Monday, March 30, 2015

Welcome To The Party




Hi. Nice to be back. Fully and completely.

I was going to spend my day off tomorrow coming back to re-visit the entire (and ongoing) Ebola outbreak.

But today, on his blog, American Mercenary went there.

And I replied in comments.

Which occasioned his reply, and now this post, mainly because it's too frickin' big to fit there, it bogged down and choked his Disqus comment server, and I needed this excuse to kick me in the pants and drag me back to this whole thing eventually.

So, let's get to the rat killin':

AM wrote:
Aesop, I've always been a bit in awe of your ability to be so certain about the uncertain.

Why stop with "less than one third"? Why not say "Less than one tenth?" or "Less than one hundredth?" or "Less than one thousandth?"

What gives you any certainty for a bounding of the uncertainty?

It is an interesting question, isn't it?

The same question, asked a different way, is "how do you know even a fourth of cases were accurately reported?" Or "How do you know the population of any given country isn't complete and total bullshit?

And to go one step further, do you know if the rate of under reporting has been consistent or inconsistent? If it is consistent we can still use under reported numbers to track the progress of the epidemic. If the reporting has not been consistent, how would you know?

Here goes: 

I appreciate the snark, but unfortunately, WHO selected the "less than 1/3rd" Fudge Factor for those reports some months back, which has been noted on this blog over and over. In fact, it was revised to a higher Fudge Factor the worse things got, because it became crystal clear even to the doorknobs at UN/WHO that the locals were deliberately spinning numbers out their fourth points of contact as a matter of course, and the previous "1/2" Fudge Factor was No Longer Operative. Unless they've taken away the breadcrumbs, the memos to that effect from last July/August/September/October etc. are still available online, including on the Wikipedia page concerning the outbreak:
Note: the CDC currently estimates that actual cases in Liberia, Sierra Leone, and Guinea are two to three times higher than officially reported numbers[1][2][3].
(Nota bene that those Wiki links are to a CDC report, Science Magazine, and the LATimes' story on the subject, respectively. From last September and October. So as far as plucking that factor out of my back end, my hands are clean. I can still pull this out of recollection, and I haven't touched the issue since about last December, IIRC.)

And the exact point of the exercise has been that the reporting of the statistics by the involved countries was and is complete and utter bullshit, then, now, and ever shall be. The "1/3" Fudge Factor is the minimum of how much BS to ascribe to the stated numbers. The upper limit is bounded only by the actual population on the ground.

IIRC, I think I only made that exact point about 200 times amidst the height of the crisis.

Like here.
Here.
Or here.
Or probably any number of 200 other posts anytime after July/August.

The entire key to managing this outbreak, by all expert testimony of world-class epidemiologists, is having and effectively deploying sufficient resources, which requires accurate intelligence about where and how bad the outbreak is, and is progressing.
And actually having those resources.

Which has been and continues to be the exact things that have never existed anywhere in West Africa from 400 B.C. to date, inclusive, not least of which because you're dealing with people who have to wear open-toe sandals to count to 20, and have literacy and numeracy rates that make Appalachia look like Caltech and MIT, before we even get into the self-serving corruption problems endemic to the world's poorest former colonies, or the abysmal lack of response to this from outside until well past the date it would have mattered.

Has this spiraled along the exponential mathematical pathway it could have done by now?
Thankfully not. Huzzah.

Why didn't it?
No frickin' idea, anywhere, from anyone.
(Personally, I'm hoping it was the Invisible Hand of the Grim Reaper, selecting out hordes of Darwin Award Finalists with carefree abandon, particularly the 50% of people there who are sure Ebola is caused by witchcraft, and the 45% who are convinced it was cause by the United States practicing witchcraft, but that's just my mischievous nature breaking free for a moment in the sun.)

How bad was it then, and how bad is it now?
We know precisely two things: Jack, and sh*t.

So we don't know where it went, why, or how, and we don't know why it stopped or what was most effective, because we never had any accurate data to go by, not even roughly.
Which makes any official pronouncements about the outbreak from on high as relevant to actual medical science, as astrology forecasts are to actual astronomy.

PAY ATTENTION PLEASE:
This is exactly why the original pronouncements from CDC, like "This will never get to the US", and "We KNOW how to handle and contain this disease", turned out to be total horseshit too: they don't know what they're talking about, and they don't KNOW they don't know what they're talking about, because they're basing their announcements on prognostication based on happygas, NOT on actual scientific analysis or sober reflection on reality and facts on the ground. AND THEY NEVER HAVE.

In the West, it's even worse: the US effectively managed, what, eight or nine simultaneous cases?
And our max capacity is...eleven. Then, now, ever.
A number they had in any of those three countries last March or April, three months before anyone much cared.

The greater point was and is:
This disease will get out, again, and we'll fail to deal with it, again, and next time, once it gets to 12 cases, that city is f*cked.
Twelve cases apiece in two cities and that state is f*cked.
Twelve cases apiece in three cities, and we're Liberia, or Sierra Leone.

And if we're very very lucky then, you might contain it in just one time zone, or on one side or the other of the Mississippi, or the Rockies. At gunpoint.
If it gets to any African or Asian megalopolii, where the number of potentially infected people who can spring for a plane ticket is seven orders of magnitude more than all of the rural African continent, fuggedaboudit.

So how many MOPP suits does your unit have, and how long do they work saturated in infected blood?
And what happens if the factory where they make them is inside the Hot Zone next time?

This is a temporary pause, and we neither know why it paused, or when or where it will return.
We only know that unlike all previous outbreaks, it hasn't burned out and gone away, and now it probably never will.

And as long as we allow them to do so, it's a virtual certainty that people will continue to get on planes there, and bring the disease here and to Europe, in both of which where it won't be noticed immediately, and will thus propagate and spread, and that if it does so over the 21-day incubation period, it could infect enough people that, just like in Liberia, Sierra Leone, and Guinea, it will suddenly spread like wildfire, in cities where the Black Death isn't just hyperbole, but merely dimly recalled actual history. And then the Gods Of The Copybook Headings are going to have their due, in exponential hyperbolic glee, without any nod towards dismissive snark, nor any regard of officialdom's irrational exuberance.

And the bodies of the disbelieving and the slow to respond will be stacked like cordwood.



AM is, by any manner of measurement I can devise, a thoughtful and bright guy, a serving Army officer who evidently survived the recent purges, and in all likelihood, destined for field-grade and higher service.

So while we may occasionally disagree on finer points, I wish him nothing but the best, and hold his posts and comments in high regard, because he takes things apart and puts them together rationally, as a rule, if not entirely. If it were otherwise, he wouldn't be over there to the right on my Blog Roll.

But having noted that the figures we were being fed were pure crapola, pretty much for half a year, which was reported on, sourced, and fact-checked ad infinitum, I can't begin to understand when or why he ever thought it was otherwise, which is the kiss of death to ever understanding this disease or combating it scientifically and efficiently, let alone putting the slightest shred of faith in any official pronouncements regarding it.

My official policy remains: "Ebola: Run for your lives." because the signal to noise ratio is at best 1:10,000, and worse than that the higher up on the panjandrum scale any of TPTB reside.

I didn't think this was news anywhere, but evidently I haven't made the point frequently, forcefully, and cogently enough yet. My apologies for tiptoeing around that 800-pound gorilla.

Comments are wide open.

14 comments:

GamegetterII said...

Good to see you're back-fully and completely.
Have you by chance read this...

FREETOWN (Reuters) – A three-day lockdown in Sierra Leone has exposed hundreds of potential new cases of Ebola, aiding efforts to bring to an end an epidemic that has already killed 3,000 people in the country.

Officials ordered the country’s 6 million residents to stay indoors or face arrest during the period that ended late on Sunday as hundreds of health officials went door-to-door looking for hidden patients and educating residents about the virus.

Reports to authorities of sick people increased by 191 percent in Western Area, which includes the capital, during the lockdown compared with the previous weekend, said Obi Sesay of the National Ebola Response Center.

“Tests are being carried out on their blood samples, and the results will be in by Wednesday,” Sesay said, adding that 173 of the patients in Freetown met an initial case definition for Ebola.

In the rest of the country, there was a 50 percent increase in sick people reported in the lockdown’s first two days, Sesay said.

Sierra Leone has reported nearly 12,000 cases since the worst Ebola epidemic in history was detected in neighboring Guinea a year ago. In all, more than 10,000 people have died in the two countries plus Liberia.

New cases have fallen since a peak of more than 500 a week in December, but the government said the lockdown, its second, would help identify the last cases and reduce complacency.

A source who declined to be identified said there were 961 death alerts nationwide during the lockdown’s first two days and 495 reports of illness of which 235 were suspected Ebola."

"So we don't know where it went, why, or how, and we don't know why it stopped"

I don't think it ever stopped,sure it slowed some,leaders proclaimed the outbreak over-and people went right back to eating BBQ'd bats and rats and monkeys.

Look for "the surge" in reported cases in Liberia,Guinea, and Sierra Leone.



SordidPanda said...

Aesop,

You realize you just wrote, "the official numbers are BS because the people compiling the official numbers estimate the uncertainty to be thus."

The snark was a blatant attempt to point out the obvious, that we can minize some uncertainty but we cannot eliminate uncertainty. It's a statistical analysis problem for any population. In essense, "how big are your error bars?"

What the error bars do is give you broader range, and as long as reality falls inside the range of the error bars, then your statistical sampling method is legit. When reality creeps outside your error bars your sampling methodology is wrong.

What I was trying to lead you to was the conclusion that the official numbers don't have to be "exact" to be "useful."

After all, it isn't like there are a whole lot of alternatives for analysis. As long as we are consistently measuring between half and one third of all cases (not exceeding the maximum detectable rate for the test) then we are good to use that test and methodology.

The uncertainty isn't a problem for me, except that I think the uncertainty was overplayed to play to peoples fears so they would demand politicians spend more money to send more aid.

Aesop said...

No, I'm not.

You should be clear about a couple of things:

The people compiling the numbers are the government functionaries of Guinea, Liberia, and Sierra Leone.

The WHO, and by extension, the CDC, are evaluating those compilations, and they are the ones making the "1/2" and "1/3" utility estimations, for their own reasons, some of which may be reports and evidentiary, and some of which are entirely wishful thinking and trying to make horseshit substitute for jet fuel, as if mere wishing could make it so.

Their evaluation is nonetheless that they're bullshit numbers, but consistent bullshit that is useful and predictable.

IMHO, that estimation is pure fancy, and merely a convenient lie based upon facts not in evidence.

As firsthand report after firsthand report demonstrates, the numbers are merely the lowest possible number of cases, times 1/3, while the upper limit is best described as "the entire population of country X, Y, or Z" based on all available evidence.

When the range of cases is "somewhere within one SWAG of 10,000-20M", any actual utility is like trying to find the chicken in government chicken soup. You'll need a microscope, and a healthy dose of optimism.

And probably, you'll need to supply your own chicken.

Hence my lack of enthusiasm for Liberia's supposed "one" new case.

Is it really "one"?
Is "one" actually "three"?
Or is it "three hundred"?
Or "three thousand"?
And was it ever actually "one"?

This is why there's no utility to the provided figures:
Their people can't count.
They have no reason to count accurately.
They have a strong impetus to lie about what they count, even were they able and willing to do so.
And none of them, from bottom to top tier, seems to grasp the fundamental necessity of accurately counting and reporting the true numbers to properly direct resources to fight the outbreak.

Which is why the outbreak hasn't been contained, and therefore likely never will be.
QED

Knowing the true situation on the ground is thus, should lead to an entirely different set of priorities and reactions from WHO/CDC and the NGOs involved, not least of which to treating everything coming out of the respective governments like we used to treat press releases from TASS and Pravda, 24/7/365.

The "official" numbers are nothing other than the pronouncements of Baghdad Bob, and precisely as useful.

Anonymous said...

Has this spiraled along the exponential mathematical pathway it could have done by now?
Thankfully not. Huzzah.

Why didn't it?
No frickin' idea, anywhere, from anyone.


Ill take a stab at it: It is and always has been a scam. Convenient how it happened RIGHT before the African delegation went to the White House to parcel out their resources to the west. The leaders profit handsomely and the people get the shaft. SSDD.

Next USA gets the excuse to send BILLIONS of aid in the form of CONTRACTORS right to be stationed right by the resources that we 'gained' from the African leaders in order to protect our interests. Of these BILLIONS the US sent, where are all the pictures of beds? Wouldnt this be a great follow up feel good story? Alas, nary a peep.

This is a scam. Just ask Nina Pham and her new found love handles.

Aesop said...

Brilliant analysis, hampered only by the utter dearth of anything like evidence in support, and those unfortunate stacks of bodies now buried in graveyards outside every village and town there. Maybe they've faked all that, just like the moon landings?

But hey, thanks for sharing.

The only difference between your hypothesis, and the West African belief that it's caused by witchcraft, is that they've actually got a reason for what they think.

Perhaps you could write us after your visit to Freetown or Monrovia, and tell us from on-site how much Ebola there isn't in those hospitals?

Keep in touch.

SordidPanda said...

Aesop, if I ever get shot I want a guy like you who is stone cold confident in his own abilities to help with the patching. However, I certainly don't want you teaching statistical analysis to college kids.

This might help: http://www.fao.org/wairdocs/ilri/x5436e/x5436e07.htm

If you can follow along the math to understand the confidence interval for the diseased chicken example, you can come to the conclusion why someone in the CDC declared such a high uncertainty. It isn't an admission that they have no clue, it is an explanation of confidence in the numbers they have. After all, there will always be SOME uncounted disease/death that doesn't make it into the official numbers.

Aesop said...

Math is math, and there's no arguing with actual data.

But we aren't dealing with just math or data here.
This is math based on the reliability of the observers, which is virtually nil.

When you have a witness in court who lies, a reasonable standard for anyone is to assume that any or even all of what they're testifying about is also a lie.

And even granting the WHO/CDC's "confidence" in the testimony of the observers there, they're admitting that 66% of what they're not reporting is a falsehood.

It's been awhile since I was in school, so remind me what the usual letter grade is for someone who scores a 34% on an exam.

Or, playing more in your ballpark, we could use the "Source and Reliability Matrix" found in Appendix B of FM 2-22.3 - Human Intelligence Collector Operations
http://fas.org/irp/doddir/army/fm2-22-3.pdf

A casual glance would seem to put the information on the outbreak squarely in the D-5 category.
(If you'd like to reassess and assign them a different grade, I'm open to counter-proposals, assuming you'll show your work to a novice like myself.)

You will, hands down, have vastly more experience with staff planning and intel analysis than I ever will, so you tell me:
On a scale of A-1 (24K gold) to E-5 (whale shit), after excluding the agnosticism of the "Unknown" categories on the matrix, what adjective word(s) between "24K gold" and "whale shit" would you assign to level D-5 (the 20th worst out of 25 possibilities)?

That's the intel that's been driving the entire Ebola response right along, and it's been precisely the conflicting reports from more reliable groups, like the NGOs there, that have underlined the reliability (or rather the utter lack thereof) of the sources.

We can usefully determine from their reports that there is, in fact, Ebola in each of the three countries. We can even grant that it is for a certainty no lower than the lowest numbers reported, a fair amount of the time, and once those numbers are greater than zero. And we know not everyone in any country has Ebola.

But there doesn't seem to be a lot of argument arguing for much more precision than that, no matter what anyone anywhere would wish otherwise.

It's like relying on global warming believers because they're the only ones publishing.

Or looking for a lost wallet under the streetlamp because "the light is better there".

It may be comforting, it may feel good, or even seem like science, but what it isn't is anything medically useful.

Modern Western medicine is all about statistics, going back to Napoleonic times. What treatment heals what percentage of persons, etc.

Thus false witnesses in medicine are functionally worth about the same as they were to Old Testament prophets: objects for target practice and brimstone.

The reports from West Africa about Ebola are, at best, like a "check engine light": they tell us there may be a problem. What, how big, or how expensive, we don't know. (It may even be a problem with the "check engine" light itself.)

But the involved West African governments fail the metaphor, because they're like a "check engine" light that flickers, goes out randomly, and occasionally switches to an "Everything's Spiffy!" light. And then back again.

A compass that's off by 17 degrees can be used, if you know that correction. A compass that spins every which way is useless, and should be disregarded.
That's exactly what trained pilots do with no-good instruments in flight, and it's exactly what should be done (and should have been done right along)in this instance with anything coming from those governments.

AM said...

Aesop,

The question isn't whether the observers can report with 100% accuracy. The question is whether the observers can report with conssitency. If they only ever get 25% of cases, but consistently get 25% of cases, then you can use their numbers to track the course of the epidemic.

If they get better over time, then your ability to track goes up, and your error bars go down.

If they get worse over time, then your ability to track goes down, and error bars go up.

So yes, math is math, I'm glad you agree. When asked about the math you respond with "well people are liars."

Yes people lie. But do people lie so consistently across multiple governments and international organizations that this can be one big con? I think that is highly unlikely.

Aesop said...

Then you have your answer: even the WHO and CDC had to revise their guesstimates of the inaccuracy upwards over time, because it was clear that the gap between reports and reality was growing wider by the day, week, and month. So the lies have grown larger from the get-go, and given the span of time involved, can now be reasonably assumed to have no bearing whatsoever on reality on that basis alone, nor any necessary consistency from one country to the next.

I don't think this is just a con, I think the governments there lie reflexively, about everything, since forever, both to deflect the public's wrath at their incompetence, and to simultaneously try to sucker as much aid (which they can then embezzle) from the West, while still trying to keep their other embezzlement gravy train of trade going. Given their reasoning skills, the fact that the two are mutually incompatible is merely an new opportunity to lie more creatively.

People who lie pathologically aren't that hard to figure out, and it's not like we're unaccustomed to either sociopaths or crooked politicians (but I repeat myself) here either.

But over there, they lack any semblance of checks and balances, a free press, or any other restraint on their shenanigans. In fact, avoiding such is generally precisely the path by which one gets to power there, going back to 5 seconds after the colonial administrations departed.

You did tours in SWAsia, so you can't pretend the breathtaking duplicity and larceny of third world administrators is some hitherto undiscovered phenomenon.

And it's worse in West Africa, because they're literally at the absolute bottom of the list of worst sh*tholes, and the cheapest thing there is human life. If it were otherwise, they wouldn't be at the bottom of every list of prosperity capable of quantification.

And these are the observers upon whose reliability anyone sane would attempt to fight one of the most virulent and lethal diseases in human history?

Best wishes with that. It's nothing less than a recipe for endless disappointment, and certain catastrophe.

Ask Lufthansa what happens when you trust in the inherent reliability of pathologically warped people.

TrT said...

"The question isn't whether the observers can report with 100% accuracy. The question is whether the observers can report with conssitency. If they only ever get 25% of cases, but consistently get 25% of cases, then you can use their numbers to track the course of the epidemic."

The problem with that line of thought is you need to know if you are counting 25% of cases, 2.5% of cases, or 250% of cases.

How do we know we are consistently counting 1 in 4?

Aesop said...

As noted, we already have seen our officials admit that their officials aren't consistent about their inaccuracy, because they saw fit to revise the fudge factor from "1/2" to "1/3".
And in fact to note the under-reporting was "at least" that much, with no upper limit of unreliability given (other than the obvious unstated one of "all available population").

No attempt was ever made to classify why the inaccuracy, because the variables are innumeracy x illiteracy x superstition x personal motivations x purposeful vs. inadvertent inaccuracy x official motivations x pathological inability to tell the truth = accuracy, and the values for each of those in that region are completely open and unknown.
And because pointing out the bucket of fail inherent in relying on the locals embarrasses those locals, the UN, and their people specifically trying to make cake out of water and bullshit, and we can't be having any of that.

Telling me someone knows how many Ebola cases there were anywhere in Africa without that acknowledgement is as useless as the Drake Equation for determining the probability of life elsewhere in the universe.

As skewered in the brilliant essay by Michael Crichton in 2003
"Aliens Cause Global Warming"
http://stephenschneider.stanford.edu/Publications/PDF_Papers/Crichton2003.pdf
such contentions are thus nothing but junk science masquerading as useful data, and deserve to be mocked and ridiculed as such ruthlessly by anyone seriously looking for answers.

As the "surprise" pop-up return of Ebola to Liberia proves, we can't even depend on their data when the question is reduced to a simple binary one of
"Is there or isn't there Ebola in your country?", where they have the maximum 50-50 break for simply blind guessing.

If any West African government source tells you a body fell in the river, clearly the likeliest place to look for it is upstream, based on available evidence. If their lips are moving, they're lying.

And when getting ahead of the outbreak depends - by the best epidemiological minds available - on tracking, treating, and burying X% of cases, when you never know what the number of cases is, all efforts are doomed to abysmal failure.

And so Ebola remains rampant in West Africa, for months now, despite the media moving on after our government's officially encouraged information blackout.

QED

Anonymous said...

"Where is this headed? How can we stop the spread?"
Here I go again...
Napalm. Good to see you back, and glad you're back winning bread, Aesop.
- Grandpa

Anonymous said...

I am not seeing anything anywhere that could possibly come remotely close to being interpretable as "500 new cases a week".

That statement is bullshit until and unless you provide evidence to the contrary.

Given that that statement is bullshit, much of the rest of what you say is equally so, and what isn't bullshit is irrelevant. We don't understand what happened. Fine. We don't understand the flu or the common cold too good either else they would have been put back in Pandora's box a long time ago. We don't understand it exactly and it doesn't matter because the outbreak has indeed effectively burned itself out.

The evidence and the numbers and the actual first-person reports from the ground do not lend themselves to any other conclusion at this time.

None of which is to say that all the things you say SHOULD be done, should not. Yes, they should. It would be better if they were. Screaming about how the world will end if they're not done, when it manifestly didn't and isn't ending, does you no favors.

Aesop said...

Evidently you majored in ranting, but math not so much.

"A" for enthusiasm, "D" for acumen.

Maybe drag your ignorant ass over to nothing more difficult to comprehend than the Wikipedia page, which summarizes the WHO reports regularly:

http://en.wikipedia.org/wiki/Ebola_virus_epidemic_in_West_Africa

They even have pictures for the math-impaired.

Then note that the "official" number of cases is running at 300-400 new cases a week since January, and that those numbers have been widely ascribed as being probably 1/3rd the actual number of new cases by everyone but the Liberian Ministry Of Tourism.

In the most recently posted data for the week of 29 Mar 2015, for instance, we observe a total of 306 new cases "officially", which extrapolates according to the accepted West African Fudge Factor to 612-918 actual no-shit cases. As Casey Stengel used to say, "You could look it up."

That would be a tad bit MORE than 500 new cases a week, and the same is true going back virtually forever to the height of the outbreak last fall.

Which basically shoots all your bleating into your diaper.

Thanks for playing.
Try paying attention.