Lunar trajectory

(c) NASA

When Neil Armstrong and Buzz Aldrin stepped out of the Lunar Excursion Module in 1969 I remember a feeling a wonder, awe and hope.  I was eight years old and had followed the Apollo program from Apollo 8, the first manned flight crewed by astronauts Borman, Lovell and Anders who were the first men ever to fly around the moon. 

In what seemed an eternity later but was in fact only six months, Armstrong and Aldrin planted the American flag on the lunar surface.  The Apollo program was only three years old, and less than a year since the first Apollo astronauts had left earth^*

Five more moon landing were made until the program was terminated after the flight of Apollo 17 in 1972.  It had taken only three years from the first manned flight of the Saturn V to the sixth lunar landings. In hindsight, the speed with which the program moved forward was astonishing. That was fifty one years ago. 

In 1989 the Soviet Union collapsed and the "space race" effectively ended. The Space Shuttle never seemed to fulfill its original promise and didn't capture the public's imagination as the moon landings had. America turned inwards to become increasingly consumed in what we now think of as the culture wars. 

Existential challenges confront the country and the world yet America's increasingly chronic inability to tackle anything consequential, at least with any success, imperils the country and the wider world. Two wars, in Afghanistan and Iraq, failed to achieve their objectives, the wealth gap has widened and trust in institutions has all but evaporated. 

The feeling I had as I watched man set foot on the moon, the sense that there was almost no limit to what could be accomplished, has given way to the realization that America's sclerotic politic and societal dysfunction has turned the dream of a bright future, brimming with possibilities, into to a bleak and deeply depressing nightmare         

^*Tragically Gus Grissom, Ed White and Roger Chaffee, the crew of Apollo 1, died in a fire on the launch pad in Feb 21st, 1967.

Elizabeth Holmes and Theranos

Elizabeth Holmes was convicted on four counts of wire fraud in a court in San Jose yesterday. The case should have been about the appropriateness of business models widely used in Silicon Valley for software development to healthcare.  

Writing buggy code and fixing it after it breaks, a model arguably pioneered by Bill Gates with Windows, is, if not acceptable, reluctantly accepted by most of us since the consequences of an application behave in unexpected ways are usually not too terrible. 

But in healthcare (and self-driving cars) the consequences of mistakes from rushing a product to market are far more serious, raising the question "is a business model that emerged at the inception of the PC industry appropriate in setting where the consequences of mistakes may be deadly"?

The focus of Holmes' trial on the hurt to shareholders is unfortunate; perhaps that's what the prosecution could most easily prove. But in missing the distinction between cases where the harm is potentially lethal rather than simply annoying (or expensive), the case fails in building a precedent that might protect people from physical harm from poorly designed products that their producers know to be faulty.  At the same time it creates a needless damper on innovation in sectors where the consequences of bad design are irritating but not life-threatening.

No time to loose

A few years ago I bought several cheap but stylish watches. Each was about $20, so I decided to have one for summer time and one for daylight savings time. That way I'd never have to adjust my watch when the clocks changed. 

Early this morning the clocks went back and I put on my daylight savings time watch. I hadn't worn it since March 13^th and almost certainly hadn't adjusted for several weeks before then. 

Checking against my phone and my desktop clocks, both of which are synced to internet time, my cheap $20 watch is seven seconds fast. Since at some point last year I did make sure is was within a second or so of internet time, it has gained only seven seconds in six months or longer. 

When I think back to my second watch, a Tissot my father gave me when his colleagues gave him a new one on his retirement, that Swiss movement made by a company renowned for its accurate time pieces, had to be adjusted practically every day.  That's the kind of remarkable progress which comes with the passage of time. 

Thumbs down

Facebook, WhatsApp and Instgram were all off-line this morning. According to the New York Times "it was unlikely that a cyber-attack caused the issues. That’s because the technology behind the apps was still different enough that one hack was not likely to affect all of them at once."  That suggests that all three platforms were taken down deliberately by someone inside Facebook, perhaps even by the company itself. Why might they do that? Perhaps its intent is to rile up its users to by showing how much people rely on these platforms in the hope that their dismay at begin deprived of Facebook's services can be channeled into preventing Congress (or the DoJ) from legislating or ruling to curtail the company's out-sized influence.  

Utilities

Utility, noun: "an organization supplying the community with electricity, gas, water, or sewerage".  Over the course of a century the term "utility" has come to be synonymous with "always on". Yet as PG&E is now demonstrating that is a misnomer and a misconception.  The adjective, "the state of being useful, profitable, or beneficial" (which is closer to its use in economics) suggest that anything which we want or find to be of use has "utility". But to expect it to be "always on" is only a function of habit and a developed country habit at that. There are may parts of the world in which electricity supply is intermittent; and California is now one of those places.   

TikTok, Donald

Why is Donald Trump fuming about TikTok and threatening to ban it? Wired has a long piece on the threat of Chinese eavesdropping.

Here's a much simpler explanation; revenge.

TikTok users were instrumental in over-booking Trump's last campaign rally (and the only one he's likely be able to hold, absent some particularly reckless decisions on his part) before the election. So instead of a packed stadium we had images of rows and rows of empty seats.

He was made to look foolish; which in his world is the most egregious sin imaginable. So he's out for his pound of flesh. I think that's a far more Occam's razor-ish explanation than Wired's. 

Google's mantra

(c) Time Magazine; https://time.com/4023367/google-china/

Google's slogan / mantra / motto / values statement, "Do no evil", sounds great. And not withstanding Eric Schmidt's somewhat jaded observation that "Evil is whatever Sergey says is evil", it seems like a good place to start.

But there's another way of thinking about it; specifically in terms of freedom of action. A seemingly converse statement "Do good" is in fact far more constraining. There are lots of courses of action that might be neither good nor evil, which are precluded under the latter dictum but available to Google. "Don't break the law" is very different from "be kind to everyone".

While Facebook seems to neither care about the morality of its decisions, nor about the morality of obfuscating and dissembling to Congress and the public, Google was, for many, a beacon of hope that at least one of the the tech giants might be the bridgehead to a more socially responsible flavor of capitalism. I'd say the jury is decidedly still out on that question.

Learning to drive

There isn't a Tesla autopilot in each Model 3. There is just one, cloned across all models on the road. And Tesla owners are all teaching it to drive.  There are more than 600k Model 3's with autopilot on the roads, all collecting data to improve the autopilot's intelligence. While autopilot isn't as smart as a person, it has had far more driving "experience" (about 16 billion mile) than any single human driver, which will eventually make it safer than the safest human driver.

The need for AI's to gather oodles of data to perform well creates an enormous network externality. As more products incorporate AI, the data economies of scale will turn competitive markets into create winner-take-all industries which without antitrust regulation will mean a series of uncompetitive monopolies. Google is already at that point in search. Only where function follows fashion, for example in social media will industries see the unseating of incumbents.  For free market enthusiasts that presents a problem. It's time to create an government entity to deal with market power and concentration.   

Solid state

The "Last Build" is finished and running. Central is the 8 core 16 processor CPU; it is maxed-out on with 64 GB of RAM and an m.2 solid state 960GB disk. It is hampered, however by the slow speed of the Samba NAS and the even slowed speed of the cloud Google drive.

Two 4k monitors mounted on a stand that keeps them off the desk provide more screen real-estate and more room for piles of papers underneath them. Two 120mm fans attached to the radiator provide reasonably quiet cooling, the simple graphics car has no fan, and there are no disks (and therefore motors) in the case to drown out the sounds of the crickets outside and the ticking of my watch inside.     
 

The Last Build

For several years I have been contemplating building a new desktop computer. Looking back through my Newegg order history, I found that I had  been using the Intel Quad core Q9300 since May 2008!  Time flies ever faster these days.

The Last Build has been a wish-list on Newegg for at least two years; this May, with some heavy data analysis for CATOE on the horizon, I decided to pull the trigger, on an almost no compromise machine. That didn't include going crazy with whizzy graphics card, but did include:

• AMD Rayzen 7 1700X 8 core processor
• 64GB of RAM 
• Water cooling
• 960MB M.2 SSD (and no Winchester (mechanical) disks!) 
• Mint 19.1
• GIMP 2.10 (which supports 16 bit image files)

Issues so far include a problem with keyboard lag which for the moment is solved by plugging the mouse's USB receiver (!) into the front rather than the back of the box, Code::blocks crashing every 3 minutes or so, and the DVI-display port cable not working.

Otherwise the system is fast and quieter then the old machine. I am not going to overclock and have no case fans other than the two on the radiator, all in the interests of lowering the background noise. 

This configuration should last me at least until retirement (and probably the rest of my life). That's a sobering thought. 

Less Autonomous Vehicles

Autonomous vehicles (self driving cars) are supposed to be arriving in about 2 years. There are safety concerns of course, as highlighted by the unfortunate death of a pedestrian in Arizona last week. But until legislation arrives their progress and roll out will continue unabated.

Safety features are of course critical to their deployment; it has been suggested that because so many accidents are caused by errors in human judgement, self driving cars will actually made the roads safer.

Ultimately that's likely to be true but in the phase in which self driving cars and human-driven cars coexist on the same roads, autonomous vehicles will need an over-abundance of safety devices to cope with unpredictable humans in their environment.

In time human driven cars may become rare enough that autonomous vehicles will be far more likley to be surrounded by other self driving cars then their traditional predecessors. Coordination between cars - a network of vehicles talking to one another - might then provide information about road and traffic conditions as well as the ability to 'flock' (as in for example, computer simulations of flying geese).   

At that point, autonomous vehicles may be a misnomer; while today's cars are tightly coupled to their drivers but autonomous from other cars on the road, self-driving cars will be autonomous from their occupants but closely coupled to other cars, road furniture and other highway features. 

S/360 legacy

IBM dominated the computer industry in the 1960s and 1970s with its S/360 architecture mainframes. With the introduction of the PC, IBM's ability to control accounts from the center - a key part of its sales strategy - disappeared, and the company has been trying to find its way in a "post S/360 world" ever since.

But despite the rise of client-server and the web, S/360 will be around for a while. Companies invested in COBOL applications and customized MRP systems that can't easily be replaced and indeed in might not be cost justifiable to do so. So MVS, VSE, VM, IMS, CICS and CMS are likley to be around for quite some time.       

Corporations face a dilemma - when their core business is shrinking slowly but not going away, they are held hostage by the past, committed to supporting those legacy systems and customers and unable to make a clean break (even if they could come up with another 'killer app", which is anyway far from certain).

Microsoft is in the same position today that IBM was twenty years ago.  And Apple may well be there twenty years from now.

As the 21st century matures, the tech landscape may well become littered with semi-zombie companies that are dying but haven't quite stopped moving yet...   

Atrophy

Early reports suggested that the pilots on Asiana flight 214 didn't check their air speed on their final approach because they thought the that plane's automated systems were taking care of that. (Is it still astonishing that only three people out of over three hundred on board perished in this terrible accident).



In the discussions following the crash, several experts have noted the issue of pilots losing their touch as electronics do more and more of the flying. While this is clearly of concern to air travelers, it is possibly the thin end of the wedge of a bigger problem that will affect a lot more people; I'm thinking here about Google's driver-less car.

I'd love to have a car that drove itself (and me) to the office. I could get a lot of work done in those two and a half hours (each way).  But software isn't, and those who create it aren't, perfect, and there will be situations in which a pilot or a driver may have to take back control from the auto-pilot.

There are to issues here. First, if pilots and drivers don't get the hours under their belts they may not become sufficiently skilled to act decisively and appropriately to avoid an accident.

Second, we need to thing carefully about designing machines that are too complex and unstable that they can only be operated with a computer intermediary. Modern fly-by-wire jet fighters, the F-117 in particular, are like this; they are so aerodynamically unstable that without computers to continually monitor plane's motion and apply minute but critical adjustment to the the control services, it would fall out of the sky; human pilots simply couldn't response fast enough and in the right way to keep the machine aloft. Imagine if we started building cars that we couldn't drive without technological assistance to control them.

In fact we probably we have already. From power assisted brakes and the simple end of the spectrum, to the Prius' fly-by-wire throttle at the other, we're already moving determinedly down that path. Remember all the kerfuffle about the Prius' throttle jamming open? That doesn't augur well for a fully automated, self-driving car.  Do we want the most efficient (high value, low cost) possible system that works 99.9% of the time, while risking calamity in the 0.1% of the time it fails?

Returning to a related theme, that's exactly what we've done with our food supply and our manufacturing supply chain. So the answer is probably "yes".

The last mile becomes the last 72 inches

Fifteen years ago, when I was collecting data for my thesis from a European telco, the threat all the company's employees, particularly the senior managers, were worrying about was VOIP. I confess I had only a vague idea they were talking about; I knew what it meant but somehow it didn't really grab my imagination. Yesterday, after installing my own analog telephone adaptor (ATA), I now realize in an oddly viscerally way what they meant. More colloquially, now I get it.   

In the 90s, the trunk network, the backbone that links local exchanges together, was transforming from analog to digital. I talk about this in class when we look at Alcatel in Spain. Huge rotary telephone switches in the locals exchanges were being replaced by small(ish) computers. But for residential customers, the last mile has remained stubbornly analog. Until now - at least for me. After setting up my ATA, my analog portion is now down to about 6 feet; and if I had a shorter cable it would be 3.