each federal state). Writing an article about it now.
EDIT: And here is the new article: https://vigilance.pervaers.com/p/german-excess-mortality-part-3
Non-COVID excess deaths correlate very strongly with newly administered vaccine doses in all German federal states during the time between calendar week 8/2021 and calendar week 27/2021 (Pearson correlation coefficient on the national level: 0.980).
95.8% of the variance in weekly, nationwide non-COVID excess deaths is explained by the number of administered vaccine doses in Germany per week (Standard Error: 78.7 deaths per week).
The correlation is strong enough to suggest a causal link. The insights we have gained into the pharmacodynamics of the nucleotide-based COVID-19 vaccines throughout the past years support the notion of a causal link between non-COVID excess deaths and administered vaccine doses.
One excess death occurred for every 1642 administered vaccine doses (95% CI: 1477-1808 doses).
From the time of their introduction to the market until the end of 2022 a total of 191,029,491 COVID-19 vaccine doses were administered in Germany according to RKI figures. This translates to 116,309 vaccinees who died as a direct consequence of receiving a dose of the nucleotide-based vaccines against COVID-19 in Germany (95% CI: 105,669 - 129,331).
The rationale looks so obvious when it's so clearly explained like this. Critical thinking can be hard and in a world where attention spans seem short (and are often celebrated for being so) it's not surprising that students and others are shocked to learn that models are imperfect. I follow a bit of stuff on climate change and it amazes me how much trust people put in those models. Anyway, another lovely article that helps us all think more critically. Well done chaps!
Thankyou for a great piece clearly investigating the system view of excess deaths. As an engineer who has spent my career on expermental model validation, and so have experienced the effects of the overall system and context on measured data. It is good to know that this analysis has been done and in the public domain even if the media ignore it.
Very interesting article. My first reaction is to provide an alternative aphorism to your consideration:
"Some models are right but are not useful".
What I mean is that a model like yours, that takes into consideration much more variables that have an influence on an aggregate figure cannot be constructed, because many of the relationships between the variables are not quantifiable (for example, they might be not constant, varying in time or space).
For that reason it is only possible to use simplified models, that are necessarily wrong, but at least can be used to test hypothesis and, possibly, reject them. Unfortunately in some cases, simplified models are not useful because they cannot capture a complex phenomenon.
Trying to determine if the vaccines are the main underlying cause in an observed increase in all-cause death using only these two variables, proportion of vaccinated people and observed deaths in many different places and different times is not useful because the required mathematical model to capture the phenomenon is unfeasible as you brilliantly show in the article.
Therefore, in those cases we have to proceed in a different way, forget about modelling complex issues that are not appropriate to that kind of analysis and stop fooling ourselves.
Good point.
It depends on the country. In the UK everyone was deprived of decent healthcare regardless in order to protect the NHS (when it's supposed to be there to protect us). I appreciate that in other countries things were very different with vaccine mandates etc. and discrimination was one way.
Disentangling these issues is the proverbial Gordian knot.
Writing from France. It's a very curious thing, as I happened to look at some stats for excess death in the EU here:
https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Excess_mortality_-_statistics
They summarize as follows:
"In November 2022, excess mortality varied across the EU. Bulgaria, Italy, Romania and Slovakia recorded little or no excess deaths. The most affected countries, Cyprus and Finland recorded excess mortality rates of 23.8 % and 20.5 % respectively, followed by Germany (15.6 %)."
As the UK is no longer a part of the EU (so sad!) the Eurostat is mum on those numbers. But those numbers just baffle me. I'm trying to make sense of them in terms of the causal model of excess death, but I clearly just don't know enough about what the nitty-gritty details of each particular country to make sense of why Finland should have such a high number of excess deaths, but why Italy should have so little. Some of these countries are highly vaccinated (like Italy) but others not so much (Romania for example). It's as if whatever causal story might work for one country is just falsified by another. At any rate, their excess mortality map is quite nice - nicely color-coded, except for sad little UK just in grey, excluded from "our" club!
Understanding a bi-variate model is a stretch for most policy makers. Even then they prefer to have someone identified as an 'expert' to summarise a conclusion. It is surprising just how often that summarised conclusion corresponds to the projected result desired by the policy maker!
So, I disagree. Not all models are 'wrong'. Many models give precisely the 'right' answer.
Thanks for this great explanation. Just by going through the exercise of enunciating the various nodes it really highlights the uncertainty. I'm sure no-one in any government health department has gone through such an exercise. As you point out you can't assign probabilities to the nodes because we don't have data. If doing experiments, in say manufacturing, you could adjust a single variable and see what happens. But say we stopped vaccination now, and it was true there were adverse effects, we wouldn't know for how long that was occurring. There could be a long lag time in the system. It's the ultimate "wicked problem".
Another thought I had was, taking it a level higher, if one were really doing risk engineering, you would look at the outcome, and at some point you would have to say things aren't getting better. We better stop whatever we are doing. No-one is prepared to set a limit. Eg if excess mortality gets to this level we know we have done something wrong. Its like in a clinical trial where you are supposed to set the endpoints in advance of the trial. When excess non-COVID mortaility got to some level like 10% there should have been a pause. Instead a new explanation is pulled out like everyone needs statins. Seems hopeless when noone in authority has any concept of systems thinking.
Whether and how the "vaccines" trigger cancer is covered here by Marc Giradot. "...some critical cells are naturally protected from immune attacks; stem cells and progenitor cells. At the core of the cellular foundation, these “royal” cells are too valuable to the body to be scuttled: They are indispensable for blood, for immunity and for regeneration. Moreover, they can replicate considerably given they are the closest to our original foetal cells.
This is a major loop-hole vaccinologists haven’t apparently considered: What happens if such a cell survives vaccine transfection? Inevitably, given their high replicability, they could trigger cancer - in a form or another".
https://covidmythbuster.substack.com/p/are-vaccines-triggering-cancers-vaccine
And Sasha Latypova says: 'Nobody Knows What is in the Vials. Covid-19 injections are dangerous, non-compliant biological materials. Their production must be stopped until a full investigation can be done'. So that is another factor to take in to account. They may be varying wildly. https://sashalatypova.substack.com/p/nobody-knows-what-is-in-the-vials
I remember that initially the Pfizer vaccine had to be administered within 5 days of being removed from ultra-low temperature freezers. Then that went out the window. As no proper trials were carried out or even legally required, the whole thing is a charade of safety and science.