A lot of my discussion these last several weeks has centered around the various predictive models used to inform us about the virus and what to expect. And of that discussion, most of my observations have been at the colossal errors the models seem to contain and the extraordinary revisions made to them, with few people pausing to consider what it means to the validity of a model when it needs to be revised so materially.
I was thinking of this in another context over the weekend. Each Easter the British radio station Classic FM publishes and plays the result of its annual surveys of the most popular classical music, featuring the 300 top scoring pieces. It has been doing this for 25 years, and claims it is the world’s biggest poll of classical music tastes.
Now, the thing about this survey is there is essentially no new classical music (the station does have a very wide definition of what classical music is and includes film scores, but in general, the major part of the classical music field is unchanging and hasn’t changed for 100 or more years). It is a very wide survey of a relatively unchanging topic, and to a relatively unchanging audience (people who listen to the station). And so you’d expect a close consistency of results each year.
But, every year, there is a mix of pieces that stay about the same, and some that move somewhat up or down the rankings. That’s all to be expected. But, in addition, there are huge shifts of rankings with some pieces, too. For example this year, Max Bruch’s Scottish Fantasy dropped from #100 to #264. Some Mozart and Beethoven pieces also dropped over 100 places.
The largest rise was from “not even on the list last year” to #182 for a Bach organ work (Toccata/Adagio/Fugue BWV 564).
Some of the pieces near the top also shifted position massively, for example, the Peer Gynt suite (Grieg) dropped from #20 to #64 and The Carnival of the Animals (Saint-Saens) dropped from #41 to #114, while Beethoven’s Ninth Symphony went from #8 to #2 (this probably represents more of a shift of vote percent than some of the larger movements of the less prominent pieces).
So a musicologist might now start trying to understand why there have been such “seismic shifts” in popularity this year. But a statistician would hopefully be able to tell the musicologist that sometimes even very important and significant seeming shifts are actually meaningless and random.
And therein lies much of the challenge in trying to “read the tea leaves” of all the data that is flooding in about this virus pandemic. How much of it looks significant but is actually meaningless and random? How can one distinguish between that type of data and perhaps the more obscured but important data?
As one more example, sometimes the most important data is obscured – for example, to me, the most significant thing was akin to the Sherlock Holmes mystery where the great detective works out the mystery not by what happened, but by what did not happen – the dog that did not bark. I thought it significant that the top scoring most popular piece of music stayed steady in its #1 position yet again this year (The Lark Ascending by Ralph Vaughan Williams). I believe it has held that position consistently for some years now.
I talk further about three pieces of Vaughan Williams’ music (all of which were in highly rated in the Top 300, in places 1, 5 and 132) in the closing notes of today’s diary entry. I do hope you’ll choose to read (or skip) down to those comments.
If you’d like to listen to Classic FM (wall to wall light classical music with nice friendly announcers and English news on the hour most hours), you could add the Global Player skill to your Alexa device, and then can play it through there, or you can play it through your computer from the Classic FM linked home page, above.
Here are the rankings for the seven states of any size with the highest infection rates. There is no change in ranking, and four of the countries again didn’t register any new cases today. The other three had modest increases.
- San Marino/356 cases/the equivalent of 10,492 cases per million people (unchanged)
- Vatican City/8 cases/9,988 cases per million (unchanged)
- Gibraltar/129/3,829 (unchanged)
- Faroe Islands/184/3,766 (unchanged)
Here are the top six major countries, showing death rates per million of population in the country. As has been expected, the UK has now dislodged The Netherlands for the fifth place, although it has a long way to go before it can reach France :
- Spain/17,756 deaths/380 deaths per million
- Italy/20,465 deaths/338 deaths per million
- United Kingdom/11,329/167
To put those numbers into context, the death rates per million in the US/Canada are 71/20. The world average (not a very reliable number) is 15.3.
For major countries and/or outbreaks, and in general :
|Total Deaths/Percent of all Resolved Cases||74,654||114,194/21.3%||119,598|
|Active Cases (ie not yet died or cured)||992,847||1,314,752||1,360,249|
|US Cases/Deaths/Case rate per million||367,004/10,871/1,109||560,300/22,105/1,693||586/866/23,621/1,773|
|UK Cases/Deaths/Case rate per million||51,608/5,373/760||84,279/10,612/1,241||88,621/11,329/1,305|
|Canada Cases/Deaths/Case rate per million||16,667/323/442||24,383/717/646||25,680/780/680|
|Worst affected major country/case rate||Spain/2,923||Spain/3,568||Spain/3,638|
|Second worst country affected||Switzerland/2,502||Switzerland/2,937||Switzerland/2,968|
Two changes above – Belgium has now displaced Italy and taken the third place for most cases/population, and Ireland has now dislodged France for the fifth place.
I have several times observed some obvious problems with how the recovered/discharged patient count is calculated, and today I noticed that Worldometers is now showing “N/A” rather than an actual number for the UK – the country I’ve singled out as worst offender. So I’m discontinuing that measurement, due to it being way too unreliable for any purpose at all.
Yesterday we noticed the flattening of daily new cases reported, and as you can see, there is a probable lessening of total worldwide daily deaths being reported as well. Both trends are appearing in the US numbers, too.
Sweden continues to attract media attention for its much less restrictive approach to social distancing, and this article says their Prime Minister is now thinking of tightening controls. This is at a time when some countries in Europe are now starting to relax controls.
But as can be seen, it would be ironic if Sweden tightened controls not only when other countries are now going the opposite way, but when the last three days have seen a collapse of the former noticeable discrepancy in death rates between Norway and Sweden.
In total, Sweden is experiencing deaths at about 3.7 times the rate of Norway, which is a number that continues to fascinate me. Depending on if your glass is half full or half empty, this is either a very small or a huge difference.
Certainly, it is an appreciable difference, and probably it is statistically significant. But when you think about how some of the model studies have excused collapses in their projected death rates that have dropped in the order of 50 or more times lower than initial projections as being due to social distancing being better than they thought it would be, one is left with the uncomfortable but very familiar feeling that these models are as much snake oil as they are reality.
Yesterday saw a new record for low numbers of air passengers in the US, with the daily count, per TSA, dropping to 3.7% of last year’s comparable day volume.
I Am Not a Doctor, But….
Some people have taken me to task for my regular positive references to the potential being displayed by hydroxychloroquine as a possible treatment for Covid-19 sufferers. They choose to ignore the layers of caution I appended to the topic when I first brought it up (although, in fairness to them, I doubt if anyone reads every word of every diary entry!).
I have no axe to grind on this matter other than the desire to separate real truth from political and professional posturing. And so today here is news of a study that was testing high doses of chloroquine and was terminated prior to completion due to some patients developing irregular heart rates.
But, do you see the correct context in which to interpret this study? First, it is a test of chloroquine, not hydroxychloroquine. Similar name, similar drug, but not identical. Second, the problem was with patients taking high (ie above recommended) levels of chloroquine.
Nothing in this study invalidates the potential for regular doses of hydroxychloroquine. And, as I said before, if you are countenancing this, of course you should make yourself familiar with the appropriate dosages, the side effects and possible negative conditions and reactions with other drugs you might be experiencing/taking.
I continue to feel this may be a promising treatment, and have yet to see any convincing study to the contrary, just a flow of inconclusive but positive outcomes.
Talking about promising treatments, here’s a nice professional journal piece updating on various studies currently underway.
Do you have to go to work at present? If so, here’s something to show your boss – some possibly easy steps to reduce the risks you are all confronting.
Who Should Pay?
So you fail to observe the rules for social distancing? What should your punishment be? If you’re in India, it might feel like you’re back in elementary school.
Here’s an interesting article – where are the trillions of dollars in government support actually going? It seems we may never accurately know the full answer to that question. Nothing like unaccountable government spending, is there…..
Timings And Numbers
The widely respected (in the US) IHME study updated its projections today. It shows a slightly greater number of US deaths in total, now predicting 68,841 deaths in total, with the peak death day today, with a projected 2,150 deaths today. It seems the actual count for today may be more like 1,535, which is much better news than the projection.
The study is quite likely not so respected in the UK. It was projecting 62,500 deaths, then reduced that to 37,500, and now has reduced that again down to 23,800. These changes all happened in less than a week, with no changes to social distancing in the UK, and most importantly, both of the revisions downwards were outside of the 95% confidence range of the earlier numbers stated. For example, when they were projecting 37,500 deaths, they were 95% confident that the deaths would be at the very least, 26,150. But now they are saying the total deaths will be 23,800, a number that is way below their 95% confidence range.
So they were 95% confident of being sort of right, but were actually wrong, twice. There is only one chance in 400 that they could be that wrong, and yet that is the reality of their revisions.
Does that make you feel confident about the accuracy of their models?
Here’s an interesting article that links to a McKinsey study of how the travel industry might recover. Here is a direct link to the McKinsey report.
How vulnerable is your state to the economic effects of our shutdown? This is an interesting table.
We continue to treat most data from China as being without value, but perhaps there is some representation of reality, whatever it actually is, in this article and the data within it. The interesting point to me is how most new cases (but not all!) in China are from people flying in to the country.
This suggests what we might expect in the future, ourselves. If we ever get to a point where the disease is almost eliminated, we will still be under constant threat of it being brought back in by people arriving (or returning) from other countries. Unless we have an ultra-rigorous procedure for isolating all such people for quarantine, or highly accurate testing that can be administered at the airport, we’ll be allowing the virus to re-establish itself again.
But who wants to travel internationally if it means a 14 day quarantine in each country they visit, plus another 14 day quarantine upon returning home again?
Supermarkets are continuing to reduce the number of people they’ll allow in a store at one time, and to come up with new ways to ensure we keep well apart from each other. For me, the greatest part of shopping now is the growing use of masks.
When should we start to think about removing some of our social distancing restrictions? WHO have come up with six requirements they say should be met before lockdowns can be lifted.
- Transmission is controlled;
- Health system capacities are in place to test, isolate, and treat every case and trace every contact;
- Outbreak risks are minimized in special settings like health facilities and nursing homes;
- Preventive measures are in place in workplaces, schools, and other essential places;
- Importation risks can be managed;
- Communities are fully educated, engaged and empowered to adjust to the new norm.
The WHO list reads well, but is nonsensical (sort of inevitable, due to where it is coming from). How can you evaluate and give pass/fail grades to most of these requirements? Start with the first point – what does “transmission is controlled” actually mean? Zero new cases a day? One? One hundred? Or???
The second point is actually a nice absolute measure.
But what does it mean to “minimize” risks in point three? How much reduction is acceptable, and how much is not? The same for what is a “preventive measure” in point four and how successful must it be?
The point five phrasing of “importation risks can be managed” is curious. It doesn’t say “eliminated”. It says “managed”. How many new cases are we allowed to slip into the country each day and still rate a pass on this point?
And lastly, what a beautiful bit of politically correct nonsense the last point is. What does it actually mean, and how is it evaluated and scored?
Yet again, WHO succeed only in highlighting their woeful and utter incompetence and irrelevance. The only thing they’re good at is making politically correct and meaningless (or am I repeating myself) statements.
Should I add that on the points that might allow for pass/fail scoring, I’d consistently rank us as currently failing.
Here’s a fascinating article that hints at more than it says about how the Mossad helped to ensure Israel got everything it needed for its own national battle against the virus.
Logic? What Logic?
Here is a great article that starts off seeming as if it will quickly become impenetrably complicated, but which succeeds in explaining some fairly esoteric issues to do with statistics.
I wrote, some days ago, about the laughable nature of the generally favored IHME projections that showed no further deaths from Covid-19, anywhere in the US, any time after 21 June. This seems to fly in the face of common sense.
Today the Wall St Journal picked up on this curious element of the IHME projection. But it doesn’t press the point, merely saying that IHME explains this assumption in the FAQ.
This is what IHME says to support their claim of zero deaths after 21 June
Our forecasts of zero deaths in July and August assume that appropriate measures are put in place to guard against the reintroduction of COVID-19 from another state or country. These measures may include mass screening, contact tracing, testing of all individuals entering the country, and quarantine of people who test positive. Details on what these strategies need to be will be analyzed in future editions of the forecasts.
In other words, “we assume a perfect outcome based on an assumption of perfect conditions, but we’ve no idea what these perfect conditions might be”. This is a circular argument and distressingly nonsensical.
I’m as guilty as anyone else for urging on the development of a vaccine for the virus as quickly as possible. But we should keep in mind that risk minimization takes time, and if we rush too much, who only knows what problems might assail us with the vaccine. This is particularly a concern when it is a vaccine that would likely be rushed into service, with hundreds of millions of doses very quickly administered to people.
I’m not an anti-vaxxer at all, but I’ll not be rushing to be the first vaccinated with whatever vaccine may eventually be released.
After a great week last week, the Dow started this week off negatively; giving back 329 of the points gained last week, and after a 1.4% drop, closing at 23,391.
The headlines boasted how a trade union could do a better job of finding and buying masks than the federal government. There turned out to be just one small problem.
The 39 million N95 masks that SEIU were about to buy didn’t exist. Indeed, it was obvious right from the get-go the masks were a fiction – they were described as genuine 3M N95 masks, but 3M only made 20 million such masks last year in total.
Delta has made a change to its boarding process, which slightly reduces the risk to passengers during the boarding process (assuming there actually still are any passengers on the flights it continues to operate).
I mentioned music by Ralph Vaughan Williams in my opening comments. When listening to the most popular piece of music on the Classic FM Top 300 list, his The Lark Ascending, I was again struck by its peace and tranquility. This is quite a contrast to the troubled world at present, and was originally written as a contrast to the troubled world at the time he wrote the piece – in 1914 at the start of World War One.
I’ve always felt the English countryside to be the most beautiful in the world, and this music is a perfect expression of its beauty. It may also calm one in these troubled times.
Two of his other popular pieces have a similar effect. If you like The Lark Ascending, try his Fantasia on a Theme of Thomas Tallis, written after the composer had spent a year learning orchestration with one of the all-time great masters (Maurice Ravel). Listen to the langorous lushness of this piece (it gets off to a slow start then swells), and again draw hope in our path back to the world we knew.
The third is based on a tune that some say may have been written by King Henry VIII when he was courting Anne Boleyn. It was already well known in Shakespeare’s time, mentioned twice in his plays. The tune itself usually scores in the top very few of any survey of “the world’s best tunes” but was the third of Vaughan Williams’ pieces in the Top 300. I refer, of course, to his Fantasia on Greensleeves.
The links to the three pieces take you to YouTube recorded performances. There are many to choose from of course, so have fun roaming widely to find your respective favorites. And as a bonus for Anglophiles, here’s one more piece, offered equally for the upbeat music and for the lovely pictures.
I better stop before I go quite over the top with English music (ooops, I just did!). Let’s all hurry back to England, just as soon as we can (whenever that may be).
Please stay happy and healthy; all going well, I’ll be back again tomorrow