Yet another long diary entry today (4810 words). If you’re losing the will to read right through it, here’s a one sentence summary. My sense is the news is turning positive on multiple fronts.
But, how positive? I’ve written 30,000+ words in my daily diary entries so far (about half the amount of a full size novel), and I’ve been obsessively consuming every possible piece of information about the virus, way more than 12 hours a day, every day without pause, for the last several weeks. I’ve struggled to revive my college stats and calculus. I’ve had the benefit of comments and perspective from some of the brightest people I know, and some of you (also bright, I’m sure) have sent in great comments, ideas, thoughts, suggestions, too. Thank you – even to the people who disagreed with me. It’s only by testing our ideas and getting alternate perspectives that we can improve them – I’d much prefer a dozen helpful corrections and alternate interpretations than a series of echo-chamber agreements.
I say this not to boast. Quite the opposite. After all of this, I still feel and fear I don’t have the slightest idea what our future will be. Not what, not when, not how, not who, not where, not why. Is this virus the greatest threat our modern civilization has ever faced? Or is it the greatest nothing-burger of all time, something we’ll laugh about for generations to come?
Now, I know I’ve been chanting, every day, the mantra that “the exponential growth makes it impossible for us to conceptually perceive the future as it will become”. I also understand that some of the possible outcomes we are threatened with are so apocalyptic in nature as to be impossible to grasp – one of the great things about mankind is a complete inability to perceive our eventual death.
On the face of it, the numbers appear dire, and the worst possible scenarios seem credible. Most of all, the credibility of these worst case scenarios is supported by the near unanimity of the “experts”, who are all more or less telling us the same thing. Whether it be the outstandingly excellent Dr Fauci, or faceless data modelers and epidemiologists, the refrain is very similar – this thing is going to get a lot worse, we’re only at the beginning of the crisis, our hospitals will be overwhelmed and we can expect millions of Americans to die.
And here’s the thing. The experts aren’t unanimous, no more than is the case with climate change. And some of the god-like experts are proving themselves to have feet of clay, and actually to be mere flawed mortals. Perhaps that is only to be expected – pretty much every expert’s complex modeling is based on “best guess” assumptions that have little or no solid support because the needed data is not available.
It is a bit like being shown a room of 100 people and told that half of them are going to vote Republican and the other half Democrat in the next Presidential election, and then being asked to extrapolate from that as to the outcome of the election. But you don’t know if the 100 people are just returned from their local trade union meeting, their fundamentalist church service, or perhaps, from a Democrat or Republican party-member conference. You don’t know where in the country they’re from. You have no other demographic data other than what you can observe via a one-way mirror looking into the room. The data you do have is close to meaningless without more context.
And so it is with many of these models. They say things like “If most people get the virus but show no symptoms, this would mean…..”. Or “If each person infects three more people within three to four days, that would mean…..”. Or “If the death rate is 0.5% of all people infected, then we can expect…..”.
The problem is that the assumptions are just one paragraph of what becomes a multi-page document, covered with charts and really impressive mathematical equations full of symbols and Greek letters that make them totally impossible to comprehend for us lesser mortals. The problem is magnified because the results of the impressively complicated modeling are usually extremely dependent on the assumptions. As has been a computer aphorism forever, “garbage in, garbage out”.
Think of a garden hose – you are holding a hose nozzle and watering the garden. You just need to twist the nozzle an inch or two, and the spray of water moves several feet from one side to the other in the garden, going from one border to the other, all from a slight twist. It can be the same with the model assumptions – change your assumptions a tiny amount, and the model answer varies enormously.
Sometimes, a model might have half a dozen assumptions, and if they each have a +/- 25% error, then when you add up all the errors, if they are all on the same side, you end up not with a 150% error with a +/- 400% error (yes, the error could be even greater than this if there are exponential functions involved, and even some simple equations can vary enormously with small shifts in the values).
There’s another factor as well which came to light today. Programmers will understand this, but if you’re not a programmer, perhaps you have developed an Excel spreadsheet and created calculations in some of the cells.
Have you made the calculation a bit complicated with several sets of nested brackets, and a conditional “IF” statement or two? And then, after continuing to develop the spreadsheet some more, adding a few rows and columns, moving some data around, and so on and so forth, what are the chances that your formulas were correct in the first place and are still correct now? In my case, embarrassingly low!
Well, consider that then multiply it by almost literally a million to imagine the complexity in programming. There are many techniques used in programming to minimize and mitigate the inevitable abundance of errors that creep in, and the first and most basic of all of these is that the programming must be well documented. It has to be easy to understand, have a nice flow to it, and be sufficiently logical and intuitive so that not only can other people look at the programming and understand what it being done, but the original programmer too can keep track of what he is doing.
All of this has been a very lengthy introduction, to set the scene. I’m now coming to the point.
Earlier today, one of the people who had created what has become the most influential model and prediction of the impacts of the Covid-19 virus quietly recanted his findings. The London based research team had created their model and out the other end, the computer predicted 500,000 people would die in Britain.
The revised number? Now – barely a week later, the researcher is saying 20,000. That’s not a small revision. It isn’t even a big revision. It is a total game-changing stunning contradiction between the first number and the second number.
This extraordinary revision of course suffers from an internal challenge. If the first number was so wrong, how can we now trust the second number?
Which brings up the other impossible-to-accept part of this revision. Many researchers, with different conclusions, have asked for a chance to examine the actual computer code that this group has used to generate their projections. That is the underlying source, just like, for a Christian, the fundamental underlying truth comes from the Bible, not from priests.
But the researchers are refusing to release the source code for their model! They said, to justify this refusal, that the code was very confusing and complex, had not been documented, was thousands of lines long, and couldn’t be understood by anyone else.
There’s no good part of that statement, starting with the arrogant “we’re too clever and you’re too stupid to understand what we’ve done” assertion.
The thought of thousands of lines of confusing/complex, and – most of all – undocumented code makes my head hurt and almost burst! That is a beyond-bad-practice. As explained in the introduction, complicated code will inevitably come complete with a wide number of bugs – some obvious, and some “corner case” bugs that are less obvious but still impactful.
Is the entire model output based on erroneous model calculations (as well as bad assumptions)? Does the computer code actually do what the programmers expect it to do? We don’t only have a string of uncertain assumptions, but we then process it through code of unknown provenance, and the people who published their findings are refusing to allow anyone else to verify their process.
This model has been strongly influential on both the US and UK governments and the actions they have been taking to anticipate and respond to the impacts of the virus. What if it is entirely totally wrong?
Here’s the article that revealed this.
(Note – if you skipped some of this, the bottom line issue is that perhaps the death model being used by both the US and UK governments to justify the draconian shut-down orders may have been overstated by a factor of 25 times.)
The top countries in terms of infection density are essentially unchanged in position from yesterday, apart from Andorra and the Faeroes swapping places :
- San Marino/208 cases/the equivalent of 6,130 cases per million people (unchanged from yesterday)
- Vatican City/4/4,994 (unchanged)
- Faero Islands/140/2,865
Here are the top six major countries, showing death rates per million of population in the country. France moved up one place from yesterday, with the Netherlands reciprocally dropping one :
- San Marino/21 total deaths/the equivalent of 619 deaths per million (no change from yesterday)
- Italy/8,215 deaths/136 per million
To put those numbers into context, the death rates per million in US/UK/Canada are 4/9/1.
For major countries and/or outbreaks, and in general :
|Total Deaths/Percent of all Resolved Cases||10,030/10.3%||21,185/15.7%||23,976/16.3%|
|Active Cases (ie not yet died or cured)||147,362||333,497||382,258|
|US Cases/Deaths/Case rate per million||13,795/207/42||66,048/944/200||83,672/1,209/253|
|UK Cases/Deaths/Case rate per million||3,269/144/48||9,529/465/140||11,658/578/172|
|Canada Cases/Deaths/Case rate per million||873/12/23||3,409/36/90||4,043/39/107|
|Worst affected major country/case rate||Italy/679||Switzerland/1259||Switzerland/1365|
|Second worst country affected||Switzerland/488||Italy/1230||Italy/1333|
It is interesting to see how the “top five” countries have largely stayed the same over the last eight days. But look at the differences between them in terms of case growth, and compare them to the US/UK/Canada.
|Country||Case growth percent|
What can we learn from this? I’m particularly focused on Italy, and how its case growth rate is lower than the world average. There were four days in a row this week with new cases below 6,000 for Italy, but today saw it grow back up to 6,203 new cases, its second worst day ever. Nonetheless, it is no longer showing fearful increases every day.
The US growth is terrible, but keep in mind that as a percent of population, we’re much lower than the leading five countries. This also encourages me. Why? Because it makes me wonder if all of the five worst countries are seeing their growth curves start to level out.
The huge question though is how much of this leveling out (if indeed there is any) is due to lockdowns and social distancing, and how much is due to the virus starting to hit up against growth limits? The answer to that is totally unguessable (at least, by me) but whatever the answer might be, I feel it to be more on the good side of the curve than on the bad side.
I also keep coming back to another point, which with each passing day is, I feel, becoming more significant. No country, nowhere, has case rates even at a level of 1% of the population. The worst micro-countries have case rates of about 0.3 – 0.6%. The worst major countries have case rates no higher than 0.14% – in other words, one person in every 700 has the virus. Even New York City is thought to have no more than a 0.1% case rate.
Yes, I know that it is likely that for every person known to have the virus, maybe as many as ten or even more other people also have the virus and don’t realize it. But that is a great and very happy making thing. The more people that have the virus and don’t realize it, the greater our chances are, should we also contract it, of experiencing such a mild dose as not to realize it either. Plus all those people getting the virus, then recovering from it, are now people who won’t get the virus again, which will start to build up the “herd immunity” that promises to ultimately stop the virus in its tracks. The sooner the better, right? 🙂
New topic. As I seem to say most days, “it is the exponential thing”. In this case, yesterday, I made what I felt to be a realistic estimate for total new fatalities announced in the US on Wednesday. I guessed 180. The actual number was 247. We’re now at the point, with daily fatalities over 250, that Covid-19 has become the sixth most common cause of death in the US, and not yet showing any signs of stopping its rise up the scale.
Whatever the total will be/become, it is clearly not a trivial matter.
I predicted yesterday that in two or three days, the US would move from its then third place and become the number one country with the most Covid-19 cases. Well, what can I say. It is the exponential thing. The US moved convincingly into the lead today, with, as of the time of writing, already 83,672 total cases now reported, compared to China with 81,285.
Who Should Pay?
Apparently the $2 trillion bailout bill will be passed tomorrow.
And – unsurprisingly because yesterday there were no public copies of the bill available to read (not so sure about today) it appears the airlines are actually going to get every penny they asked for. $25 billion as a “grant” (aka gift) to use to pay staff, and another $25 billion in the form of loans or equity.
Yet again, the airline lobbyists have triumphed. There is no logic nor fairness, at all, to gifting the airlines $25 billion to pay their staff, but not doing the same for hotels, rental car companies, cruise lines, tour operators, bus companies, amusement park operators, or any other employer in the general travel and tourism field. For that matter, why should an airline be more important and more deserving than an office or a factory or a farm?
Why do the airlines get preferential treatment ahead of every other industry in the country? Should we mention also that the airlines, perhaps the most oligopolist of all the sectors of the travel industry, also treat us – the tax paying ultimate source of their $25 billion gift and further $25 billion loan, worse than any other company – certainly in the travel industry, and perhaps entirely across the board.
Even hotel resort fees pale in comparison to airline change fees, for example. How can it cost more money to change a ticket than it does to buy a new ticket? What would happen if Amazon started charging a fee to refund/return an item which was greater than the original cost of the item? Why didn’t Congress attach some “good behavior” requirements to the $25 billion gift?
And – most of all – why are the airlines balking at giving us refunds when they cancel their flights at present. That’s a legal requirement, but still they duck and dive and divert and delay.
Timings And Numbers
Los Angeles Mayor Eric Garcetti says he intends to keep his city on lockdown for another two months and maybe longer.
I wrote in my introduction about the problem with bad models and the results they provide us. The problem gets magnified when journalists then summarize and simplify the model output, and usually leave out the discussion about model assumptions and accuracy.
Here’s an example of a really bad story. On the face of it, it seems factual – it opens with the bold and clear statement that Germany’s death rate of 0.5% is lower than that of any other country. Except that, ooops – it isn’t.
As well as some small countries with zero death rates – statistical anomalies due to small sample sizes – there are countries with high case counts but with lower death rate percentages than Germany. Norway has a death rate of 0.4%. Australia is also 0.4%, and Israel is 0.3%.
Oh, one other number to look at. Germany’s death rate isn’t 0.5%. It is actually 0.61%.
So the entire article’s basic premise – that there is something unusual about Germany’s uniquely low death rate – is wrong right from its first sentence.
We don’t agree that of course more testing means that more cases of Covid-19 are detected, including very mild cases that might not otherwise be reported. This does mean that with more mild cases of the virus being reported, the death rate does decline. But what the article never tells us is how many people Germany has tested, and compared that to how many people have been tested in other countries (we vaguely understand that both Iceland and Italy have done a lot of testing too). Is Germany randomly stopping people on the street and testing them? Or is it, like every other country, only testing the people who ask to be or need to be tested? That is important to understand, too.
The total article is incomplete as well as wrong. And that’s from a generally respected source – NPR.
I visited the local Trader Joe’s and one of the local supermarkets today.
Both stores had much healthier supplies of everything than was the case when I was there on Saturday. Even toilet paper and pasta, but no disinfectant (just cleaner, which is a different product entirely).
Logic? What Logic?
President Truman had a sign on his desk “The buck stops here”. That’s a concept that has been enthusiastically but inappropriately adopted by everyone in the US and used to absolve themselves of blame for anything and everything. It is used as an excuse by everyone for everything. “It isn’t my fault”.
In particular, hospital administrators, earning an average of $343,000 a year and often more, say it isn’t their fault their hospitals are running out of personal protective equipment for their staff and out of respirators and consumables for their patients.
It is apparently no-one’s fault that hospitals failed to stock sufficient masks, or any of the other items that are now in critical supply, other than perhaps this or the previous federal administration.
People are dying, and many more will die in the future (see this terrible article) because no-one in any hospital chose to hold a larger supply of such things in inventory. Worse still, no-one responded to the first early warning signs about the dangers of the coronavirus when first expressed back in December by individuals and by Taiwan. When did hospitals first start worrying about their supplies of such equipment? In most cases, it seems like it is just in the last week or two that hospitals are starting to realize the new world in which they are expected to save lives, and now they’re scrambling to blame someone/anyone for their screwups.
I’ll concede that it is a sensible business decision to maintain inventory at appropriately low levels. And I’ll concede that rates of usage now are skyrocketing way beyond the normal levels of consumption. But – and this is the point that must be stressed – if a hospital made a calculated business decision to minimize its stocks of inventory, their administrators needed to also make a matching decision that they’ll have a hair trigger alert for any catastrophic event that might require a sudden huge increase in demand.
The concept of a global pandemic is not new. Disaster agencies models such events on a regular, almost annual, basis, and the typical scenario is a contagious respiratory virus, exactly as we now have. There’s been nothing to suggest that the Covid-19 virus isn’t anything other than a reasonably typical type of pandemic such as is regularly modeled and analyzed. Exponential growth is a core part of any such model. Maximum and above maximum demands on health care services, and in particular, every patient requiring abundant PPE, is one of the absolute standard elements in any such exercise.
There’s been a chorus of people shouting blame at President Trump, on “the buck stops here” basis. His administration, in turn, is quick to point out that they inherited the national emergency stock levels that were in place – in 2009 emergency stocks were seriously depleted and the previous administration, for the seven years that followed, consistently decided not to rebuilt the reserves. Why blame the present administration and its three years in office without sharing that blame with the previous administration and the seven years it passed over this item too?
But we don’t believe the stopped buck actually should travel that far. There’s a legal theory about “proximate cause”. There has to be a reasonably close and direct connection between a thing and an event to be able to include the thing in the range of potentially liable parties.
For example, if you fell asleep at the wheel of your car, drifted over the center line, crashed into an incoming car and killed all the people inside, their estate could sue you for falling asleep. But it couldn’t sue your spouse’s boss for unfairly criticizing your spouse at work the previous day, with the result being your spouse came home unhappy and got in an argument with you last night which caused you to spend a sleepless night, which is why you fell asleep behind the wheel of your car the next day. And similarly, it couldn’t sue the incumbent President, no matter who he or she is, or what party they belong to.
Unless there was a national law or regulation forbidding hospitals from holding quantities of things above a certain minimum, or a county/state or federal guarantee that any time hospitals needed more, it would be immediately available for them, no questions asked; the proximate cause for hospitals running out of product is the incompetency of their administrations in failing to make adequate contingency plans for the foreseeable scenario where they might need a huge increase in consumable items.
The chief executives of hospitals are paid as much as, and sometimes more than, $10 million a year (please see this article to understand exactly how extraordinarily high these people’s earnings are). They’re not even doctors, usually. Just administrators. How many more masks would a hospital have if each year the hospital redirected a single one of those ten million dollars away from its CEO and used it for mask and other consumable purchases, instead?
Or how about the hospitals with an abundance of hospital helicopters. This article makes the memorable statement that there are more hospital helicopters in the Dallas-Fort Worth Area than in all of Canada and Australia, while observing they are seldom needed or offer real benefits compared to regular ambulance transportation. How many masks and ventilators could the hospitals have acquired by just selling one of them.
There are class action lawsuits being filed against China for causing this virus. But how about some class action wrongful death lawsuits against hospital administrators for gross incompetence and a colossal failure of their “duty of care” for their patients? It is currently only a matter for conjecture as to how many people in total will die due to insufficient hospital supply inventories, but that number will definitely be more than zero.
One more thing – even if there was an obligation on this and the previous administration in DC to bulk up national reserve supplies of these items, that doesn’t let the hospitals and their administrators off the hook. It was not a secret that the reserves had been diminished and not restored. Why didn’t the hospitals make this a huge big issue, or why didn’t they recognize the problem and become either self-reliant or create industry groups and as a group do the same thing?
Stop blaming this or the previous President. Blame the nameless overpaid person who heads your local hospital.
Virus? What Virus?
Here’s an interesting paper that models a possible set of situations and suggests that each time we relax our movement controls and social distancing, the virus will rush back into the community, requiring us to then reintroduce those controls again. This on-again/off-again process would naturally span up to twelve or more cycles and two years, and only end when there was sufficient “herd immunity” in the country so that future infections never got off the ground, because most people had already acquired immunity and so didn’t get re-infected and continue the virus spread.
The “short-cut” which would massively improve this projection is either a good vaccine being broadly available, or easy/quick treatment that increases the number of patients our hospitals can handle.
Needless to say, the thought of two years of disruption, lurching from crisis to crisis to the next crisis, is terribly unappealing.
Unstated in this paper is the interesting point that developing herd immunity should be the goal of any program and the only viable solution, either through exposure or vaccine. Maybe the first crazy concept of Boris Johnson and his advisors in Britain – to just let the virus run its course – wasn’t entirely wrong, other than in the implementation thereof.
Also unstated – perhaps because it is obvious – is that if we simply observe a reduction in new cases and say “Mission Complete” and abandon all caution, we’ll be as spectacularly wrong as was President Bush when indirectly making that claim with the Gulf conflicts. Until we have herd immunity, any respite will only be temporary and as soon as a new “Patient Zero” arrives into the US, we’ll restart a similar cycle yet again. And with illegal immigration, even if we do 100% testing of every arriving “normal” person (which is wildly impossible anyway), the people crossing the border outside of such controls will inevitably bring the infection back into the country.
The other interesting thing is that the modelers are using a trigger point for switching on the lockdown measures as being 3750 cases per million “people”. We’re not sure if they mean all people, or just adults (elsewhere they refer to adults in particular). So maybe the trigger point is about 7500 per million people if half of all people are adults.
The trigger point for switching off the lockdowns is 1,000 cases per million adults – call it 2,000 per million of the population as a whole.
How realistic are these numbers? At present, the US as a whole has a case rate of 253 per million people, so that is 30 times lower than the point where we should start locking the country down. There are regional variations, of course, but even the hot spot of New York City has only something like 25,000 cases in a population of about 25 million people, or a case rate of 1,000 per million. So, even NYC has a case rate massively below the lockdown number (7500) and indeed, its case rate is also below the end-lockdown number (2000).
We’re neither endorsing nor lampooning the study. Some of its assumptions are sensible or reasonable, others are unknowable. But it does paint a totally different picture to that portrayed elsewhere.
Some good news in the last 24 hours. It appears that the virus is slowly rather than rapidly mutating. The reason this is good news is because it means that if a vaccine can be developed (and that is still an unknown “if”) then it is likely to last for more than just one or even two years.
Viruses mutate at different rates. Smallpox, for example, changes very slowly, which is why a smallpox vaccination can last for so long. Regular ‘flu (which is not a coronavirus, by the way) mutates quite quickly, which is why a different ‘flu shot is needed every year.
The Dow has now had three positive days in a row – the first such time this has happened since things became dire. Today saw another rise of 6.4%, closing at 22,552.
We remain a very long way away from where it was before things came crashing down, but there’s nothing bad that can be said about a 6.4% rise in a single day. Could we have some more of the same, please.
Masks, masks, and respirators. A topic we’re all obsessing about at present. But have you ever paused to wonder about the origin of this currently vital thing? Here’s an interesting article to enjoy.
Please stay happy and healthy; all going well, I’ll be back again tomorrow.
Please click here for a listing of all our Covid-19 articles.
1 thought on “Covid-19 Diary : Thursday 26 March 2020”
Thank you for all your hard work and thoughtful analysis. They say that information can help diminish fear, and your diary is useful food for thought.