Please God, Make It Look Like the Good One

Bob Muenchen has been keeping track of Point-Click veneers to R. These are not RStudio alternates, but attempts to behave like SPSS. This is, I guess, support for my notion that the principle strength of R is its bevy of built-in (including contributions) commands, rather than as a programming platform. We'll see.

Figures Don't Lie, But Liars Figure

Prodded, in significant measure, by the inane behavior of The Manchurian President, I grow ever closer to believing in the NYT rather econometric models for the purpose of 'predicting' the future.

I do not think that the currently popular DSGE models pass the smell test. ... The advocates no doubt believe what they say, but they seem to have stopped sniffing or to have lost their sense of smell altogether.
-- Robert Solow/2010

Of course, DSGE has been the Flavor-of-the-Month within the data driven economic prognostication cabal for some time. Solow's experience notwithstanding.

Which situation leads me to consider another mode of analysis: game theory. There's the famous book, co-written by an economist. If one views non-Mother-Nature-based data driven analysis as on the razor's edge of credibility, as is macro-analysis, then one's analytical method must needs be event based. Thus reliance on the NYT, and history, to guide the way.

All of which led me to a recent (as in, a few years rather than decades) post delving into the problem.

The problem, Boumans explains, is that Morgenstern's benchmark was observations in the natural sciences, which were planned, designed, and guided by theory. Errors were the unsystematic result of human mistake, for nature "doesn't lie deliberately." But economic agents do, Morgenstern lamented, which adds to the lack of designed experiments, errors from questionnaires and instruments, and lack of definition.

Sound a bit familiar? So, yes, mine is not a new or unique observation, just independently arrived at. Why is it that the mainstream pundits remain befuddled by the continuing rush to Treasuries, and the resultant dive in long term interest rates? Apparently, such behavior is not amenable to analysis from existing time series. Who woulda thunk it?

Or, in varying syntax, it is said: 'econometrics has predicted 9 of the last 5 recessions'.

Free Radical

Finally, a mainstream pundit edges a tad closer to answering the puzzlement: why do Treasuries keep going up in price and down in realized interest? Why, oh why? The answer was, is, and always will be: more moolah chasing a (more or less) fixed supply of Treasuries.

The reality is that China and about 20 other nations are already doing so. By using public capital to purchase huge quantities of United States government securities over the past two decades, they have driven up the value of the dollar to make their own exports supercompetitive.

Capitalists, the world over, are scaredy cats when it comes to turning financial capital into physical capital. We've had 'historically low' real interest rate for about a decade, and they still avoid real investment. 'Gimme Treasuries or give me Debt!'

And, yes, productivity has been going up like a Saturn rocket since the Great Recession. It's just that workers haven't gotten the benefit of that increase. As the op-ed piece makes clear:

Wall Street didn't object to the dollar gaining value. It enabled a flood of cheap imports, and has driven huge profits for the companies that sell them, especially Walmart, Amazon, Nike and Apple. It also led to a reduction in wages for 80 million American workers competing with countries whose labor has been cheapened by these undervalued currencies.

Not to forget: aided by authoritarian governments which grind workers under the heal of bidness. Capital never met a Fascist it didn't love. Not to forget The Manchurian President who adores the beautiful letters he gets from them.

Data Science - part the second

So, today the other shoe dropped. USDA Perdue is shipping the data folks to the arid desert of Kansas City. "Arid", you ask? Here's what Perdue had to say to those who complain that this is just a slow motion purge:

He added that this decision was not made "with disruption in mind," and that federal employees have many other opportunities within the government should they choose to stay.
[my bold]

Well, let's see. What we really want are the number of Federal locations by SMA by GS occupation series. What I can find, so far, is this list by state of Federal employment. It's a proxy, at best. What we really want to know is: what happens to these data folks when Perdue starts culling out the undesirables?

So, in Missouri there are: 35,549. USDA is 3,573 For DC the numbers are 141,514 total and USDA 6,171. Somewhere, deep inside OPM (where I once toiled) there are the numbers of GS occupations that fit the USDA folks being shunted into the desert. The chance that there are the numbers of open data science slots in the Kansas City MSA as DC is ludicrous. For the USDA folks, they're now in a company town. And you all know what happens then? Of course you do. Those that don't toe the line with regard to climate change or any other externality generating private sector activity will be RIFed. With, of course, no where to go. Hell of a way to alter reality.

The Damn Gummint - part the first

For what is likely to be a series of expletives, herein the first explanation of why any society needs a Damn Gummint.

The opioid pusher Insys has filed bankruptcy, and the right-wing anarchists will hail this as the free market working as it should. Ah, that it were so. A share once sold for just over $13, as I type it's going for $.39.

What this really demonstrates is the perversion of the corporate construct: the corporation perpetrates any manner of negative externalities on the public, raising its profit and sloughing the costs on the rest of us. Since 'owners' are exempt from paying for the corporation's evil, they get to keep the moolah. When caught, little to no assets remain to undo the damage, so bankruptcy ensues. If the damage is mitigated, the taxpayer foots the bill. See Superfund. The Manchurian President has done so many times in the past, so I suppose that makes it the smart thing to do. He is, by his own account, a stable genius.

Book Em Danno - part the second

Yet another rant against the book publishing machine.

I decided to look at python, once again, and given my preference for dead-trees information and Rep Kover binding for soft-covers I was led to a No Starch. O'Reilly abandoned all pretext of caring about its market when it went to standard perfect binding. No Starch, on the other hand, was still providing the, now, standard (tape not cloth attachment and low quality glue) Rep Kover. Or so I thought. The last couple I got, one for R and another for python, had the characteristics of 'real' Rep Kover: a tape strip (RIP, cambric) to attach the covers to the page block, scores on the covers to allow them to flop open like a ballerina's legs, and (they claim) PUR glue between the tape and the page block. This latest python had no scores on the covers, no tape to attach cover and page block, and glue as thick as any perfect bound. The cover is just glued to the first and last pages. You can guess what happened?

I still have O'Reillys from the late 90s, with cambric and PUR of microscopic thickness, and they've not lost a page, and really lay-flat. The only substantial difference between what is sold today as Rep Kover (by No Starch, anyway) and perfect binding is that the cover spine isn't glued to the page block.

Anyway, in trying to get the covers of that latest python book to actually 'lay-flat', the back cover detached from the page. So, 'knowing' that No Starch books had 'real' Rep Kover, I emailed No Starch asking for a properly bound unit in exchange. They said, no, since I hadn't bought the book from the website. About a week ago I got an email from someone at No Starch asking if I had received a replacement. I described what had happened, and he wrote that the printer was replacing defectives. Did I want one? Of course.

It finally arrived, and lo and behold, exactly the same construction. Emails ensued. Me complaining that the new unit had the same flaws as the broken one. What was going on?

Eventually, the CEO replied that he had decided that dispensing with tape attachment and cover scoring were benefits to the consumer. Yeah, sure. It was just another cheapening of product to save a couple of pennies/unit. No more No Starch for me either.

I've chosen to work with a printer that does *NOT* use tape to attach our covers because many of our readers don't like the look and feel that the book looks like it has been repaired.

So, for decades O'Reilly advertised its 'real' Rep Kover bindings, and sold lots o units because of it. And now No Starch customers think that it makes the book look "repaired"???? Fake news.

It Can't Happen Again?

Yet more agita for the Boeing folks. Turns out that the 737 has more problems. The reporting, alas, is wrong on at least one point:

Leading edge slats are an aerodynamic control surface that extend from the front of the wing. Some the tracks may not meet manufacturing standards and may need to be replaced, Boeing and the FAA said. They said if the parts are found to be defective, airlines should replace them before returning the planes to service.
...
The faulty parts could fail prematurely or crack. The FAA said a part failure would not bring down a plane, it could damage an aircraft while in flight.

While strictly speaking, hopefully, true, at least one 737 has burned to the ground due to a compromised slat track mechanism. "Air Disasters" had an episode, bolstered with amateur footage of the burning of the plane. Boeing and FAA seem to be still in bed together. A puncture at flight level wouldn't (with 99.9999% confidence) cause the fire, due to dissipation of fuel. On the other hand, fuel loss could be fatal. And, of course, the China Air incident occurred after engine shutdown on the tarmac. The broken part caused the tank rupture as the slat retracted while taxiing to the terminal.

Still, a tad too cavalier? Yet again? Nobody needed to know about MCAS, now did they?

For some time, there have been stories that the Boeing/McDonnell merger was more like a coup by McDonnell suits. Here's one take

For me, the biggest change came with the McDonnell Douglas merger. All the processes and procedures that we had as a heritage Boeing Company were slowly changing and becoming the McDonnell Douglas process or the McDonnell Douglas procedure.

I felt it was wrong. I felt it was going to take the company in the wrong direction. I thought that quality would suffer and the integrity of the product would suffer. Anybody I know that's a heritage Boeing employee has said the same kind of things.

Some times, Social Darwinism is the wrong way to go. The main purpose, at least internally, for having a Damn Gummint is to enforce payment, up front, for externalities. In other words: if we allow corporations to do as they please, post-damage recovery will likely exceed the assets of a corporation and just declare bankruptcy. The Suits have already gotten their moolah, so what do they care? Shareholders and the injured public get the shaft. An ounce of prevention and all that. We don't need no EPA!