All the good stuff, caveats, code, data sources and explanations are linked, some in Update III, and the most important in Update II, Update IV, Update V, Update VI, Update VII, Update VIII, Update IX, Update X, Update XI, Bayes Theorem & Coronavirus, and the Sanity Check Perspective, so go to them first before asking what-about-this-and-that. Skip to the bottom for the latest model. Thanks to everybody emailing me sources, including Ted Poppke, Jeff Jorgensen, Jim Fedako, Joe Bastardi, Philip Pilkington, John Buckner, Harry Goff, John Goetz, Warren McGee, Robert Kinney III. https://wmbriggs.com/post/30606/. Sorry I’m slow answering emails.
“You can discover what your enemy fears most by observing the means he uses to frighten you.”
— Eric Hoffer
Juicing Numbers & Fear
Incidentally, the doc video we told you to watch last week before it was purged—was purged. YouTube must have CDC zampolit on staff to decide which videos are “legitimate” and which must be memory-holed. I’m too lazy to search for the video elsewhere, but I’d guess it’s on BitChute by now.
— Alex Berenson (@AlexBerenson) April 30, 2020
“1/ 61,000 and counting. And counting. Colorado is now overruling attending physicians to classify nursing home deaths as #COVID.”
It wasn’t only death models that were off. It was hospital resources, too.
Hospitals in California have been half empty this whole time pic.twitter.com/8GL2BLypqV
— Elon Musk (@elonmusk) April 30, 2020
Naturally, some will claim the lockdowns account for this. But they say this by assuming it is true. I mean, they begin by assuming lockdowns must work, and then conclude that lockdowns worked. When, in fact, lockdowns cause harm, and the models were just wrong. The only other possible claim is to say lockdowns worked “much better” than predicted, which is absurd. The models assumed lockdowns. Since they assumed it, and they were wrong, it must mean the lockdown assumption was wrong, or other parts of the model were.
In any case, something was wrong, meaning we should not have trusted the models.
Later this week, or maybe next, I’ll show you that WHO agrees with this.
From FoxRiverFlood comes this quip: “Flatten the curve” is good marketing. If they just called it “Extend the Epidemic Indefinitely” people might not have bought in.
300+ workers at a Missouri pork plant have tested positive for #COVID19. Most had no symptoms.
Meat plant workers often work shoulder to shoulder and are overwhelmingly immigrants or minorities. President Trump has ordered plants to stay open, despite at least 20 deaths. pic.twitter.com/vnjRT1OFqy
— AJ+ (@ajplus) May 4, 2020
Most people would take this as good news, seeing the majority who get it feel little or nothing.
Not our stalwart media, though. They managed to find a way to turn it into FEAR.
Whose Right Numbers
We’ve been using the COVID Tracking Project for our reports. As I’ve told you from day one, other sources differ. As of 8 PM Monday EST, our source says 62,806 reported deaths.
The CDC, on one of its sites, says, as of 4 May, 38,576.
Later this week, we’ll have a guest post on all this. For now we can note this is a bit of a difference, no?
The CDC also reports deaths with pneumonia and COVID at 17,122. Now I’d think this is a proper “and”, meaning you can’t add these to the COVID total, but I don’t know. It may be that their report of COVID alone is without pneumonia. I can’t make out from their description what they mean. But if that’s right, then there would 55,698 total reported deaths. Still 7 thousand less than our source.
Kind of odd, too, there are only 5,886 reported flu deaths since 1 February. There were 92,615 reports deaths “with” (their word) pneumonia , flu, “or” COVID.
As I have been saying, to sort all this, we’re really going to have to look to all cause-deaths. CDC says 739,600. I guess from 1 February, since that’s when the chart starts. But see this post.
We’ll be real careful about “excess deaths”, though, as this other official plot from the CDC indicates.
In other data, the 2017-2018 flu season looks worse, but that’s neither here no there for now. It’s the idea of “excess deaths” we have to be careful about. Deaths not above the model are counted as “excess”. But the threshold undulates, acknowledging the seasonal winter peak. The “excesses” are usually ascribed to flu, now coronavirus.
But what’s causing those other deaths in the peak? Not flu? Not corona? By assumption, yes: those deaths are caused by other things, even in the peak. How we do know the assumption is correct, given how variable death certificates are with flu? We don’t. It’s always an estimate.
In other words, and as I have been saying, it’s a mess.
Here, anyway, are the per 100,000 weekly deaths going back to 2009, with the three official sources. The latest numbers, as CDC says, will be delayed somewhat. Dashed line all deaths minus COVID deaths. Details and sources here.
With under-reporting, there is still a chance the dreaded coronavirus will beat 2017-2018 flu season. Getting harder and harder, though. Be careful how you place your bets.
Models of Doom
The MOBS model from the Network Science Institute at Northeastern University also estimates that there will be about 89,000 deaths by mid-May if stay-at-home orders remain in place.
That death toll would increase to over one million in an unmitigated scenario, according to the projections that are among those used by the CDC to forecast the pandemic.
Nice source for various predictions. Compare these to the actual totals below. The colors are slightly confusing (to my eye), but it appears UT Austin might be closest. These are all snapshots from maybe a week or two ago, as all models change continuously.
How about this breathless headline (from a “conservative” site)? Leaked CDC Model Projects: Next Month Daily Deaths Will Reach 3,000.
The Times somehow got hold of a FEMA chart based on numbers put together by the CDC. (How they got hold of it is itself an interesting question.) Most of us are already grimly inured to daily reports of 1,500-2,000 dead and 30,000 more infected. Imagine how much more inured we’ll be on June 1 if this data pans out. One death is a tragedy; 3,000 deaths every 24 hours is a statistic.
So this is a model from Department of Homeland Groping, apparently, run on 1 May.
Wowza! Bring out your dead! What I find funny is the nervous guy writing about this (at that link and the AI bot at NYT) focuses on the forecast of daily deaths increasing increasing increasing, with no end in sight!, while not noticing the model has been absolute shit up to this point.
Somehow that the model has predicted so badly so far is proof IT’S WORSE THAN WE THOUGHT. You can make similar claims about our naive model, since it consistently under-predicts, but at least we’re not projecting forever growth, and we got the shape right. This FEMA model makes it look like this virus, unique among all others in history, will accelerate, not in a second wave, but right now. As many as, what, 30,000 deaths a day by 1 June? That smell right to you?
You can read all the political hand-wringing blah-blah-blah at either side about how this model was “leaked”. I refuse to think about it.
I repeat, for the nth time, that these are naive models of reports, not actual totals.
New totals forecasted: 3.9 million cases, 275 thousand deaths. Last week “3.3 million; new total reported deaths 242 thousand.” About the same under-prediction on reported deaths as the week before. Not too bad for a naive model.
As I’ve been saying week after week after…(how many weeks now?)…the true cases must be much larger than the reported cases. I will repeat that word-for-word for the US. In any case, here’s how we know. This is the reported deaths divided by reported cases.
There is no way the virus is killing north of 7% of those it infects. This graph, using official numbers, proves that (1) reported cases are too low, or (2) reported deaths are too high, or (3) both.
Using the same trick as last time, and using the total reported deaths (so far), assuming the virus actually kills 1% or 0.5% of those it infects, which are on the high end, then there must be 25 to 51 million actual cases in the world. More if kill rate is lower. Assuming the reports deaths match the actual totals. If they’re too high, then so are these estimates.
Again, watch for the media to tout reported cases over deaths in the next couple of weeks. Deaths (as we’ll see) doesn’t look like it will help them keep the panic going.
Reported daily cases:
The increased testing it clear enough. It also means, as with US below, this model is inadequate for reported cases. See US discussion for more on this.
Reported daily cases:
The up-and-down we’ve discussed many times, but it’s perhaps clearer with the US. I’ve explained it below.
Anyway, ignore the model and look only at the data. The trend is clear enough, though noisy.
USA! USA! USA!
New forecasted totals: 1.23 million reported cases, 67 thousand reported deaths. Last week: “New total reported cases 1.1 million; new total reported deaths 57 thousand.”
Obviously with testing ramping up, this simple model is inadequate at capturing reported cases. Watch the news media, though. If they haven’t deaths to report, they will instead emphasize cases, hoping to show how rapidly the bug is spreading. When, of course, at least a good portion of tests reveal cases that were already there.
As I’ve been saying week after week after…(how many weeks now?)…the true cases must be much larger than the reported cases. We’ll prove this again for the US after the daily reported cases.
Reported daily cases:
You can see the weekly signal in the dailies: these are, after all, reports, and even under lockdown, we’re still operating on a seven-day basis. Even with increased testing, the trend is down.
It’s also clear the model will be way off in the total reported cases. That total depends on our dear leaders and how much testing they will require. I’m open to suggestions about that.
Reported deaths divided by reported cases:
This virus is not killing more than 5% of the people it infects in the US, as the evidence above, and everywhere else, indicates. This graph, using official numbers, proves that (1) reported cases are too low, or (2) reported deaths are too high, or (3) both.
Using the same trick as last time, and using the total reported deaths (so far), assuming the virus actually kills 1% or 0.5% of those it infects, which are on the high end, then there must be 6.2 to 12.4 million actual cases in the US now. There could and will be more later.
This assumes the reported deaths match the actual totals. If they’re too high, then so are these estimates.
Reported daily deaths (reminder of source for those who will gasp).
In a way, I had bad luck picking Tuesdays as update day. It just happened the first time, and then it made sense to stick to a weekly schedule for fair comparisons. But because reports are also weekly, and tend to be lower for deaths on weekends, the reports are always lowest Monday nights. So we’re always hitting low points, which paint a brighter picture.
Meaning I am not predicting that new reported deaths will drop to 0 in a few days time. In fact, reports will almost certainly spike back up—but not way back up. The general trend is correct. And in any case, the black line is the data itself.
To support this site and its wholly independent host using credit card or PayPal (in any amount) click here