Computers and Visual Data

Regular readers will know I am skeptical self-driving cars are just around the corner, in part due to the limitations on computers being able to interpret visual data. As I said before:

The recognition of objects by computers is still only in its infancy, and nowhere near robust enough to deploy in any safety-critical system.  Given the pace of development of other areas of computing abilities, such as sound recognition in apps like Shazam, object recognition is seriously lagging behind and I suspect for very good reasons: software, being made up of pre-programmed algorithms, simply isn’t very good at it.

Computers are unable to do this for anything other than the most basic of pre-determined objects and scenarios, while the environment in which humans operate their cars is fiendishly complex.

Via Tim Worstall, I found this story:

Amazon’s facial recognition technology falsely identified 28 members of Congress as people who have been arrested for crimes, according to the American Civil Liberties Union (ACLU).

There are clearly instances where facial recognition technology works, but it’s still not very good and I suspect will remain that way for any conditions which are not perfect (i.e. good light, looking face-on, and of an expected skin colour) for some time yet. I’d be interested to know if the Chinese system which was deployed to fine jaywalkers works as claimed; it’s hard to believe anything which comes out of that country.


Chinese Cars

I’m not convinced by this:

Fifteen years ago, if you rode around in a Chinese domestic branded car, they felt like copies of Japanese, Korean, or German vehicles.  That’s because in most cases they were.  Chinese car companies that were not operating through a joint venture with a foreign brand and who therefore had to rely on their own designs largely copied popular models that were selling well.  The copying was legendary; a colleague at Toyota told me that they had a lot of respect for one of the Chinese companies in particular – their copy of one of Toyota’s popular models was so good you could take the doors off of it and put it on the corresponding Toyota and the door seals would work perfectly.  Firms in China first copied designs, then sourced parts, and finally built a factory and mass produced the car.


While some of that still goes on, leading Chinese automakers have moved on.  They have been aided by sophisticated computerized design tools that allow them to do their own design and modeling, a phenomenon that is becoming more and more important as know-how gets embedded in tools.

Designs of what? Bodywork? Engines? Gearboxes? Engine management systems? Without an example, he might be describing a copied Toyota with different bodywork.

These days, if you ride around in some of the leading Chinese brands, you will find sophisticated engines, turbochargers, complex automatic transmissions, and high levels of interior detailing.

Yes, I’m sure you will: but are they copies? And could the author tell if they weren’t? There’s a sleight of hand at play here which makes me wonder if I’m being told a load of bollocks.

The Chinese government is using the transition to electric as a way to surmount the historical advantages of incumbent car companies and nations.  It won’t be saddled by existing infrastructure for making internal combustion engines and old ways of working; it starts with essentially a clean slate.

This assumes electric cars aren’t some giant white elephant, hoovering up resources which would be better applied elsewhere. As I’ve argued before, without a major step change in battery technology, electric cars are a non-starter.

BMW recently announced plans to source batteries from Contemporary Amperx Technology Ltd. (CATL), and Volkswagen is focusing its electric vehicle development in China.

Is this because Chinese technology is wonderful or because their environmental legislation is non-existent, labour costs are low, and there are a billion potential customers outside the factory gate?

Tesla just announced plans for a new factory in China that will produce 500,000 vehicles a year.

Tesla announces grand plans on a weekly basis; few of them ever come to fruition. As an author of articles on electric vehicles, you should know this.

Chinese automakers and parts suppliers will harness the Internet in ways that haven’t been imagined in the rest of the world

Really? Those clever Chinese, eh? How will they go about this, then?

One of the things I wanted the students to see is how the Internet is evolving very differently in China, because of its “isolation” and the very different boundary conditions there.  For example, Government regulation tends to come after a new business idea has developed in the market, rather than beforehand as is generally the case in the rest of the world.

So the Chinese – who lock down the internet more than anyone else – are at an advantage because they can develop new internet-based business ideas before regulations catch up with them, and this will give car manufacturers an edge…somehow. Like many great ideas, it’s obvious once explained.

Also, physical delivery in major Chinese cities is very inexpensive, typically costing just over $1.  So it is remarkable how much Chinese consumers order things online – merchandise of course, but also lunch and al kinds of services.

Unlike in the US, Europe, and Japan where internet retail is a decade away and everyone traipses around shops. And what this has to do with car manufacturing is anyone’s guess.

We visited an automaker that offers a built-in WiFi hotspot with lifetime provisioning included.

Well that does it, I’m off to buy a Donfeng.

You can imagine the implications for turning the car into a platform.  Connected and self-driving car technologies are a natural amalgamation of complementary technologies that China seeks to dominate.

I think the guy who wrote this has come from rural China where he rode a bicycle all day. Western-branded cars have been connected to the internet for years – my own comes with a built-in SIM card – and it can best be described as a solution searching for a problem. Other than the real-time update of traffic conditions, an online search function in the GPS, and some remote-controlled functions it is largely useless. When I bought the car in 2014 the brochure boasted I could update Twitter from the car. The other day I received an email asking if I wanted to install MS-Office in my car, presumably so I could compile spreadsheets from the outside lane of the Autoroute. Any other online services I need while in my car can be accessed from my phone: none of this is new.

While the American public has a world view of the global auto industry centered on North America, Europe, and Japan, we should pay more attention to what’s going on in China. That’s where a lot of the action is going to be coming from.

I’m wondering who paid this guy and how much.


Getting wood over wood at The Economist

Part of the decline of The Economist, aside from the fact its employees write drivel, is its wholesale adoption of the environmentalist religion. With their latest video they seem to be plumbing new depths of woo-embracement:

The answer, of course, is no: wood has been used as a construction material since the dawn of time, and in the modern age there is probably not a thing we don’t know about it. Concrete and steel replaced wood for very good reasons, and unless wood has undergone some revolutionary step-change (e.g. trees grown with carbon-fibre grafted into them), those reasons still apply. If it made technological sense to use wood instead of steel, people would be doing it. If it made economic sense, the same would be true. But let’s take a look at the video (I’ll paraphrase rather than write the whole transcript).

0:25 The world’s population is increasing, by 2050 it will be 10bn most of whom will be living in cities in skyscrapers with a large carbon footprint.

The video shows Tokyo and other developed world cities, but almost all that population growth will come from Africa. Are they going to be living in high rises? Having seen the sprawling shanty towns of Lagos in person, I doubt it. And if “carbon footprints” are a problem, maybe its time to stop subsidising that population explosion in Africa? One of the main reasons Nigeria’s population is exploding is the lack of reliable electricity, which in turn is a direct result of corrupt government practices. What I’m trying to say is, if increasing populations are a concern, building materials are an odd thing to focus on.

0:30 Our view is all buildings should be made from timber, and we should look at steel and concrete as we do diesel and petrol.

I have no idea who this chap is, but he’s looking at a Landcruiser and trying to say a horse would be better. I suspect he’s saying this because his salary depends on it.

0:44 I think it’s realistic someone will build a wooden skyscraper in the coming years. There is a lot of potential that is unrealised for using timber at a very large scale.

It’s as if engineers are unaware of wood’s limitations in compression. Hell, even the Romans knew over a certain size you had to use stone and concrete.

1:00 Throughout history buildings have been made of wood But it has one drawback, it acts as kindling.

Don’t ever say Economist videos aren’t informative.

1:32 If concrete were ever to arrive as a new material on “Dragon’s Den”…but then you say we need a whole new fleet of trucks to move it around…

You can tell this guy is an academic. Firstly, there are transport costs associated with wood; they don’t grow trees on potential building sites and wait a hundred years. Secondly, the cost savings associated with using concrete obliterates the additional cost of needing specialist concrete trucks. It’s one thing to play devil’s advocate for some future hypothetical, but this guy is doing it for something that’s already happened: he’s already been proven wrong.

1:51 I don’t think it would be a compelling case.

The richest man in Africa is a Nigerian called Aliko Dangote; the bulk of his wealth comes from his owning Africa’s largest cement company. The invention of concrete revolutionised construction, and made an awful lot of people incredibly rich. But here we have an academic saying if it came along nowadays, nobody would be interested because you need to add steel and buy some specialist trucks.

1:58 Concrete and steel are costly to produce and heavy to transport.

Compared to what? This is like saying the weather is good.

2:05 Wood, however, can be grown sustainably and is lighter than concrete.

Weight doesn’t matter much in buildings, because they tend to be stationary objects supported by the ground. You also have a lot of glass curtain walling these days. If weight is a concern you use steel – as the Manhattan skyline nicely demonstrates. Insofar as transportation costs go, aggregate can be shipped cheaply in bulk from anywhere, and you can install a concrete batch plant on or near to the construction site. A someone who lived in Dubai during the construction boom, I saw a lot of this.

2:07 And crucially, as trees grow, they absorb carbon dioxide from the air, locking it into the timber.

This is crucial? Not to construction considerations it isn’t. If you want trees to absorb carbon dioxide then plant more trees, but to put this forward as an advantage for using wood in construction? You might as well say forests are nice places to walk a dog. In any case, unless these buildings will stand for centuries, at some point the wood will rot or burn releasing all that carbon dioxide into the atmosphere anyway. Why not leave the trees standing?

2:18 One study showed that by using timber to construction a 125-metre skyscraper could reduce the building’s carbon footprint by up to 75%.

One study…could…by up to. Well I don’t know about you, but I’m convinced! Note all this assumes a building’s “carbon footprint” is something we should be concerned about.

2:42 Wood isn’t strong enough to build high, but engineers have come up with a solution: cross-lamination.


2:45 It’s cross laminated so layers of wood are glued at 90-degrees to one another.


3:17 But what about fire?

They demonstrate how a skyscraper made from wood will withstand a fire by holding a blowtorch to a piece of plywood before claiming it will extinguish itself after losing “some structural mass”.

3:25 We’ve actually seen steel roofs collapse in fires when wooden ones have not.

Assuming this is true, this is an argument for making sheds from wood, not skyscrapers.

3:52 Once these wooden panels arrive on site we’re building a floor a week.

Right, but it’s essentially a 5-storey plywood box. Are you sure this method is going to work for skyscrapers with 50 plus floors?

3:57 This is maybe twice as fast as concrete.

The guys in Dubai were pouring a floor every few days. I’d like to see how fast these wooden panels go in when they’re a hundred metres above the pavement.

4:23 Andrew and his collagues designed Britain’s first wooden high-rise apartment block.

It’s ten floors, hardly high-rise.

4:51 As yet, nobody has used CLT (plywood) beyond 55 metres.

The building they refer to is Brock Commons tower in Vancouver:

The structure is concealed behind drywall and concrete topping, mainly to comply with the accepted fire-safety codes and consequently speed up approval from building authorities.

So it needs concrete to stop it turning into a matchbox, incinerating everyone inside. But wait, what’s this?

Due to concerns about structural stability, the American Wood Council and the International Code Council currently limit wood structures to a maximum of six stories above grade, depending on occupancy type.

For good reasons, I’d imagine.

To reach its height of 18 stories, Brock Commons used a slightly different approach. It follows in the shoes of the supertall skyscrapers we’ve seen cropping up across Asia and the United Arab Emirates (UAE), which use a central structural core to take the stress off of the building’s exterior.

Oh! What type of central core?

Two concrete “trunks” on a concrete podium form the core of the structure, with the rest of its 18 stories being constructed of cross-laminated timber (CLT) flooring and glue-laminated timber (GLT, or glulam) columns.

So this groundbreaking tower block which demonstrates the viability of wooden skyscrapers is held up by two, bog-standard concrete cores? The Economist never mentioned that.

This entire video is basically a puff-piece for a London-based architectural firm with its eye no doubt on government monies earmarked for eye-catching green “solutions”. Wood can be used effectively for construction, but it has severe limitations which are well known: warping due to heat, rotting due to damp, termites, separation of lamination with time – and the ubiquitous fire hazard. I’d love to see how well this Brock Commons tower is holding up in a decade’s time, and hear it from the poor sods who have to live in it, not the architects. This is before we even address such issues as increased land use to grow the trees, not to mention the wastage. The good thing about steel and concrete is it can be moulded to the shape you want without wastage, but wood has the tendency to be grown tree-shaped and from there you need to chop, saw, shave, and sand it into something useful – all of which creates mountains of waste product (when I was a kid, timber merchants used to give away wood shavings and sawdust for free). So what happens to that?

How many trees occupying how much land are needed to build a 100m building, and how much waste is involved? And how much chemical treatment does the wood require? Some numbers would have been nice, but this is The Economist: when it comes to the environment they sound more like The Watchtower.


Password Managers

Can anyone recommend me a good password manager? I’ve not got any problem keeping track of my passwords, it’s just I’ve had the same two or three passwords across dozens of platforms for several years and I probably need to start updating them all. So rather than do that, I’m looking at password managers.

Are they secure, i.e. better than normal passwords? How do they work if you have multiple computers plus an iPad? I don’t mind paying for one if the free versions aren’t very good.

I’d be grateful for any advice.


Thin Skin

It seems Elon Musk’s submarine really was crap:

A British cave explorer who recently helped rescue 12 boys and their soccer coach from a Thailand cave network slammed Elon Musk’s child-sized submarine proposal as a “PR stunt.”

The Tesla CEO had traveled to Thailand earlier this week, touting his “mini-sub” that he said was built with rocket parts and could be carried by two divers.

But British caver Vernon Unsworth told CNN that Musk “can stick his submarine where it hurts.”

Unsworth, who lives in Thailand and has spent years exploring the Tham Luang cave system where the boys were trapped, said Musk’s submarine wouldn’t have made it past the first 50 meters into the cave after the dive start point.

“It just had absolutely no chance of working. He had no conception of what the cave passage was like,” Unsworth said. “The submarine, I believe, was about five-foot-six long, rigid, so it wouldn’t have gone round corners or round any obstacles.”

When a CNN journalist pointed out that Musk had been inside the cave, Unsworth shrugged.

“And was asked to leave very quickly. And so he should have been,” he said.

I expect we’re going to hear more feedback along these lines as the rescuers start telling their stories. Given the way Elon Musk has reacted to Unsworth’s comments, he seems to have struck a nerve:

Musk lashed out on Sunday, saying he would make a video proving that his “mini-sub” would have been successful and adding: “Sorry pedo guy, you really did ask for it.”

The accusation, presented without evidence or context, was directed at Vern Unsworth, a British cave explorer who recently said Musk’s attempt to help the rescue effort was a “PR stunt”. No evidence has emerged to substantiate Musk’s claim of pedophilia.

On Sunday, when a Twitter user pointed out that Musk was “calling the guy who found the children a pedo”, the billionaire responded: “Bet ya a signed dollar it’s true.”

Intense criticism followed. Some Twitter users pointed out how “dangerous” and irresponsible it was to make such a serious allegation and to broadcast a potentially libelous insult to his 22 million followers.

Spokesmen for Musk and Tesla did not immediately respond to requests for comment. Later on Sunday, Musk deleted the “pedo” tweet and its follow-up.

Publicly calling someone a pedophile because they didn’t like his submarine PR stunt is pretty low from Musk, and there’s a good chance he’s going to get sued for libel. There’s really something not quite right about the man, at least to my eyes; I’d not be surprised if he does a stretch behind bars at some point in his life.


Elon Musk: help or hindrance?

Elon Musk has busied himself over the last few days tweeting out possible innovative, hi-tech solutions for rescuing the Thai kids from the cave.

Of course, many people think this is great, likening him to a real life Tony Stark deploying secret technologies in humanity’s hour of need. Ever the skeptic, I’m not so sure. I have no doubt that Musk means well and genuinely wants to help, but he’ll not be unaware of the PR value of this (and he seems to be as good at PR and self-promotion as anything else, to be honest). But he’s not there at the scene, and I suspect a few of the rescue team are getting a bit annoyed about having to divert time and attention to respond to whacky ideas from a billionaire sat on the other side of the world. Imagine you’re trying to work out how to get kids through a 15″ tunnel filled with water when this arrives:

Okay, that wasn’t from Musk, but you get the idea. I’m sure he means well, but I suspect his input is less a help than a hindrance.


Printer Advice

Does any of my readers know anything about printers?

I’m looking to buy a printer for home or small office use, so massive speeds and loads aren’t important. I also need a scanner, so would want one which also has that capability. I’m thinking of going for laser rather than inkjet, mainly because I find the quality much better on lasers and inkjets tend to dry out unless used regularly, and this printer might be sitting idle for long periods. Colour printing isn’t important – I can get by on black and white – but is the price difference these days such that I may as well get a colour one? Any brands to avoid? Finally, I don’t need some huge thing: compactness would be desirable. Budget: not entry level but not hi-end pro either. Low-end pro is about where I’m at.

Any advice or recommendations?


Death by Self-Driving Car

I’m not surprised by this:

THE POLICE HAVE released video showing the final moments before an Uber self-driving car struck and killed 49-year-old Elaine Herzberg, who was crossing the street, on Sunday night in Tempe, Arizona.

The video includes views of the human safety driver and her view of the road, and it shows Herzberg emerging from the dark just seconds before the car hits her. And based on this evidence, it’s difficult to understand why Uber’s self-driving system—with its lidar laser sensor that sees in the dark—failed to avoid hitting Herzberg, who was slowly, steadily crossing the street, pushing a bicycle.

The video itself is here:

As I said after that self-driving Tesla crashed in 2016 having failed to differentiate between the sky and the side of a lorry:

The mistake people make is to assume every action in driving is one of simple measurement, and conclude that computers are far better at measuring things than humans are in terms of speed and accuracy.  However, driving is often about judgement as opposed to pure measurement (and this is why it takes a while to become a good driver, judgement improves with experience), and much of this judgement relates to the interpretation of visual information.  The recognition of objects by computers is still only in its infancy, and nowhere near robust enough to deploy in any safety-critical system.  Given the pace of development of other areas of computing abilities, such as sound recognition in apps like Shazam, object recognition is seriously lagging behind and I suspect for very good reasons: software, being made up of pre-programmed algorithms, simply isn’t very good at it.  And even then object recognition isn’t enough, a self-driving car would need to be able to not only accurately acquire visual data but also interpret it before initiating an action (or not).  Computers are unable to do this for anything other than the most basic of pre-determined objects and scenarios, while the environment in which humans operate their cars is fiendishly complex.

This latest incident doesn’t do much to convince me otherwise. And here’s Longrider:

A human driver can look at a situation developing and reading body language and little subliminal cues to react long before the situation has become dangerous, long before the AI has even suspected that a situation is developing.

Quite. Also:

This woman was unlawfully killed and someone, somewhere should be prosecuted for it. If it was a human driver, then that would be straightforward. However, now we get into who programmed the damned thing, how was it set up, what failed? But bottom line here is that someone was criminally negligent and should be doing jail time.

After the Tesla crash, I said:

What does amaze me though is that computers are being put into cars with the belief that they can do things they demonstrably can’t.  A hefty lawsuit and tighter regulations can’t be too far away.

Both politicians and the public seem keen on self-driving cars being rolled out onto public roads while still dangerously unsafe, wooed by the idea that “technology” is safer than humans and any errors – even those resulting in someone’s death – can be rectified with a few lines of code. Someone’s going to make a lot of money out of these things, and it won’t be the manufacturers.


A Lightbulb Moment

So today I go into France’s biggest DIY chain to buy a lightbulb. The type I’m after is one of the small, spiral-wound fluorescent bulbs which fits neatly into a ceiling lamp. Only they no longer sell fluorescent light bulbs: they’ve all been replaced by LED bulbs, which are better. But they’re not the same shape, having a more bulbous glass piece (similar to the old incandescent ones) for the same connection size, and it doesn’t fit the lamp.

No doubt I can buy the bulb elsewhere, but answer me this: which fucking dickhead persuaded governments worldwide to force hundreds of millions of people to switch to fluorescent bulbs when LED bulbs, which would make them obsolete, were just a few years over the horizon? Can we get hold of them and string them up by the balls please, followed by each and every politician who went along with this idiocy?


Material Technologies

Thanks everyone for sharing enough stories to make me feel comfortable that I was not alone in driving a dangerously unsafe car back in my youth. Like I said at the beginning of the post, there was a time when taking possession of your first banger was a rite of passage for young men. (It may have been different for women: the first car of my ex-housemate was a Nissan Micra that, although in very good condition, was ginger in colour.)

Back around 2001 I had to commute between south Manchester and Warrington, so bought a brand new Renault Clio, 1.2l at the very bottom of the range (I got a new one because I knew it would work and the finance deal was pretty good). It was about as much a girl’s car as you could find, especially being bright red, but having done that job delivering cars all over Manchester I knew what it was like to drive compared to Fiestas, Polos, and the other cars in its class. I went for the bottom of the range of a small car simply to save money. I had that car for about two years and not a thing went wrong with it, it was perfect except for being a little to small for my legs. On long journeys, my knee would hurt.

What I noticed when I lovingly washed it every weekend was how much of it was actually plastic. The front wings were, and the sills covered in a rough plastic coating which didn’t chip easily. Parked outside in Manchester weather, there wasn’t a spot of rust on it even after two years. When I walked to work this morning, I tried to spot a rusty car on the way. I didn’t see one. But back when I was growing up in the 1980s? Oh boy. I read stories about how British Leyland would stamp out car panels in one factory, load them onto an uncovered flatbed truck, and drive them through the rain to be painted and installed elsewhere. Little wonder they started rusting from the moment you took it off the forecourt. It wasn’t just British cars, though. My parents had a VW Beetle which one of my school chums nicknamed “measles”, and we had two successive Mark I Golfs whose wheel arches rusted through in a few years. Back in those days, Halfords used to sell sheets of wire mesh that you’d fix over the gaping holes in the bodywork and cover with a sort of polyfilla, then sand it smooth. If you were lucky, you’d find some paint to match but a lot of people just left it at that. I doubt anyone does this any more, save for those working on classic cars.

What’s changed, aside from the demise of state-run disasters such as British Leyland, is materials technology. The steel will be of a higher quality these days, and I expect all cars are galvanised as standard. Plastics are used wherever possible, and the coating and painting systems will have advanced beyond recognition. The paint on old cars used to be very brittle, and would flake off around a stone chip. Nowadays the paint is more rubbery, and stone chips cause small pitting but don’t usually penetrate to the metal.

In my lifetime, the two massive advances in technology have been the internet and mobile technology. These have overshadowed other advances which are possibly of equal importance in terms of quality of life and wealth. Being an engineer, I have a habit of looking at modern equipment and comparing it to the kit I grew up around. The difference is incredible. Clothing is an obvious example. When I was an army cadet between 1992-1996, we were decked out in Falklands-era uniforms: heavy cotton smocks and trousers, woolen military jumpers, 58-pattern webbing made of a sort of woven canvas. This stuff was only waterproof if you sprayed it, and although it kept you warm even if wet, it trebled in weight and took a week to dry. A few of us got hold of Norwegian army shirts made from towelling, which were very warm but if they got wet the arms would increase in length by about fifty percent. By the mid-90s Gore-Tex was well established in civilian clothing lines, and fleeces were starting their period of dominance which continues to this day. Nowadays when I go hiking or skiiing, I’m amazed at how warm, light, and waterproof everything is, and not just the clothes. Footwear, tents, rucksacks, head torches, straps, buckles, and every other piece of equipment is now made from plastics optimised for that precise application. I’m sure the same is true for other pursuits, too. I don’t know much about sailing and nothing at all about golf, cycling, or motorbikes, but I’m confident the material technology in these areas is space-age compared to what it was in the 1980s.

Back then, when we went on holiday, my parents used to pack our clothes into these brown vinyl suitcases in the traditional style. They were awkward, not very strong, and the straps were splitting. Wander into a Samsonite store now and you’ll see suitcases which look to be made of body armour that weigh nothing. Even the arms of the glasses I’m wearing now are made from carbon fibre. In fact, it’s hard to imagine a piece of equipment whose usability, quality, durability, weight, and ergonomics hasn’t been improved massively thanks to the invention and adoption of new material technologies. It’s something a lot of people probably miss, blinded by the more obvious technological changes around them. We probably ought to give a small nod to the men and women who brought it about though, especially when people start railing against hydrocarbons.