Saturday, July 27, 2013

Not every good idea is a good idea

PayPal, Tesla and SpaceX founder Elon Musk has been generating buzz for a concept he calls “hyperloop” for intercity transport, supposedly faster than subsonic aircraft (approx. one and a half times), and cheaper. Apparently the technology, to be unveiled 12 August, is based on air pressure/vacuum and magnetic acceleration. Musk has twitted that this is not a vacuum tunnel, so it must be a vacuum above-ground enclosed tube—otherwise air resistance would prevent the claimed speed and would require continuous power. Los Angeles to San Francisco is his suggestion of a first application.

Disbelief has to forcibly choked, not merely suspended, to avoid the conclusion that whatever it is, it would be far more expensive than the conventional high-speed line now being planned between these very cities as a result of a matching-funds deadly embrace between the effectively bankrupt state of California and the rescission-bound federal government. No-one claims that that project makes sense by any commercial standard. The likely ticket prices will target mostly airline passengers, of whom there are only 13,000 per day between the multiple airports in each region in both directions, split among more than half a dozen airlines—thus ensuring minimal prices. With traffic so low and priced so close to cost, one is at a loss to justify the existing rail project as sensible even in serving some significant unmet need. Musk is proposing something with far higher capital costs than air service, probably higher operating costs, and only inconsequentially higher speeds.

Yet this as-yet undeclared idea is spam du jour. Musk has transcended his real achievements (first commonly accepted electronic payment service, first and second proper electric car, first private mission to the ISS) and filled the…vacuum formerly occupied by Steve Jobs — that of the unassailable genius pronouncing ex cathedra.

It is not that Musk isn’t that, but rather that his two well-known, business-jeopardising mistakes were both due to false economies—behaviour antithetical to a high-capex project such as a hyperloop. PayPal’s inability to bring itself to pay for any meaningful customer service quickly made it the thief’s friend, before Ebay purchased it and cleaned it up. And years later at Tesla, to save an inconsequential amount, he built charging stations that were A Bridge Too Far—one rather than two on each of Washington–New York and New York–Boston, and a little too far apart between the California coast and Las Vegas. He then gave reporters $100k+ Model S cars with the best batteries and they got stuck. And stuck. And almost stuck twice. Unsurprisingly, that is what they then wrote, for the opposite effect from the desired validation of Tesla’s fitness for the real world.

The moral of this story: don’t buy hopelessly romantic technological dreams that would make Mitterand blanch from an entrepreneur who almost impaled two otherwise successful companies on fits of irrational tightfistedness.

Friday, July 26, 2013

Design and death

In Spain’s worst rail disaster in almost 70 years, 78 people are dead and 32 are in serious condition.

The high-speed ALVIA service uses a preexisting tight curve at that location, three km south of Santiago de Compostella on the Madrid-El Ferrol line. Straightening the route would have been costly, so train drivers are required to slow sharply from 200km/h to 80.

Train drivers, like car drivers, often do not do as they should. On Spain’s dedicated high-speed lines, as in many other parts of Europe, a system named ERTMS would then apply the train’s brakes. But there was no ERTMS on this section of the route, nor any alternative automatic breaking system triggered by excessive speed, such as that used extensively in the U.K. (following a spate of horrific accidents there). Instead, Spain’s old ASFA signaling system was used, which merely informs the driver that he should slow down. But the driver hadn’t been paying adequate attention.

So the driver’s failure to slow down in time on a tight curve such as this automatically and ineluctably led to an accident. It was then up to the design of the rolling stock to minimize the harm from the result.

There is a best-practices design that has successfully done this for 32 years, with only two fatalities on a single train. It is Alstom’s TGV, used in several countries including, ironically, Spain’s own first high-speed line between Madrid and Seville, where they still run. Adjacent carriages are joined by rigid membranes and ride common bogies (trucks). The entire weight of the train acts to keep every part of it upright. It also prevents the carriages from jackknifing (folding on each other), as happened both in Spain and in the similar accident in Germany in 1998. The carriages’ departure from the track at high speed into a concrete wall, their falling down on their sides, and their jackknifing were major contributors to the carnage.

The Bombardier-Talgo S730 trainset that crashed in Spain represents a nod to some of those principles but not enough of them to prevent a bloodbath. It does have a low centre of gravity. It is not made of aluminium extrusions; those tend to tear along the windows like stamps along perforations, and apparently contributed substantially to the toll in the German ICE accident. However, it is made of light aluminum that shattered on impact, the car ends were torn from their tilting suspensions, and the aft gas turbine generator engine was thrown from its dedicated carriage, landing in its entirety on the embankment above. It does use joint bogies, but these do not by themselves guarantee stability or integrity in the absence of perhaps the most important safety factor: rigid membranes between carriages. On the S730, the latter are fully articulated as on a conventional train. They are free to fall on their sides, fly into walls, collide with each other and fold like harmonicas.

This was not, as many terrible accidents are, the result of an unlikely combination of factors. Rather, this was a blatantly design-caused, foreseeable disaster. It would be unfair to say that nobody cared. However, by default we all labour under an overwhelming weight of lack of awareness of the meaning of our choices. Here, people from four companies — the train operator RENFE, track owner ADIF, Bombardier and Talgo — allowed a design to be developed and placed into service that relied entirely on the driver’s slowing down when entering that curve. If he did not pay full and timely attention — as they themselves had failed to do — or the train’s braking system failed, there would be no automation to stop him and nothing to prevent a terrible outcome.

Thursday, July 25, 2013

Leaving Las Vegas

Television — with which the average person (of course not you or I) supposedly spends hours per day — is changing much faster than befits such a dominant, time-tested institution. As media use becomes individualised, the little fish plus the Web plus streaming services are ganging up on the big fish.

The latest trend is for big terrestrial networks to threaten to switch to pay-TV distribution only (cable and DBS). So long as pay TV has near-total penetration, and pay-TV-only channels are challenging the major broadcast networks’ ratings, they say, why should we be subjected to the injustices and iniquities of regulation?

It is an obvious bluff. But not because the bluffers won’t carry out their threat if they don’t get what they want. It’s because they will carry it out even if they do. And not a moment earlier either way.

ITV, long the U.K.’s top network with a 31 full-day viewer share in 1992, has plummeted to 16 by last year (its only comparably sized competitor, BBC1, went from 25 to 20 at the end of last decade but since rose back to 23). It operates on limited-time franchises issued by a public authority. It still produces some expensive “quality” drama (although much less than it used to) and a fair bit of national and local news. Having adjudged these to be bad for its business, it started carrying the line that it was seriously thinking about leaving terrestrial distribution unless the relevant regulations were seriously relaxed. The government did not want to deal with this issue and extended ITV’s licenses until 2024.

Would ITV leave the public spectrum? What would happen to it? Its next nearest competitors BBC2 and Channel 4/S4C have shares around 7; Channel 5 has 5. The rest are at 2 or lower. And ITV is quite profitable, with a ~10% net margin last year.

Content may be king — some of that expensive programming that differentiates ITV may also permit it to soar above all but one other channel, as it still does — but it is a constitutional, primus enter pares king: pay-TV-only channels with arguably better programming, which they use against each other, cannot touch ITV, apparently because of its privileged terrestrial position. If it were to leave, it would lose access to the 46% of U.K. households that have no pay-TV, but also its disproportional appeal to those who do — based on its terrestrial distribution, ostensibly irrelevant to them. Perhaps it is the very public-service-obligation programming that it is trying to shirk that confers upon it this special status, although it may not be evident in each programme’s direct ROI. (Such indirect effects are NBC’s argument for paying for the Olympics even when they run at a loss.)

So ITV won’t leave under current conditions (barring the group insanity of its management). But it may well continue to cut programming costs and dumb down their appeal in response to declining ratings; it is the global poster child for this disease (see the previous post). And unless it stops, it may reach a point at which there won’t be much benefit to staying free-to-air.

The latest big bluffer is Fox in the U.S. It is unhappy that Aereo, a startup that streams terrestrial channels from personal antennas, refuses to pay it retransmission fees. These now approach $2.4bln but media rep SNL Kagan projects a rise to $6bln by 2018. U.S. terrestrial TV also takes in around $47bln in advertising. Fox was second this season, usually several times higher than the top pay-TV-only channel. There is no precedent for the switch of a high-cost, high-rating terrestrial network to pay-only distribution. There was, however, a high-profile failure to do what that would almost certainly require: sharply control costs. It was NBC’s replacement of drama (ER) with Jay Leno, and it failed disastrously, with affiliates criticising the network in public and replacements having to be hastily procured midseason. Network TV may be down at heels but not that down.

Aereo, precisely to avoid having to pay for retransmission, offers terrestrial channels only. As such, it supports the small but highly profitable — for terrestrial broadcasters — cohort that does not use pay TV or prefers not to use it. Aereo offers essentially an extension of conventional terrestrial distribution, for which, of course, there are no retransmission fees. The good news for the networks and their affiliates are the huge profits they make, burdened by little or no competition, off those 15% of U.S. households (and a larger percentage of viewers, now that viewing has largely become individual). The day may come when the networks conclude they have nothing to lose by abandoning terrestrial distribution; that day is still very far, and services like Aereo would not be a factor because, far from harming the networks, they restore their glory days of oligopoly — among subscribers. Until then, Fox would be nuts to move to pay TV, but it may be right in considering members of Congress and federal appellate judges daft enough to think that it would.

Monday, July 22, 2013

No news is better than good news

Large terrestrial television networks have for many years been losing audience across the industrialised world. Those with the highest programming costs — in the U.S. and the U.K. — are on a vicious cycle of declining revenues that lead to less attractive programming, rinse, repeat. ITV in the U.K. is barely recognisable from twenty years ago, La Cinq in France is long dead, The WB and UPN in the U.S. were shut down and merged into the barely profitable CW, and the country’s most-watched network on Fridays, and in the crucial 18–34 demo on other days, is now routinely the Spanish-language Univisión. Heading a major terrestrial network has become a very stressful job.

The same is happening in Russia. The top three terrestrial networks, Channel One, Russia 1 and NTV, had a combined 55.4 full-day audience share in 2005 (full-year and regardless of means of distribution) in the country’s TNS TV Index. By 2008, this dropped to 51.3. Then the rate of decline doubled, and in 2012, the three networks had a 41.0 share. At this rate, they would drop to zero in 16 years (of course they won’t but they can easily engage a vicious circle that would render them irrelevant to most).

The main reason for the big three’s decline over time seems to be their heavy bias towards older viewers. They had a 53.4 share in Jan.–Oct. 2012 among people 55+ (life expectancy there is 64.3 for men and 76.1 for women, so this cohort is proportionally smaller than in most other industrialised countries), 39.4 among people of ages 35–54, 29.9 among those 18–34, and 17.0 among children 4–17.

This is interesting not just from the perspective of broadcast management. Almost from his election in 2000, Vladimir Putin has based his autocracy on near-total control of these three networks, which he quickly established, and on the dissuasion of other terrestrial networks from any informational role (the runner-up TNT and STS networks are tolerated because they have no news programming; the once-critical REN was acquired by a progovernment owner and largely neutered). On the big three, the opposition that has emerged since the fraudulent parliamentary elections in December 2011 is occasionally pilloried as foreign agents but usually not mentioned at all. Of the several protest events since then that brought out more than 100,000 Muscovites, only one was mentioned — briefly — on two of those channels, and only because their news staff threatened to strike. The third canceled its main newscast that day. The leading opposition figure Alexei Navalny is blacklisted from any mention; his arrest last week after being sentenced to five years in prison on a trumped-up charge was covered — briefly — on just one of the big three. The Russia visible on these screens has no meaningful political choice at all.

Putin has relied heavily on the support of older, poorer and provincial Russians, most of whom vote either for him and the majority United Russia Party or for the declawed and controlled Communist Party. These are thought to watch mostly the  terrestrial channels: many cannot afford pay TV and have no interest in other channels anyway; few of them use the Internet. He is known to review only the newscasts of the big three networks, usually to the exclusion of any other media.

If the régime doesn’t topple earlier, the big three could lose half of their remaining share by the time Putin seeks reelection in 2018. If they do, television news will cease being a regular item on the menu of most Russians — an unprecedented development in the industrialised world. What will be the effect of the loss of the country’s main source of information?

The TNS report, in Russian, is at .

Sunday, July 21, 2013

Living by one’s wits

Conventional wisdom often can’t keep up with reality. The U.S. may have lost much of its manufacturing base but still exports a lot of high-tech manufactured goods (aerospace, computers, pharmaceuticals, scientific instruments, electrical machinery, etc.). Right?

Not as much as it used to and, relative to its population, less than practically all other major industrialised countries. Here are such exports per capita (click to enlarge):
Source: World Bank analysis of the United Nations Comtrade database, . Population data compiled in Wikipedia, which see for sources, .

Singapore earns half its GDP from high-tech exports; the U.S.—less than 1%. This would not be so alarming had the U.S. not had a huge balance-of-trade deficit. Singapore, of course, has a large surplus.

Now here’s how the major exporters of high-tech manufactured goods (above $50bln/year) are trending:
  1. China is on a tear, exporting half the total of the other majors combined and growing rapidly. High-tech goods now make up 26% of all its manufactured exports; the U.S. fraction is only 18%.
  2. Almost all the other majors are growing except the U.S., which is now third behind Germany, a country almost four times less populous.
  3. The U.S. topped out in the middle of the last decade and dropped sharply in the first year of the economic crisis, from where it hasn’t recovered.
Furthermore, the majors are not very major, having accounted for only 23% of high-tech manufactured exports (not consumption) by value in 2010 (the latest year for which complete data are available). Manufacturing dependent on high tech is already widely spread around the industrialised countries — and many others.

Like most statistics, these numbers lie, and in a very specific way. When one country (e.g. China) exports manufactured goods (e.g. computers) made possible by R&D conducted in other countries (e.g., the U.S. and Europe), it is the exporting country that books the value of the finished goods (computers). The foreign R&D contributors book little: a much smaller value for microchips exported to China for inclusion in the computer, and no value at all for the motherboard and graphics card made there under license or, probably, any of the software. The same principle applies to exports of pharmaceuticals (from, e.g., India) developed elsewhere (e.g., the U.S. or Europe). This is how China can rack up these enormous numbers while not really being competitive (yet) with the U.S. in producing high-end high-tech manufactured goods (aerospace, medical devices etc.).

However, recognising this lie results in another lie. Money, in the balance of trade, is money. Unlike China, the U.S. has a huge deficit in its trade with the rest of the world (specifically, with China). It doesn’t matter that the U.S. developed a product of interest to other countries if it cannot use most of its value to cover that gap. Of course, it could develop more products and try to live by its R&D capacity alone, but that is clearly not happening.

Thursday, July 18, 2013

Microsoft rumours of own demise not greatly exaggerated

When it defended against an antitrust suit by the FTC and 14 states in the late 1990s, Microsoft, then controlling more than 90% of the operating-system installed base, spoke of itself as if it were an ephemeral butterfly. The courts did not have to intervene because this was tech, and some better-faster-nimbler competitor could arrive at any moment and steal its lunch, it said. And it cited some marginal markets where someone was energetic enough to drive a truck through the yawning maw of the bored monolith.

Yeah, right, everybody said. At the time, MS was just downright scary. It had the desktop OS market in a stranglehold and was quite effectively encroaching on Unixoid OSes' dominance on servers and machines for serious computation. It ruled supreme in the main productivity-application segment, having driven out most real competition. It was astutely capitalising on Internet Explorer. And it was increasingly inserting itself into new-media software that seemed to hold future importance, such as streaming (recall RealNetworks’ claims that Microsoft stole and used its software, which Microsoft settled years later with a large sum). So although it didn’t enter every market, and suffered a high-profile failure, despite years of heavy spending, to produce a server for streaming television on-demand, it was clear to all that Microsoft didn’t suffocate only those whom it couldn’t be bothered to crush. And Windows was both evil and eternal, like the Soviet Union.

Then there was a bit of the-rest-is-history. Judge Thomas Penfield Jackson actually tried to explain to the press his decision against MS; this was deemed unjudicial by the Court of Appeals for the District of Columbia Circuit, which eviscerated his decision. The G.W. Bush administration came to power and unsurprisingly settled on terms favourable to MS, leaving the states to twist slowly in the wind. Having triumphed, the monolith was only emboldened.

Gradually, though, the tide turned. Linux resisted and eventually turned the tide of Windows on serious servers. Smartphones were born and took years to overcome an awkward childhood; some ran Windows Mobile but others didn’t. Apple didn’t die as scheduled but rather used the last of its cash to buy the second coming of Jobs along with his strong OS engineering group, which finally (on Apple’s third attempt) gave it a proper OS. Although the slightly modified Unix-based NextStep took a long time to grow market share under the Mac OS X brand beyond Apple’s gauche and effete fans (and me), it ultimately did and was a key component in the iPhone — which turned the smartphone into an object of desire.

By 2009, Windows had a market share no longer of 95% but of 70%. However, given the flood of smartphones that this number already accommodated, one could perceive it to be a forceful assertion of Microsoft’s continued relevance and stability. Which would be bolstered by Windows 7, to be released within months, fixing much of what was wrong with the hapless Vista.

And then exactly what Microsoft predicted actually happened. In four years, Android emerged from /dev/null (nothing) and became the majority OS, with more than half the market (53% last January). iOS and Mac OS X together grew to 19%. Windows-branded OSes now control only 22%. See the top chart at (green is others).

Android rode in on the power of the unsatisfied market for smartphones (often any mobile phones) worldwide, coupled to their somewhat greater affordability than iPhones (I would guess that many consumers, especially outside the Golden Billion, never considered iPhones even where they were available). So Microsoft’s position now is not as shaky as it appears because it still has a large chunk of the desktop market (90% in June) and the one for lesser servers. It just didn’t benefit much from the smartphone and tablet revolution (where it has just 1%).

But that would ignore a little factor named synergy. There may have been little if any of it in the many mergers and takeovers — especially in the media world — that were peddled to shareholders on its basis in the 1990s, but it is a critical aspect of the popularity of computing devices. Microsoft leveraged the ubiquity of its key products to a tremendous extent on its way up: the operating systems — by allowing anyone to build hardware to use them, and both the OSes and Office applications — by not resisting piracy at all until recent years and halfheartedly even now. The latter forbearance is not due to technical challenges (it’s been possible to implement bulletproof but user-friendly license protection for many years) or ideology (which is to the contrary, as demonstrated by Gates’s 1976 letter to users on the consequences of piracy to innovation), but rather a way of making lemonade out of the lemons of piracy — in industrial quantities. There were so many copies of these products lying around (while WordPerfect and Lotus 1-2-3 had user-hostile copy protection) that the product lines’ penetration grew to become the path of least resistance — to be followed even when license fees were paid. The default choice. The automatic choice you made unless you had special needs or were strange. One operating system, one productivity suite, one Reich, one Führer.

The same mechanism is in play when one swtiches one device to an OS already in use on the other, or, in the case of Apple, the brand of one device in favour of the brand on another. This is where Microsoft failed. Windows Mobile was sad, bloated, buggy and broken, and the quality of the devices so awful (a touchscreen failure is a real impediment when there is no other way to interact with the device). It became apparent to anyone who cared that riding camels through eyes of needles was not a practical means of transport. But just because MS couldn’t do it doesn’t mean others can’t.

Apple did that for itself and also for others, by insisting on a working touchscreen and by validating the very concept of a functional smartphone. Now the big winner of this trend is Android. To many, a mobile phone means a smartphone and that means Android. Which also means Android on a tablet. All that’s left is for the deadly virus to go airborne, and I would be stunned if a version for full-fledged desktops and laptops didn’t follow.

As for Windows, I don’t, for one, believe that many care now that Windows 8 might be reasonably acceptable on mobile devices. The leveraging trend is against Microsoft and would be even if the company was much better at getting functionality and usability right the first time than it actually is. I expect its market share will unwind like a yo-yo. The fact that all competitors (except the moribund Blackberry) are based on some form of Unix — the only OS in history that got more popular over a long time, because it was done right from the beginning in darkest 1968 — will only help them to leverage each other and eventually turn the lights off in Redmond even faster. Sic transit gloria mundi.

I just hope someone rescues Excel, if only out of commitment to public service. Neo/Libre/OpenOffice are Potemkin villages without so much as a Motel 6.

Employment of miners’ canaries as a share of the workforce

When I started to learn programming 30 years ago, almost all the packaged software in the world was U.S.-made. There were a few examples of “Euroware” such as 4D, à la fois brilliant and bizarre, that burned brightly but briefly. And now look: ten of the top 25 software companies by revenue are European, and lower down the ranking, the U.S.’ standing is, if anything, much worse: many of the large open-source firms with their much lower earnings per copy are either based in Europe or do much of their development there: MySQL before the Oracle takeover (Sweden) or InnoDB (Finland) or the main Linux distros (U.S.-based, largely European-programmed). And the farther you go from mass markets, the more dominant Europe becomes. In my field (Audiresys stands for “audience research systems”), it has made almost a clean sweep.

And that’s a problem, my fellow Americans, because most of our manufacturing has gone and it is not coming back and we have nothing to maintain us in the style to which we’ve become accustomed, as alimony-seeking divorcées used to say, but high tech. Aaaaand... the powerful narcotic painkiller that is our country’s unique status as issuer of the world’s reserve currency is starting to wear off. The Chinese workers and bosses who make almost all our consumer goods (a) want ever more money for their products as their standard of living rises, and (b) are increasingly less amused by the increasingly funny money we must use to pay much of the bill.

In addition to that drug, we take some others that dull the pain: the pervasive sense of exceptionalism that some of us will still believe in even when we cross paths with Turkey (which, in PPP GDP per capita, we probably will between 20 and 40 years from now), and the very large numbers of dollars and jobs that even inconsequential phenomena generate in our economy, still the world’s largest, for now. (Although in most contexts, these obfuscate more than they reveal, they remain pandemic and very hard to avoid; this tells you something about how people misuse data.)

To avoid the former bias, one has to look at other countries (duh) and their and our trends. To avoid the latter one, one has to eschew amounts and use percentages.

And when one does that, one finds statements like this:

UN Conference on Trade and Development, Information Economy Report, 2012, p. 21 (the ranking of software companies by revenues is from the same source, p. 26).

For people with less-than-perfect eyesight, 0.6% of all employed people in the U.S. work in “computer software and services”. Not just for export, mind you, where our position is even worse, but to serve our presumably huge domestic needs as well.  The only mostly industrialized country with a lower percentage is Russia. Even Costa Rica (0.8%) and South Africa (0.7%) do somewhat better than we. India has the same 0.6% as we do. Even with all the outsourcing, that boggles the mind.

I would be surprised if this weren’t based on a lot of error (for one thing, do these numbers include the vast amount of in-house development and computer services in companies with other primary activities? Apparently not), but I would also be surprised if it affected the U.S.’s number significantly disproportionately. We’re not that exceptional.

This is not to equate “computer software and services” with high tech generally. We produce a lot of high tech in which it is a modest component of the cost (pharmaceuticals, for example). But this field, whether it includes in-house employment or not, is the canaries in the mines. It is necessary for just about everything, especially other high tech. Thus, a low number of canaries indicates a low number of mines—or the use of other kinds of carbon monoxide detectors (you know, for’ners). As the UN agency says: “Boosting software employment not only helps to build up the software sector itself, it also has downstream multiplying effects. Moreover, jobs in software and IT services can help attract skilled young people”. We are going the other way.