Doctors, Doctored Numbers, and Democracy

[TBoneTX 45,571]

Article from a normally very calm author who has lost patience.  He objects, procedurally and intellectually, to operating consequent to mathematical models instead of real data about the coronavirus.  Comments?

Quote

Doctors, Doctored Numbers, and Democracy

by Angelo Codevilla - April 2, 2020

 

[...] Trouble is, though, these numbers come from just some doctors—they reflect neither reality nor broad medical opinion.  They are synthetic products that hide the (often ignorant) assumptions that they reflect, substitute for reason, foster panic, and ruin the country.

 

The following shows how "soft" and hence divorced from reality these numbers are, what it would take to produce "hard" numbers—i.e. ones reflective of reality—and the manner consistent with self-government in which such matters should be debated and decided. [...]

 

Continues here: https://amgreatness.com/2020/04/02/doctors-doctored-numbers-and-democracy/

 

 

 

 

[Dashinka 24,628]

People are beginning to wake up to the bureaucratic mess that is being created.  I wonder who will be the first governor to take the plunge and open back up (of those that have issued SIP orders)?

 

Making decisions on the basis of meaningless curves and bureaucratic authority rather than through open debate about hard facts followed by roll call votes is not just undemocratic. It’s stupid.

[Burnt Reynolds 5,058]

tl;dr rant incoming (the topic and article just made me think of stuff, which I'm going to just post):

 

One thing to understand about this recent fetish for models and modeling, whether it be in science, whether it be in sports data, etc., and why this came out of nowhere, is simply that it was becoming a social trend. 

 

People need to adequately understand the way social trends are perceived and used. It can be as innocuous as someone seeing demographic and segmented trends and modifying their products to sell more stuff. Cool. 

 

Or, it could be used the way predators do when they notice women having body language and social cues to demonstrate their vulnerability. 

 

Just like climate, if we take the sports statistics and modeling fetish that's suddenly come, you'll notice, hasn't ever accurately predicted anything. So naturally one should then turn their attention to why its done. It's done because people saw social trends changing, people idolizing, fetishizing, numbers, and the appearance of statistics as something useful when its basically like people performing magic tricks or at a circus and people simply being entertained because they want to be. But people can take that too far, to where they have personal resonance with this, and thus overly commit themselves to trusting something because of the appearance of truth, rather than actual truth being proven.

 

This resonates in the business world, anyone in management of any kind might be aware of how statistics and data/visualization apps like R, PowerBI, analytics, etc. works. However, anyone with actual expertise and real world experience (and not trying to sell things) knows that these things are rarely useful as predicates.

 

Where modeling systems most err is their excessive use of "assumptions". If you're going to guide statistical models based on assumptions, those assumptions had better damn sure be accurate. Except they're not. So while their math winds up looking sound, it fits their regression model, fits within their stated margin of error, passes fitness, and so on, its derived entirely from error, so it has no actual meaning in the real world.

 

In some ways you see this similar fetishizing with AI and algorithmic based AI. Twitter for example cited their censorship structure as algorithmic, as in, it must be some neutral AI deciding things, when in fact, its humans subjectively deciding what words to "implement" in an algorithmic fashion. It becomes like the "lie detector", where it has no real use in science, nor in fact, but merely becomes a social tool used to herd around people of less intelligence. 

 

But this is what we're faced with. There was a time right after high school where I used to believe the world would be better off without religion (I was quite rebellious against religion because I had a life epiphany that all this religion stuff I was taught was wrong), but experience has taught me the same irrationality that encompasses religion remains in many of the non-religious, it's just merely shifted. In many ways we are hard wired to be irrational.

 

In the business world, I acquiesce to the social trend that people will always want an easy outlook, and thus, why anything that *looks* intelligent will be catered to. In the private industry, especially those working for larger companies, you're likely faced with stakeholders that only care about investment ratios and not processes. This is why, no matter how much AI, analytics, and modeling is played up, in the end, what'll always drive good choices is simply the ability to recognize trends, a social intelligence factor to galvanize, and overall, heuristics based problem solving. 

 

If people want to understand why humanity seems to stride forward then take two steps back, look to our hard wired irrationality.

Edited April 3, 2020 by Burnt Reynolds

[Dashinka 24,628]

Give the governor a harrumph!

 

Edited April 3, 2020 by Dashinka

[TBoneTX 45,571]

The harping on models aside, I'd like to see Recovery/Discharged from Hospital statistics cited in the same breathlessness as "21 zillion new cases!" and "12 more deaths!"

 

I'd like to hear "infected / recovered / never hospitalized" statistics cited.

 

It's probably too late now to split all the figures out, but I'd like to know precisely how many deaths were OF (caused by) the coronavirus versus deaths from other potential/actual causes in patients who concomitantly HAD the coronavirus.

[90DayFinancier 910]

9 hours ago, TBoneTX said:

The harping on models aside, I'd like to see Recovery/Discharged from Hospital statistics cited in the same breathlessness as "21 zillion new cases!" and "12 more deaths!"

 

I'd like to hear "infected / recovered / never hospitalized" statistics cited.

 

It's probably too late now to split all the figures out, but I'd like to know precisely how many deaths were OF (caused by) the coronavirus versus deaths from other potential/actual causes in patients who concomitantly HAD the coronavirus.

Aside from the number infected, I readily found all of those stats state and country.

You will only know the number infected after this settles, and then it is only estimated.

https://www.worldometers.info/coronavirus/

 

[90DayFinancier 910]

COVID-19 Fatality Rate by COMORBIDITY:

*Death Rate = (number of deaths / number of cases) = probability of dying if infected by the virus (%). This probability differs depending on pre-existing condition. The percentage shown below does NOT represent in any way the share of deaths by pre-existing condition. Rather, it represents, for a patient with a given pre-existing condition, the risk of dying if infected by COVID-19.

PRE-EXISTING CONDITION	DEATH RATE confirmed cases	DEATH RATE all cases
Cardiovascular disease	13.2%	10.5%
Diabetes	9.2%	7.3%
Chronic respiratory disease	8.0%	6.3%
Hypertension	8.4%	6.0%
Cancer	7.6%	5.6%
no pre-existing conditions		0.9%

*Death Rate = (number of deaths / number of cases) = probability of dying if infected by the virus (%). The percentages do not have to add up to 100%, as they do NOT represent share of deaths by condition.

[Dashinka 24,628]

18 minutes ago, 90DayFinancier said:

Aside from the number infected, I readily found all of those stats state and country.

You will only know the number infected after this settles, and then it is only estimated.

https://www.worldometers.info/coronavirus/

 

Hmm, just like the stats regarding the flu.

[90DayFinancier 910]

1 minute ago, Dashinka said:

Hmm, just like the stats regarding the flu.

Yes, just more lethal

[Dashinka 24,628]

1 minute ago, 90DayFinancier said:

Yes, just more lethal

I think that is still TBD considering no one has any idea of the total number of infected people.