29 August 2014

8 Is Enough

AnandTech is out with a rundown on the latest from Intel.

If you're interested in building a database or quant machine with big cujones, the last table on the page gives you the numbers. The quote is $7,080, which is much higher than such a machine would need. You can skip the GPUs (well, at least two of them), drop a couple of hundred watts from the psu, and easily get under $5,000. Running R and any of the open source (or free, as in beer, commercial) databases in 64 gig and 2 terabytes (skip RAID? why not?) of space. That's a couple of weeks of salary for a quant or database dev. Really. You can afford it; the neverending free food'll cost you more.

28 August 2014

Reservations for ... 8 billion??

Quite recently, in various versions of these endeavors, there was an essay dealing with the Old Gold vs. New Gold situation. Not for the first time.
The hidden truth: the country with the reserve currency of the global economy will always, in fact must, run trade deficits. Think about the situation from the point of view of other economies. They have to get reserve currency in order to acquire goods and services.

Imagine my surprise, and wistfulness at not having gotten paid, to see today's NYT piece on just that subject. And about the same conclusion. Well, sort of.

Bernstein paraphrases a Treasury economist (Kenneth Austin) who has published an essay in The Journal of Post Keynesian Economics, here (it's not open, alas). The notion of post Keynes usually means full bore Randian, but not in this case:
Post-Keynesian economists are united in maintaining that Keynes's theory is seriously misrepresented by the two other principal Keynesian schools: neo-Keynesian economics which was orthodox in the 1950s and 60s -- and by New Keynesian economics, which together with various strands of neoclassical economics has been dominant in mainstream macroeconomics since the 1980s.

In any case, neither Bernstein nor Austin (filtered by Bernstein) defend the notion of the reserve country having to be a net debtor, only that the level of its currency circulating can, more or less, be under its control. As events stand now, they both argue the USofA is at the world's economies' mercy.

The Unfortunate Alternative is hard specie, and one need only read up world, and USofA, economic history from the 19th century through the Great Depression to see how foolish that is. If you think the world is not level now, you ain't seen nuthin yet.

24 August 2014

The Poor Will Always Be Us

The environs of Washington, DC are widely excoriated by The Right as being the bastion of Liberal Evil. Not least, the disparity of median income there being higher than most, if not all depending on the year measured, of the rest of the country. Having lived there for the better part of a decade, I can say with certainty that income level was and is driven by the private sector consultants and such, not the bureaucrats who have difficulty keeping up. Bureaucrats aren't living in the big houses in McLean. The academic scene is largely liberal, with the notable exception of George Mason University, in Virginia. That state is really two, with Northern Virginia (actually spelled as such) turning a once deep red state a pleasing shade of lavender. It ain't blue yet, however.

GMU provides the NYT with one of its token right wingnuts, in the person of Tyler Cowen, who regularly displays a breathtaking degree of cognitive dissonance. That the Times would continue to print his stuff is puzzling. I can only guess that the Editors are allowing the loonies to shoot themselves in the foot on full view. I mention this mostly because his essay today exceeds his usual level of incompetence and villainy.

Let's wield the sharp cutlery, shall we?
For all the talk of the Great Depression, we might look at a different exemplar for modern times, 18th- and 19th-century economic history India. That country's economic retrogression during that era may help us understand the quandary that some parts of the world face today.

The overarching theme of the piece is that 18th and 19th century India is prescriptive for today's US and Western economies generally. Baloney. The 18th and 19th century global economy was dominated by mercantilism, with India and the New World colonies being principle examples of those on the losing end of the bargain. This period was marked, more than any other way, by the discovery and pillaging of natural resources in this New World, mostly by European overseers. The second most important point was the development of science and engineering from primitive to near completion (save for Einstein and Bohr and the final entries in the periodic table). Most of the widgets that you use and prize today were invented by The Great Depression, they're just smaller and faster now. In other words: the scope of new knowledge and new resources from 1800 to 1900 overwhelms what can be expected from 2000 forward. The simple fact is, we know just about all there is to know about the physical world and where the useful parts reside. And there is no New World to pillage. (For those talking about mining other planets and such: not with chemical rockets, boyo.) And we're nearing, if not met, the carrying capacity of these, now firmly limited, resources. Wealthy Chinese are scuttling away from their air and water as fast as possible. The notion that there's some magical, as yet undiscovered, venue of employment to support a vast new middle class is a pipe dream (and that phrase refers to the Chinese opium pipe, just so you know). The journey goes from farm to factory to cubicle. Full stop. We've found the whole pie, now we need to carve it up in such a way that civilization survives.
In 1750, India accounted for one-quarter of the world's manufacturing output, but by 1900 that was down to 2 percent.
...
India just didn't do enough to move toward production on a larger scale or with better machines.

Well, let's consider this period. Slave based production of cotton and textiles by the US South dominated the period. Why ship cotton half way around the world, when your slaves can grow it here? Of course, not. Virginia (tobacco) and the Carolinas (rice and cotton) are much closer to London than anywhere in India, after all. At the time of the American Revolution, slavery production was the major source of hard currency, not manufacturing. India didn't stand a chance. Moreover, industrializing agriculture yields more poor people when there aren't alternatives. Take the modern example: robotics replacing hands in the manufacture of autos. You know the rest. The notion that India somehow missed its opportunity to be Europe's supplier of industrial output in 1850, or thereabouts, if only India had spent more on machines is asinine. The reason that Apple, and the rest, can exploit China is the 747 freighter. That aircraft didn't exist in 1850. Nor did the container ship.

Here's where the cognitive dissonance really kicks in:
International trade grew rapidly after World War II, but at least in the early postwar years most of that trade was among countries with roughly comparable technologies and real wages. And that trade spurred growth rather than damaging laggard economies.

In the last 20 years, the economic surge of Asia, especially China, has brought a large trade readjustment to the world, one with few parallels with the possible exception of the rise of the Western economies several centuries ago.

The post WWII economic surge was built on the afterglow of socialism, but not by that name, of course. The war effort was a case of "all for one, and one for all". The notion of shared responsibility, rather than Randian greed (she didn't start in earnest until the 1950s), was the order of the day. Corporations paid real taxes, unions bargained widely, and Bretton Woods made the US buck supreme. Cowen, either because he's too stupid or vile, elides the simple fact: American corporations now exploit totalitarian labor for the benefit of the few. It was brought to you by Richard Nixon in 1972.

But this is all good for China, right?
China's per capita income, less than $300 in 1984, is now in the range of $10,000.

Well, Cowen, being an econ professor, knows that even in the best of economies mean/average/per capita income overstates reality. I can't find a median income measure for China, but that's not too surprising:
Results of a wide-ranging survey of Chinese family wealth and living habits released this week by Peking University show a wide gap in income between the nation's top earners and those at the bottom, and a vast difference between earners in top-tier coastal cities and those in interior provinces.
...
In March 2012, Bo Xilai, a top party official who was trying to create a populist image for himself and was later purged, said at a news conference that the Gini coefficient had reached an alarming 0.46. His willingness to announce the number came as a surprise to many observers.

(The US gini, 2011, was .477. That's worse than China, just to be clear how the number works.)

China's .1% are doing just fine, thanks. The rest, not so much.
Average annual income for a family in 2012 was 13,000 renminbi, or about $2,100.

I wonder, how did per capita income quintuple in a year? It didn't of course. Cowen just made it up. Or the ghost of Ayn whispered it in his ear while he nodded off during a lecture at AEI.

Back to Cowen:
French citizens expect a great deal from their government, and strikes are a common response to reduced wages or benefits.
...
Chinese export growth and wage competition may have been a kind of final straw that made old ways unsustainable.

So, it would appear, Cowen's remedy is for Western labor to accept Eastern poverty as the new normal? How, exactly, is that progress or a solution?

Reading history, and data, makes the story quite clear. Economies which (ahem!) enforce equity, such as the Scandinavians, thrive overall, while those which embrace Rand fall into revolt. Cowen bemoans the prospect of stagnant/falling median income in the USofA, but at no point offers a remedy other than benign acceptance; it's God's will. He has to know that while median income has gotten poorer, the 1% have gotten richer. And he has to know that these effects derive in concert, not by coincidence. And he has to know that slack demand is the result. And that slack demand leads to lowering median income. Rinse. Repeat.

When the dominance of a service economy was being first recognized, in the 1970s, the notion was that such a move was at least as beneficial, overall, as the migration from farm to factory pre-WWII. Instead of factory workers, we'd all be office think workers, earning much more than our beaten down factory fathers. Hasn't worked out that way. We, still, mostly buy real widgets. You can't eat software.

We can't all be London Whales, crashing our small corner of the economy. And if we were, the whole economy crashes. Wait, didn't we do that?

So, you can be poor because totalitarian regimes elsewhere keep wages at bare subsistence and your job goes there or... you can be poor here as corporations implement wage arbitrage down to bare subsistence with the outsourcing as threat. Of course, in due course, there'll be no one in the USofA, or anywhere else, who can buy the all the widgets being made. What a country!! The United States of Mississippi. But, of course, you can't sell much in Mississippi; they're all really, really poor.

The fundamental problem is that wage or tax or foo arbitrage by corporations ultimately fails for structural reasons. As your Good Mother told you, "what would the world be like, if everybody behaved like you?" The second and third worlds, largely autocratic governments, provide cheap hands in order to earn hard currency (which they, mostly, keep for themselves, of course). Today, that's the US buck. The same thing goes on here, the red states ban unions and drive down wages, so corporations move to such states. The problem, of course, is that both kinds of arbitrage depend, absolutely, on a high wage population (those God hating blue staters, of course) to suck up output. What happens when there's no longer a pool of high wage earners to export to? The issue grows more dire as automation becomes both more widespread and expensive. You do remember the story about ditching 300mm wafer production for 450mm?? Hasn't happened. Likely, never will; insufficient demand for so many more chips to pay for the machinery. In due course, and since the world is non-linear (according to Dr. McElhone), demand utterly collapses when least expected.

23 August 2014

Am I Blue? Yes, I Am

While I dearly love DB2 on LUW, I've been putting off upgrading the editor needed for 10.x. Prior, the Control Center was very easy to use and not too much of an install. This newfangled "Data Studio" is 1.5 gig of download and a java/eclipse beast. A DSL killer. Well turns out, there's a way to get it without using the on-line Install Manager. This post lays it out. One point of confusion is that the repository adding happens off the initial File menu; I spent a couple of times fighting with on-line nonsense. Just fire up the download when Olbermann comes on, go to sleepy sleep, and wake up with a huge gzipped tar file. Talk about bloat?

Still can't, that I've found, use it with PG, of course. Offers up MySql, along with the usual commercial suspects. DB2 has so many dials and knobs in the engine that one can tune (and that none of the others, especially OS, have), that multi-database editors such as AQT and Aqua (haven't used Toad in a while, so I can't say) don't bother to add support for such a small market share database. Too bad about that part, since DB2 really rocks.

20 August 2014

Danger Will Robinson, Again

simple-talk has an interview with Bjarne Stroustrop just up. It's a better interview than most, by the way. Well worth the effort. Reading it reminded me of a quote from Stroustrop, but for which I cannot vouch, since I've long lost the cite. I've been looking to confirm it for years.
Object oriented programming is buzzword programming.

It may be C++. And it may never have been said by him. Or anyone. The Voices may have just assimilated me.

Anyway, the interview has me off looking, again. I did find this piece, which provides the quote which got me into so much trouble with the Indians at CSC:
As we have all learned, methods in good OO programs should be short and sweet. Lots of little methods are good for development, understanding, reuse, and so on. Well, what's the problem with that?

Well, consider that we actually spend more time reading OO code than writing it. This is what is known as productivity. Instead of spending many hours writing a lot of code to add some new functionality, we only have to write a few lines of code to get the new functionality in there, but we spend many hours trying to figure out which few lines of code to write!

One of the reasons it takes us so long is that we spend much of our time bouncing back and forth between ... lots of little methods.

This is sometimes known as the Lost in Space syndrome. It has been reported since the early days of OOP. To quote Adele Goldberg, "In Smalltalk, everything happens somewhere else."

Organic length of code, as organic normal form, is the ideal to be sought. The fascination with two or three line methods, tens or dozens of 'em, means only one thing: in order to understand the functioning of the object, one has to hold lots of stuff in one's immediate memory. Few can do that, so most OO code ends up being written in the (GUI) debugger. Unlike the Deming Dance (do it right the first time), this is pound the hammer until you bust the thing to pieces.

Not to mention the corrosive effect of the web: OOD/OOP has devolved back to function/data of FORTRAN, only we now adopt the terms ActionObject and DataObject. As Stein said about Oakland, "there's no there, there". OOD/OOP means that objects really are self-contained, and communicate with other objects with messages (which is to say, "do something, and I don't care how"), not data packets. Where did I park that DeLorean??

18 August 2014

Life is A Totem Pole

Having lived, in my first job in Boston and thence in Washington, DC (and having done my grad school in econometrics), the notion of localized inflation was well understood at least as far back as the 1970s. No matter where you work/live, the $$$ you get paid aren't what determine how well you live, but where those $$$ seat you on the local totem pole. I guess it's still news to some folks.

Interception

These endeavors have spent a lot of virtual ink over the last year or so making that case that financial quants don't know as much as they think, and that the events which led to The Great Recession and London Whale should be, but clearly haven't been, teaching moments. I've heard tell of various bots and such, but this expose` is truly mindnumbing.

As has been promulgated here, it is events which move data, not the other way 'round. While time series analysis can give one a warm and fuzzy feeling that the way of the world is clearly delineated in all-the-data-to-today, and one can happily predict off the end of this curve, it's not much more than tossing a dart in a pub.

You must read it.