* Posts by Kristian Walsh

1690 publicly visible posts • joined 10 Apr 2007

Intel starts mass production on Intel 4 node using EUV in Irish fab

Kristian Walsh

Nanometres are nothing more than marketing-speak these days. Transistor density is a better metric, but even that has caveats when you look at actual parts.

“Intel 4” used to be called “7 nm”, but it offers densities better than TSMC’s N5: basically in the range 150~180 million transistors per square mm. First application for Intel 4 will be a CPU, which means the actual density will be at the low end of the process’s density range (density is the inverse of performance; SoCs have higher overall densities than CPUs because of all the DRAM; the transistor density of the CPU cores is lower to allow better heat dissipation during faster switching).

Intel was traditionally more conservative about feature-size - it typically quoted the sparsest configuration, which is the one that gives the highest performance, while TSMC typically quoted the densest configuration, which comes at a performance penalty. Again, this makes sense when you consider that TSMC makes mainly system-on-chip dies on its smallest node, while Intel tended to use its smallest nodes solely for microprocessors. In the days when the audience was engineers, who understood this distinction, there was no problem, but in today’s media landscape with tech bloggers and marketing droids who just want a Holy Number to fixate on, it was making Intel’s shortcomings look bigger than they were.

That’s not to say Intel isn’t behind. They are, but the next two years look interesting, because Intel 4 was the big step-change into EUV, and the pipeline that follows it is very rapid, much faster than TSMC’s. The next node down, Intel 3, is to start volume production mid-2024, jumping ahead of TSMC’s current N3 node, and only a year after the TSMC process (compare with Intel 4, which was three years behind TSMC’s N5). After that, it is Intel that looks like being first to crack “2 nm” (or “20 ångstrom” if you’re Intel...) in volume production, and it could be more than a year ahead of TSMC’s equivalent node. That will probably mean that TSMC’s N2 will be denser than Intel 20A when it arrives, though, because it’s generally true in fabrication that whoever has the newest process has the smallest node.

Why can't datacenter operators stop thinking about atomic power?

Kristian Walsh

Re: Nuclear is the best option if you look at the facts

“Nuclear” what, though?

Fusion? Definitely a hope for the future, but as of today, there’s exactly one fusion reactor on the planet that exceeds its energy input, and it only exists in a lab.

Fission of Thorium isotopes? Maybe, but Thorium reactors don’t exist outside of experimental facilities... and that’s after a research history that’s almost as long as Uranium’s.

So that leaves the existing technologies. Uranium and Plutonium fission. Well, a cursory glance at the half-life values of the most common waste isotopes produced by these provides a sobering rebuttal of the definition of “long” you just used there...

The alternative to stopping climate change is untested carbon capture tech

Kristian Walsh

Re: This is bonkers

At least your first sentence admits that you are in denial.

Science is based on the presumption that the accepted consensus is wrong; the scientific method relies on disproving, not proving, a hypothesis. The group of people who have tried the hardest to debunk the theory of man-made global warming are... climate scientists. They’ve been taking measurements of historical CO2 (yes, you can, from deep ice, and yes, it is accurate - like any system, the more you sample, the more accurate it becomes), examining and re-examining the models, and in all that time, the hypothesis that human activity in the last 10 years is changing the climate cannot be disproved.

Meanwhile, we have the other camp who have ready answers for every concern people might have about their way of life needing to change. Don’t worry, the data’s wrong, it’s not man-made, we can capture the CO2, it’s a conspiracy, the scientists are on the take... you name it, there’s an answer for it. That refusal to admit doubt, to say that there’s even the slimmest chance that the other side is right... that is the first warning sign that you’re being asked to believe things that are on the wrong side of measurable reality.

You raised the cui bono? argument elsewhere.. so let’s play that out. Who does benefit if the climate-change scientists are correct? Really, nobody - things are going to be pretty bad, to be honest. I don’t see a clear, organisable group that is small enough to be able to properly coordinate such a campaign of “misinformation”, yet well-resourced enough to push it through against the evidence of reality, who would actually become enriched or more powerful if humanity weans itself off fossil-fuel energy. The only candidates I’ve ever heard proposed for this role boil down to conspiracies about New World Order and other anti-Semitic trash...

Reversing the question, it’s very easy to see the beneficiary if the climate-change science is wrong: the oil and gas producers. The shift in fossil-fuel consumption from high-volume fuelstock to low-volume industrial use (yes, we will still need oil for plastics an pharmaceuticals) would destroy the regimes in charge of certain oil-producing states, who never diversified beyond getting the shit out of the ground, and who use their petroleum incomes as a way of exerting outsized pressure on world politics. Those same actors have a lot of money, are few in number, and have a history of lobbying.

The high number of conspiracy theorists who include climate-denial in their set of beliefs should ring an alarm bell or two. But then, conspiracy theories find their most fertile ground in minds that would rather believe in a malign omnipotence than accept banal, unpleasant truths about reality.

It looks like you’re a developer. Would you like help upgrading Windows 11?

Kristian Walsh

Yes, 10.6 was definitely peak Mac. It was the last OS release to be developed before iPhone overtook Mac as Apple’s biggest product line. (It was also the last release overseen by Bertrand Serlet, who had led OSX from its very beginnings). After 2011, the desktop OS was pretty much ignored for five years, then when it was given some attention, it got filled with lots of misfeatures intended to make your Mac into an iPhone accessory, plus lockdowns on the root user’s privileges.

That said, for me WSL2 is better than a Mac for development. I can’t abide any of the Linux graphical shells, and some of the hardware and software I use isn’t supported on Linux anyway, so my choices were only ever Windows or Mac. Actually, now I think of it, when I used a Mac, I mostly ssh’d into a Linux VM anyway, rather than use the native user-space (while I do believe that BSD is a better-written OS than Linux, my customers have only ever paid me to develop services that get hosted on Linux...)

Teardown reveals iPhone 15 to be series of questionable design decisions

Kristian Walsh

Re: It cuts Apple’s costs, so they do it.

The “by value” statistic is skewed by pharmaceutical imports, which have enormous value-to-volume metrics, but yes the point is taken that in general, light and expensive things tend to go by air. The difference is that Apple is in the rare situation of having a product with high value per kilogram that it also sells in huge volumes (230 million last year). For this reason, Apple also uses surface freight for iPhones - sending every phone by air would cost a fortune, especially when manufacturing capacity exceeds demand after the first couple of months on the market.

The product stays on sale for 12-24 months after launch, with reasonably high demand after the first 3 month peak, and is sold into a price-fixed market, so the depreciation argument does not apply either. This allows Apple to send shipments, by ship, leaving China at the same time as the initial air transports, to arrive later to fill in demand. When you’re moving 200+ million units of a product whose pricing is not time-sensitive, the low cost and high capacity of shipping containers is hard to overlook.

That volume means that if Apple had large manufacturing bases in the USA or Europe, rather than China, it wouldn’t need to rely so much on air-freight. China’s advantage for Apple has been that its very low wages and short supply-chain (especially for displays) more than overcome the higher transport costs of finished goods out of the country.

Never forget that Tim Cook’s path to CEO of Apple was via Logistics.

Kristian Walsh

It cuts Apple’s costs, so they do it.

Because of the company’s insistence on not pre-filling the retail channel before a launch, and its almost exclusive use of manufacturing in China, Apple ends up needing to air-freight a lot of iPhones in the days around a new model launch. Air freight is the most expensive way to deliver goods, and air-freighting millions of anything gets really expensive, really quickly.

A fully-loaded 747 cargo holds about 500 cubic metres of cargo, which is about 500,000 boxed iPhones (not exact numbers, but that's the magnitude involved). Most of the volume being carried is air, with the phone being the biggest contributor to the overall mass, so every gramme you shave off those phones’ weight equates to half a tonne of payload weight that doesn’t have to be paid for. (once you get to surface transport, volume is the what costs you money, not weight).

Around the time of the iPhone 5 launch, someone who was involved with this operation gave me the real numbers for the air-freight savings figures from moving to the relatively heavy 4S to the cheaper-feeling, but 28 grammes lighter design of the 5... I can’t remember the exact figure, but it was in the high millions of dollars, and that was just for the 90 days after launch.

Regarding Titanium itself, yes, it’s light, but it’s a horribly expensive metal to work with, and it scratches really easily. The normal approach to the scratching problem is to plate the metal surface, but the best-performing kinds of protective plating for titanium include nickel: a metal which can cause skin rashes in a significant portion of the population - not ideal for something that people hold all day... as Apple itself discovered in 2001 when customers complained that their brand-new titanium PowerBooks were making their palms red and itchy.

No joke: Cloudflare takes aim at Google Fonts with ROFL

Kristian Walsh

Re: An idea

HTTP is a content delivery system; HTML, on the other hand, is the input format to the very popular document typesetting system we call a web browser.

Kristian Walsh

Re: It’s about time !

10,000 is a tiny number of visitors for a website, certainly for any site that would be using Cloudflare as their provider. CF has customers who get millions of site-fetches a day. That’s a lot of quarter-pounder meals.

... and if those are US dollars, you should probably ask for a pay-rise.

Core blimey, Intel's answer to AMD and Ampere's cloudy chips has 288 of them

Kristian Walsh

The performance gap between Intel’s P-Cores and E-Cores isn’t as big as the difference in die-size and power requirements. In Alder Lake, the E-Cores provided about 50~80% of the performance of the P-Cores, depending on workload. On the die, the P-Cores are about 4x the size of the E.

E-Cores are perfectly capable of running any HTTP-borne service at a couple of thousand connections per second, so from a data-centre’s point of view, it’s far more desirable to have 250 of those in a rack than having P-Cores and dividing them using software... after all, there's not much use for the P-Cores’ fancy matrix maths or hyperthreading when all your applications are sitting inside single-threaded runtimes like Node.

In a datacentre, it makes perfect sense to specialise hardware like this. On a desktop, where the one system has to handle all workloads, then mixed-core designs are the way to go.

US Department of Justice claims Google bought its way to web search dominance

Kristian Walsh

I use Edge in preference to Chrome because Edge at least controls the CPU usage of Javascript tasks in background tabs, and yes, the constant insinuation from Google that not using Chrome is dangerous gets really old. Edge Chromium is objectively a better browser than Chrome, with the exact same HTML support and protections against malicious sits as Chrome, but Google keeps telling users of Edge that they’re unsafe if they continue using it. If it walks like a FUD and quacks like a FUD...

As for Bing... I’m disappointed: for a little while there, it was actually getting pretty good. Its big advantage over Google was that the SEO dirtbags pretty much ignore Bing’s algorithms when they try to game search rankings, which means that for general queries it tended to produce better results than Google. I gave up on Google when I realised that with the single exception of API documentation searches, most other queries produced a half-page of paid insertions and junk links, with the useful ones at the bottom of the first page.

But, we can always rely on Microsoft to hamstring anything half-way decent that it produces, and so they put that fucking AI bot into it. Now, Bing results are polluted with autogenerated nonsense, so I am once again trying DuckDuckGo and hoping it has improved its indexing...

HP reveals bonkers $5k foldable tablet/laptop/desktop

Kristian Walsh

Re: Laptop, schmaptop - I say flaptop

Seconded! Excellent name!

Lightning struck: Apple switches to USB-C for iPhone 15 lineup

Kristian Walsh

Re: Where do we go from here...?

... in the USA only. In Europe, Ford (like Tesla) will continue to use CCS2 for charging.

That adoption went both ways: Tesla has been forced to open the Supercharger network to all makes of vehicle, and to change its previous charging-port licence, which was as “open” only in the way a bear-trap is until you step on it.

In exchange, the US car manufacturers will now agree to use that Tesla connector type (called NACS - North American Charging System) for future production, and CCS1 stations will switch to using it too in time. with adaptors being made available to allow CCS1-equipped vehicles to access those and the existing Supercharger stations in the near term.

The benefit? You don’t get pushed into buying a Tesla anymore just because there’s no other charging network where you live. Conversely, if you want a Tesla, but there were few Supercharger locations near you, then great news: the other charging sites near you will soon become compatible with that new Tesla.

This is a good result for customers.

Kristian Walsh

Re: Where do we go from here...?

Yes, but nobody with eyes or a brain would have believed that about the British Empire, while there is ample evidence of the EU walking the walk on enforcing market competition, often against the will of very large domestic corporations.

Anyone who’s interested in what the “free market” of the British Empire looked like could ask themselves why the Indian independence movement adopted a spinning wheel as its emblem, and go from there. But, in summary: the UK farmed its Empire: for food, for raw materials, for troops, and for customers for British-made goods. Any doubts about the “benefits” to members of being in the British Empire can be dispelled by noting how quickly everyone left once Britain’s military strength was depleted in the aftermath of WW2 to a level that made reprisal against rebellion unlikely.

Kristian Walsh

Re: Where do we go from here...?

The EU believes in one thing above all: a free market with competition. That’s not the same kind of “free market” they have in the USA, where companies are free to build monopolies, lock in customers and seek rent from them forever; it’s the actual “free market”, where new entrants can get a foothold by offering a better product, and customers can leave a product they no longer like using.

“Innovation” is a bullshit-word thrown around by tech companies when what they actually want is customer lock-in: you can innovate and remain compatible with standards (even Apple did it - the original MacOS X is a great example). In a truly free market, innovative products still reward their producer, but the difference is that the reward is temporary, and when competitors match or exceed that innovation that reward transfers to them, and everyone benefits. In a market with actual competition, there is more innovation, because you can’t sit on your ass for almost a decade doing nothing much except raising your prices, safe in the knowledge that you’ve locked in your customers with bundled services and sunk costs on incompatible anciliaries.

But seriously, who in this day and age thinks it’s not stupid to own an iPhone, a Mac and an iPad, and yet need both a USB-C and a Lightning cable. Lightning was “innovative” right until USB-C came along. Retaining it long after a better solution existed (and one which Apple used on other product lines) was just another example of Apple trying to bar the customer from leaving its walled garden.

Ford, BMW, Honda to steer bidirectional EV charging standard

Kristian Walsh

Re: "because if you unplug your car, your house goes dark"

Maybe, but spending £2000 for a 5kWh battery and high power inverter looks expensive when your car came with at least 40 kWh of battery, you’re already paying for it, and like most cars it spends over half its working life parked at your house...

Kristian Walsh

Re: Not sure I'd do this

They’re not taking it, they’re borrowing it, and they’re not taking it all. The amounts are in the order of 5~10 kWh of battery (how much is your choice). The power of V2G is in signing up hundreds of thousands of cars. 5 kWh x 100,000 cars = half a gigawatt-hour of on-demand power supply. Whatever they pay you for it, it’s cheaper than building a power-station.

You’re not really consuming charge cycles either because in use the rate of discharge is pretty slow, certainly when compared to driving; and you’re not only benefiting someone else: you’re being paid for it.

But if you want an appeal to greed, the technology behind V2G also allows you to use your car battery as a short-term energy store without re-exporting to the grid - on days you’re not travelling, you can charge at off-peak, discharge through the day and you will pay the absolute lowest price for electricity (maybe even zero if you have a lot of solar panels). But don’t think you’re getting one over on the utilities... this is exactly what they want you to do.

Kristian Walsh

Re: Voila!

If you stop reading before the end of a sentence, it will lead to you making silly comments.

“Solar + storage”. The last word was important.

Microsoft to kill off third-party printer drivers in Windows

Kristian Walsh

Not “acquire” it: the very nature of FOSS makes that impossible. They did hire its lead developer for nearly two decades, and they put a lot of development work into it in the early 2000s, but that’s all freely available thanks to the licence.

Actually, in that brief near-decade between the creation of MacOS X and iPhone becoming a hit, Apple was one of the better corporate citizens when it came to FOSS. WebKit is another project to which Apple contributed a lot, and for their internally-developed Bonjour/Zeroconf mDNS-based service discovery system they went through the IETF recommendation process (eventually RFC 6762, 6763) rather than keeping it proprietary. The XNU kernel and Darwin user-space are still open source, but now these days Apple keeps the drivers for its own hardware segregated from that..

Times changed once Apple stopped being a computer vendor, and the Jobs instinct to control everything took over.

Kristian Walsh

“The print system is directly descended from that of early Macs.”

No, not really - it just looked like it. The API and user interface for starting a print job may have been carried over, but this article is about drivers, and on that front there was a total break in 2001. For Mac OS X, Apple adopted the open-source CUPS project as the sole driver framework for printing.

Pre-OSX, print drivers had to map QuickDraw ‘PICT’ format primitives to their own output instructions, which required knowledge of QuickDraw (a rare thing in the 1990s) as well as the target printer - Mac printer drivers used to be pretty flaky for non-PostScript printers as a result. The people behind OSX’s print model made two very good decisions: first the OS adopted PDF as the native image metafile format of OSX. Second, they chose CUPS as a driver framework. These two decisions greatly simplified the driver-writer’s job, especially on printers that had PostScript support already as PDF is relatively simple to convert to and from PostScript.

But, because CUPS is open-source, any printer with a driver for macOS also automatically has a driver for Linux, which has greatly improved Linux printing support. However, the bigger problem with printing on Linux is the general dog’s dinner around graphics APIs used by applications to originate documents, with many try to everything themselves, so sometimes you end up with a classic Linux pipe-bludgeon of rasterising a postscript file, embedding it in a PDF container, and sending it to CUPS. Yes I know that’s stupid, but... “I got it to work”.

Elon Musk has beef with Bill Gates because he shorted Tesla stock, says biographer

Kristian Walsh

Short-selling = negative feedback = stabilisation = good

If you look at a stock market as a control system for finding the correct price of one share of a company’s stock, then short-selling is a fundamental requirment. Long options (the kind Musk likes) are a way of buyers indicating that the current price is too low. Short options are the opposite: by setting a future sale price below the current value, option-takers are indicating that the current price is too high.

Between them, these two signals act as the error in a closed-loop control system, and the price, theoretically, settles on a reflection of the demand for the stock. Without a negative feedback signal, the price would rise uncontrollably until it crashed. (See the Dutch Tulip Bubble for an example of a market that had no shorting).

Or that’s the theory, anyway. The problem with the stock market as a control system for finding a correct price is that the system makes it easier to send the long signal than the short one. To short a stock, you need the strike price, plus a lot of money to cover your potential losses; to go long, all you need is the price of the shares you’ve bought. This biases the feedback towards increasing prices, and it’s not really an error, but rather a designed-in part of the system.

Musk, like nearly all VCs (Musk’s role in both Tesla and SpaceX was as an early-stage investor), exploits this bias by hyping stocks, thus transferring the risk of failure to small investors who will buy and hold the shares post-IPO. The reason he’s so against short sellers is that a short-selling run is the only thing that can really correct the price of an overvalued stock, and like every other VC, he needs overvalued stock to keep other ventures afloat.

Bombshell biography: Fearing nuclear war, Musk blocked Starlink to stymie Ukraine attack on Russia

Kristian Walsh

Venn diagram would be nice...

I’d love to see the intersection of the group of people who shout loudly about “net neutrality” and the people who are now resolutely defending Elon Musk for clearly breaking that principle.. I know it’s not going to be small.

To clarify: “see” refers to the chart, not the actual people ... even one Musk fanboy is nauseating; can you imagine a mass of them? yech.

Kristian Walsh

“Help us Elon, or the Ukranians might win actually beat us and reclaim their country”

“Shit - we can’t have that, Vee-vee. Gimme a minute to open Slack here... [HOLD MUSIC].. Hey, yeah, I got the guys to make a fix, so now you can put anything you want in Crimea”

“Splendid. I owe you one, I owe you one, big guy”

“Any time for my Uncle Volodya...”

Mozilla calls cars from 25 automakers 'data privacy nightmares on wheels'

Kristian Walsh

Dashboard is not the “instrument binnacle”

I didn’t downvote you because Tesla’s manufacturing quality is so shit that there is a remote possibility that the dashboard in your has subsequently become detached and lost, but no, you’re not talking about a dashboard: the dashboard is the horizontal surface running from one side of the cabin to the other that faces the driver. You mean the “instrument binnacle”, which is a housing, mounted on the dashboard, in which displays of important information for the car’s driver are presented. Tesla is too cheap to put a second screen on the dashboard (or, more expensively, to have to design two dashboard mouldings: one for LHD, one for RHD), hence that stupid tablet.

The name, if anyone cares, is carried over from the dash-board of a horse coach - this is the plank of wood that was mounted in front of the driver, and its purpose was to prevent pebbles and other road dirt hitting the driver. (the meaning of “dash” here is “to break or smash”; the same one as in the phrase “to dash one’s hopes”).

Kristian Walsh

Re: re: Tesla warns customers:

... which gives him plenty of time with that blowtorch and lump-hammer.

That “basic tools” thing isn’t unique to Land Rovers: Toyota J40s can also be maintained with basic tools, but the difference is that most people who owned J40s never had occasion to discover this fact.

Kristian Walsh

Re: Obsolete

Er, what? Why would anyone need to do this when licence-plates and the database behind them already exists. This is how C-charging and ULEZ fines are implemented right now (or Italian ZTLs for that matter): a camera reads your plate, the backend looks up registration number in a database, and If you're exempt, happy days; if you're not, you'll be told.

And the “controversy” about the ULEZ scheme in London is a confection of the right-wing press in the UK, desperately looking for some culture-war distraction to stop their readers noticing that the current UK government has done an absolutely shit job of running the country over the last decade. ULEZ expansion is a rare example (in UK politics) of a government studying evidence and using it to guide policy, rather than just being swayed by whoever is shouting the loudest (or, in the case of the Tories, paying the rent for the villa).

Kristian Walsh

Re: A long time ago...

"A major US car manufacturer observed that it had turned from a car maker to a finance company with a sideline in cars..."

It was General Motors. In 2008. How did that go for them afterwards, I wonder...?

Microsoft tells partners unbundling Teams is a 'compromise' with the EU

Kristian Walsh

Re: Difference w/o distinction

Sorry, but my experience is that Google Meet is a ball of shit, and any medium sized call will involve at least one participant having to reboot and/or change browsers or machine.

Challenging for shitness is Slack with it’s two minute ritual of “can you hear me? I can hear you. Hold on, now can you hear me... hang on, I’ll quit Slack” once a day for every contact.

Zoom does one thing, and does it pretty well. But in general, I can’t help thinking that outside of social use, video group calls are a bigger time-waste than in-person meetings: nobody is paying attention, but they still have to be present enough that they can’t to productive work. At least when everyone’s in the room, there are lots of non-verbal signals that it’s time to do something useful, but on a video call, the most senior waffler can string it out for ages.

Grant Shapps named UK defense supremo in latest 'tech-savvy' Tory tale

Kristian Walsh

Re: Question

A shallow talent pool combined with a requirement for loyalty over ability.

The UK electoral map (like that in most two-party, first-past-the-post systems) harbours a lot of “safe seats”, where anything can get elected to Parliament so long as it belongs to the appropriate party. This results in those seats being given as rewards for favours rendered to persons who have no political ability whatsoever. That creates a pool of MPs who lack ability, but are loyal to whatever leader or group managed to get them selected for their seat.

(Historically, there are more of these safe seats in the Conservative column than Labour, but this run of Conservative governments does seem to be doing its best to rebalance that situation...)

In a more competitive (i.e., democratic) electoral system, candidates would need to actually convince the electors of their worthiness, which would select in favour of more politically capable people and against ideologues, party hacks and spoofers. Consider what would happen in a vote-transfer system where there was not one, but two candidates elected: in a safe Tory seat, the Tories would put up two candidates, but as there would not be sufficient support to return both, the electorate would ultimately choose which one gets in. This situation neatly explains why the party opposes PR: if you can’t get a chum into Parliament, then what’s the point of it all?

Kristian Walsh

Re: Xitter...

I read all of that in the aforementioned Catalan pronunciation of X (so, /'ʃɪtər/ and /'ʃɪt/), and it still works, with “squeezed out a ~” actually gaining something in the process, I feel.

Tesla's purported hands-free 'Elon mode' raises regulator's blood pressure

Kristian Walsh

Re: Exclusive club

It sounds like it is you who has pre-judged Musk, and are sticking with that rosy view in defiance of the evidence. I don’t form fixed opinions of people - I judge people mostly by their actions. So, when Elon Musk starts acting like someone who understands why there are laws, and stops ignoring them purely from exaggerated self-belief, then, and only then, I will revise my own low opinion of him.

Kristian Walsh

Re: I for one can't wait till this works

“They” can only refer to the subset of young men who ride motorbikes that have done their best to convince the general population and governments that motorcycles are lethal implements.

I know “not all bikers” and not by a long way, but you will know exactly who I’m talking about - the kid who has a bit of cash, turns 24 and gets his A-licence and immediately buys something that you know he lacks the experience to ride safely... and a year or two later you hear he’s had a massive spill. Bikes have become really cheap relative to income (and are not taxed in the way cars are), plus because they last much longer now, and high-performance bikes are especially cheap second hand, it’s cheaper to get on something dangerously fast - without something to rein in the impulses of the young male, there would be a bloodbath. This wasn’t a problem before the 1990s because typical bikes were heavier and had less power - now everything is lighter and faster, but the skill of novice riders hasn’t increased.

For cities, the two-wheeled transport that’s needed is bicycles, and it’s relatively easy to install space for them to keep cars away from bicycles, and the bicycles away from pedestrians. Electric-assisted pedal-bikes make cycle-commuting a practical proposition (generally, however far you can ride comfortably on pedal-power alone, you can nearly triple that for an e-bike).

Kristian Walsh

Re: I for one can't wait till this works

Sorry to break it to you, but private autonomous vehicles will create traffic jams, and will most likely be banned or heavily taxed in urbanised areas.

Like a lot of people, you’ve been gently guided toward thinking that in this future scenario, only you will own an autonomous car, and so when you drive to your destination, you don’t ever think about being stuck in traffic with all the autonomous cars that have been sent back to their homes.

Right now, vehicle occupancy in the USA is about 1.1 persons per car, which is why traffic is so horrible in US cities without good mass transit (i.e., nearly all of them). This ratio has slowly reduced over the decades, and you can add to this that cars are also getting bigger and longer, which means fewer people carried per foot of road-lane. Basically, the efficiency of the road network in suburban areas is declining. To solve the traffic problems in car-dependent cities in the short to medium term, traffic planners have been trying carrot-and-stick programs to get that occupancy figure up to 1.25 per vehicle, because the alternative is billion-dollar road programs that will only forestall the problem for 5 years at a time. (One metropolitan planner told me once that if the law allowed it, he would mail checks to people to get them to car-pool, because it was a better use of highway funding than actually building new roads).

But the important thing is that in high-traffic cities, occupancy is a nice (inverse) proxy for congestion: the lower that average vehicle occupancy figure gets, the closer a city gets to traffic hell. Now, imagine the effect of a whole fleet of 0.0 person vehicles would have on that average... It would only take 12% of drivers to have one for the traffic planning models that underpin your city’s road transport network to enter the “this can never happen, so don’t worry about how shit it gets” zone where occupancy rates fall below 1.0.

Kristian Walsh

Re: Much ado about Lidar

“the company producing LIDAR equipment”, as if there’s only one manufacturer...

LIDAR isn’t magic, but the last two letters, AR (And Ranging) explain why it’s really useful when you want to make sure your computer-guided 2 tonne car doesn't drive right into a solid wall (or the side of an 18-wheeler truck). But you don’t even need LIDAR - a much cheaper radar sensor of the type used in Level 1 adaptive cruise control would work too. Musk was too cheap to install either.

Tesla execs would do well to read up on the history of the US auto industry, because without some serious course correction, their company is barrelling towards its own Pinto scandal.

Uncle Sam accuses SpaceX of not considering asylees and refugees for employment

Kristian Walsh

Re: Irony..

sigh... that one was too subtle for some, obviously. For avoidance of doubt, the word “refugees” is not to be taken literally in that sentence, as the description “fleeing prosecution” should have made clear.

As I’m spelling stuff out: the irony is in comparing the situation where today’s US space industry are (allegedly) assuming that refugees applying for jobs are no more trustworthy than criminals, when that very industry was founded by genuine criminals who were spirited out of Germany before the process of identifying and trying prominent Nazi’s got going, and who were explained away to the public as mere “refugees” from post-war Germany.

Kristian Walsh


The entire US Space industry was created by refugees, many of whom were fleeing prosecution in their home country at the time...

(That isn't a typo.)

Bad software destroyed my doctor's memory

Kristian Walsh

Re: Have a hundred upvotes

100% agree. The golden rule for line-of-business apps is that you NEVER assume your user knows nothing, but consumer app design is based on the idea of a user who needs to be coaxed into choosing to use your app. This use of consumer design patterns in professional applications software is the biggest design flaw I’ve seen in this kind of software, and where it comes from highlights the root of the problem (hint: who does the development team liaise with? The users, or the management layer that is paying for the software?). These aren’t consumer products; the user of these system know a lot about their domain. They know a lot more about it than some database designer who is thinking only of how data elements relate to each other, or the “UX architect” who will inevitably try to impose “surprise and delight” and onboarding patterns onto something that its users already need to use.

Users invariably know exactly what they want, but they’re just not great at articulating it. (e.g., Try to explain to someone what you need in a text editor.. I guarantee you’ll leave out something basic like search/replace). The existing system they have is the best requirement specification for any digitisation project: that works. For them, it’s accessible, and it is adaptable, and it is the accumulation of years of small process improvements. That system could be a collection of spreadsheets, a document pile or a paper file, but it is the most important thing to study, and not just its structure, but how it is used. It is your job as a development team to preserve that knowledge, not throw it all away in in favour of the “new flashy”. Your users don’t need onboarding or guidance on how to do their job, they know how to use the tool already; they don’t even need pretty visuals - for desktop/laptop systems, information density is often more important than something looking aesthetically pleasing.

As an example, my GP uses the ugliest piece of software I’ve ever seen, but she swears by it.. they’ve it’s exactly like a paper case-file or an email client. Open my page, and you get a summary of my details, a list of active prescriptions, then the rest of the screen is basically like an inbox of all the notes and documents about me, in chronological order. It isn’t pretty, but it keeps the information organised in a way that makes sense to a doctor running a general practice, and reduces the amount of clicking and hunting needed to find things. The first few years I visited the practice, I kept thinking “that software is shit, surely I could design something better”, but after a while seeing how well it suited the workflow, I realised that I was confusing aesthetics with usability. An important lesson learned. (And my doctor’s been pretty good too).

Pope goes fire and brimstone on the dangers of AI

Kristian Walsh

As a matter of historical fact, the Catholic Church was very much in favour of the moveable-type printing press. Rome was the first city outside of Germany to house a printing press (just 13 years after Gutenberg’s), and the leaders of the church was shrewd enough to see that this would make access to the printed word far cheaper and easier, so they made sure that their message was out there first.

Certain individual Catholics, specifically the scribes who worked out of monasteries across Europe and earned quite a nice income from copying manuscripts, may have had a different view of Gutenberg’s new machine, but at the top of the church it was not seen as a threat, but rather an opportunity.

Printing certainly facilitated the spread of the Protestant Reformation, but that was a about fifty years later, by which time printing was just a part of the modern world, and seen as an instrument that was not inherently good or evil.

India launches contest to build homegrown web browser

Kristian Walsh

Re: Nationalism for the sake of Nationalism?

A purely technical argument:

India’s native languages contextually-rearranging scripts in the Brahamic script family (Brahamic scripts are not confined to India: languages of Sri Lanka, Bangladesh, Thailand, Tibet Laos also use scripts belonging to this family). These are the most complex writing systems in use today, and they present their own specific challenges regarding text editing and display on computers. It is conceivable that changes in UI to make these writing systems more natural to use with the browser could result in a slight degradation in UI for, say, Latin scripts. No global FOSS project would accept such a change, given the prevalence of Latin (and Latin-like) writing in the world, but one specifically targeting India would lean more towards equality of the two dominant writing systems in the subcontinent.

That said, this is far more likely to be another one of Modi’s Hindu-supremacist impulses.

ChatGPT's odds of getting code questions correct are worse than a coin flip

Kristian Walsh

Oppositional Networks train for plausibility, not accuracy.

Okay, this shouldn’t be a surprise at all if you dig in to how GPT and similar are trained.

First, you get an AI that can categorise files (e.g., “this is an image of two kittens playing scrabble”, “this is C source code for a quicksort”, etc.). Then you get another AI that tries to generate files. When the second one can produce documents that the first accepts as matching the description, you are ready to release.

Now, I’m sure you can already see the problem here. The first AI has no knowledge of whether: a. the original descriptions were correct (the kittens may actually have been playing ludo, the C code could have had a bug), or b. why those descriptions were correct. All it did was build an enormous, opaque network that gave the correct description for its inputs. And the second one had no knowledge either, it just kept making shit up until the first AI accepted it.

Doesn’t that sound like that guy* you hired who could do a great interview, but didn’t know anything about anything once you put them to work?

Basically, we have created machines that can bullshit better than a human. I will leave it to you to decide what level of concern that should attract...


* yes, it’s almost always a guy. My interviewing experience taught me that female candidates are almost always honest in interviews, males are about 50/50 between truthful and braggarts.

Arc: A radical fresh take on the web browser

Kristian Walsh

(small correction)

... OSX was designed in Macromedia Shockwave, not Flash.

Kristian Walsh

Re: Edge updated itself this morning...

Exactly the same, but some people are so fixated on Microsoft that have an enormous blind-spot for Google’s anticompetitive behaviour.

If you don’t use Chrome, every Google service periodically tells you to install it, claiming that the thing will run better in that browser (no, it won’t, because Chrome still can’t properly throttle resource use by background tabs).

And I also remember that when I sign in to Google web service (sheets, docs, mail, etc.) on a new machine using a non-Google browser, the “was this you?” email pushes Chrome at me too (sign-ins using Chrome use different wording).

But never mind that, Micro$$$oft hates Linux, so they must be evil.. yeah, they hate Linux so much that they give you an installer for it as part of Windows these days, in the form of WSL, and put other distros in the Windows App Store if you don’t fancy the default Ubuntu.

Kristian Walsh

Actually, macOS was never designed around the idea of using a single app in a single window. From the very beginning of Macintosh Toolbox, Mac applications are independent of their windows: originally, windows were there to contain only documents, not toolbars or controls (those would go in their own “tool” panels: originally fixed, but later evolved into floating tool windows), and an application could theoretically be designed that never had any windows open, with only the menu-bar as its control mechanism. The fact that closing a Mac app’s windows does not close the app still catches out users of other systems.

Apple’s decision to pull the menu bar out of the window is based on sound ergonomic principles (for more detail, look up “Fitts’s Law”), and keeping the menu outside of the window frame also saved screen real-estate on a system whose original display was just 512 x 342 pixels.

Because the user interface was designed around having multiple document windows per application, the control you called “minimise” did not exist. Instead, you had a control that told the application to size the window to its optimal dimensions for the contents. Press once and the window becomes wide/tall enough for its content, or full-size if that content is bigger than the window; press again, and it returns to its previous size. Later Mac System versions (7.6+) added a true minimise by double-clicking the title-bar (this was originally a shareware product called “WindowShade”, which Apple either bought in or shamelessly ripped off - they had a habit of doing both in those days).

Where things went to shit was MacOS X, which was designed after Apple had fired its Human Interface Design Group. OSX was designed in Photoshop and Macromedia Flash, by graphic designers and animators, with one goal: look cool. That meant a lot of the subtle interface features from the Classic MacOS that were based on ergonomics and cognitive science were dumped. The “Minimise” button is a particularly obnoxious piece of UI because it is annoying to undo if accidentally pressed (you can't just click the same place again, as the click-target has moved; but worse, it causes the whole visual context to disappear, so you now have to visually search for where it’s hiding, and unlike Windows’s taskbar, the OSX Dock is centrally aligned and thus grows and shrinks, dicking around with your short-term spatial memory in the process), so Apple’s decision to put that right next to “enlarge” is a little inexplicable...

Actually it’s not inexplicable once you stop thinking about how it would be used, and focus instead just on how it looks... they wanted a “traffic-light” pattern, and with Red meaning “close” (this is actually a good reinforcement), Green meaning “big” (getting a little tenuous, but it’s the opposite of closing, I guess), then “minimise” had to be amber and also had to go in the middle to abide by the visual metaphor. Makes sense visually, but it’s the wrong place: one of many examples in OSX of a fetish for visual patterns overruling functional needs. A favourite example of mine: early OSX builds had the Apple-logo smack-bang in the middle of the menu-bar.. “to match the casework”... This, thankfully, did not last more than a month or so.

This isn’t a rant against graphic designers - I am one myself from time to time - it’s a rant against the belief that “graphic design” (or its modern cousin “experience design”) and “interface design” have the same goals and are interchangeable skills. They’re not, and sometimes a good interface needs to be not-quite-aesthetically pleasing in order to be learnable and usable.

Tesla to license Full Self-Driving stack to other automakers, says Musk

Kristian Walsh

Re: Behind, always behind

Long loading and unloading times mean that rail freight is efficient if your origin and destination are far away from each other and you have no intermediate load/unload points. That’s a good description of freight between the US’s major population centres, but it’s rarely true in Western Europe or South-Eastern Asia with its dense network of short-distance freight routes between industries.

Generally, as the number of “towns” per thousand square km increases, freight shifts from rail to road as rail’s lack of reach and flexibility overcomes its low per-km cost. Conversely, as the population density of those settlements increases, rail becomes the most efficient mode of personal transportation.

There are a few hybrid systems to allow road freight to hop onto rail for long-distance or vehicle-restricted sections of their route: the Brenner RoLa is one example, where trucks drive on to rail flat-cars and are then carried through the Brenner Base Tunnel railway line under the Alps while the drivers take a little nap: on the other side, the process is reversed, and the truck drivers set off on their various ways again. And, more familiar to El Reg readers, there’s the Channel Tunnel, a rail service that also operates a roll-on, roll-off crossings for freight drivers.

There are similar proposals for transcontinental routes in the USA (here the main win would be to still be able to move freight while the drivers are on their mandated 14 hour daily breaks), but as ever in American infrastructure, politics intervenes and things get gummed up.

Microsoft kicks Calibri to the curb for Aptos as default font

Kristian Walsh

Yep. This one is Apple’s fault. The original Macintosh first exposed the general public to the idea of typefaces, and its early applications had a “Font” menu for faces, with a “Style” menu beside it for style and sizes. This made people think that the word for a typeface was “font”.

“Font” itself was an Americanised spelling of the British “fount”, with the same pronunciation, which ultimately came from the German “Fond” in its meaning of “a stock or supply of something” - in this case, the typesetter’s stock of letterforms of a given style at a given size.

Kristian Walsh

Re: I like serifs

That “Motorway font” is officially called Transport and it’s used on all UK road signage. While there is actually another face called “Motorway”, it only defines the narrower digits 0-9, M,N,S,W,E and parentheses as used on motorway route numbers.

One really nice, but subtle, feature of UK signage is that there are two different typeface weights used: Transport Heavy is used on white-backed signage, and Transport Medium is used on dark-backed signage (such as A-roads and Motorways). The reason for using two weights is to compensate for a visual phenomenon known as halation, where very bright areas appear bleed into surrounding dark areas, so on a reflective white signboard, the background appears to glow, making the black type seem thinner; on a reflective dark-background sign, it is the type that glows, making it seem thicker.

One year after Roe v Wade overturned and 'uterus surveillance' looks grim

Kristian Walsh

Re: Negative data

Had this exact argument before about the need to hide a user’s home location from our GPS logs. Enforcing a (e.g.) 100 m radius around their home would require that we made the system learn where they resided: something we had no business reason to know, and the exact thing we were trying to avoid, and then there would be an obvious missing circle from any plot of their location data.

In the end, we used an opt-in algorithm: we had areas that we had a legitimate business reason to know they were, and which they had consented to, so we only recorded points within those regions. Sadly, that meant we were requesting GPS outside those areas (guess what information your phone OS asks for as soon as you set a geo-fence...), in order to decide if we were going to retain that information, but Apple’s “here’s where [this app] tracked you” map doesn’t know that, and we still got complaints from customers. But, contrary to the received wisdom within tech, most people are not idiots, and the majority of our customers were able to grasp the idea that you can’t know whether you should report a location without first knowing the location.

Kristian Walsh

As it happens, the right of privacy does protect people from being charged with drugs offences simply because they are under treatment for the effects of that consumption: The cops can’t just call in to the local hospital emergency department and demand a list of everyone who’s been given naloxone that week.

Kristian Walsh

That murder analogy is claptrap, and I suspect you know it.

First, a murder has nothing to do with the privacy of medical information, which was the grounds for a US constitutional prohibition on banning abortion. Second, any right to privacy of the murderer is pretty obviously trumped by the pre-existing right to life of the person being murdered. (This right-to-life argument, incidentally, is the one with which anti-abortion campaigners have been most successful outside of the US: if you can secure a legal ruling that an embryo has equal rights to a born child or adult, then suddenly you can argue that early-term abortion is homicide. This is fine example of making an argument from a faulty premise, but that’s kind of par for the course from a lobby that has a large-crossover with young-earth creationism.)

As it happens, I live in a country with legalised abortion, and judging by your username, I suspect you do too, or at least did (few Americans would understand the reference in your username). And yes, I did vote for it, directly. But unlike Americans, I live in an actual democracy. And it’s one with an electoral system that hammers extremists, and while we have a written constitution, it is one that is routinely amended by public vote. Yes, that means we have put some dumb things in our basic law over the years, but we also get to emphatically remove dumb things from our basic law too.

I agree with you that the US places too much emphasis on its constitution, especially now that the practice of amending that constitution to respond to changing events appears to have died off. The last amendment was in 1971 - over half a century ago (and all it did was reduce the voting age to 18). In the 20th century, the US Constitution was amended 11 times (or 9 if you discard Prohibition and its repeal). What used to be a living document of basic law is at risk of being ossified into a snapshot of early-20th century beliefs.

But there is nothing in the US constitution that actually prohibits abortion - it’s just that the Constitution is one of the few legal instruments that the conservative Right are wary of interfering with. Many US states have legal abortion simply because they passed laws that allow it (this is the same as the UK). However, in many other states, religious conservatives, imposing the morality of their particular sect upon the entire population, retained or strengthened laws that banned the practice. In the US legal system, the only way to counter repressive laws at the State level is to show that they are incompatible with the Constitution. This used to work, but Roe v. Wade was both a victory and the start of a defeat for the progressive faction in the US. But Roe v. Wade didn’t assert the right to bodily autonomy explicitly, all it did was assert a very specific right, but one that was incompatible with any attempt to criminalise abortion. I agree that this is an arse-backwards way of doing things, but if you spend any time in the US (and outside of the cities), you’ll understand that this is a country that even today is deeply religious, and very culturally conservative, and we’re talking about 50 years before today’s more liberal outlook: after all, much more “liberal” countries in Europe, taking a direct human rights argument that women, as human beings, have the right to full bodily autonomy took longer to legalise abortion.

I don’t disagree for a moment that “privacy” is a weak hook to hang the right to abortion upon; but it was the only hook that could be used in 1971, given the absence of an explicit right to bodily autonomy in the US Constitution - the US Bill of Rights does not have that concept, partly due to it being written in 1791, and partly due to the existence of chattel slavery across the USA at this time. That need to use a roundabout way of asserting a human right is why the current mess exists. (The 1948 Universal Declaration of Human Rights was ratified by the USA, but as a declaration, it has no legal force)

So that was the victory, but the beginning of defeat was that Roe v Wade spooked the Religious Right so badly that they formulated a long-term plan to load the US Supreme Court with the most right-wing pro-religious candidates they could find over the coming years: this way they could neutralise the parts of the Constitution that don’t agree with their ideology without having to open the Pandora’s Box of actually amending it, and it paid off last year, when a judgement came down to say that no, there’s no right to medical privacy, so no, there’s nothing to stop States passing laws that make abortion a crime.

Oddly, you seem to be labouring under the impression that the USA is a functioning, representative democracy. I would disagree, given how many barriers there are to ensuring that the choices of the electorate are reflected in their government. State-level gerrymandering (sorry “redistricting”), voter suppression (sorry, “anti-fraud measures”), even the Electoral College mechanism are all designed to prevent the actual voice of the electorate being heard. At least the Electoral College may have made sense in the days of pony post, but it has no purpose now; the other two are modern tactics to lock-in supermajority results, often from a minority voting base.

Kristian Walsh

Privacy is relevant, because up to a certain point, a woman’s pregnancy is private information, known to her alone. Also subject to privacy is any patient’s interaction with their medical carers. This is the basis under which Roe v. Wade was won.

Making abortion a criminal offence would mean that, in order to prosecute, there must be a report of the “crime” occurring and evidence of this. In the case of early-term abortions, the smallest number of people who would typically know that the act has occurred is two: the woman, and her physician. The woman would not disclose the information, and the physician is bound by doctor-patient confidentiality. So, the matter is private, and the record of the act having occurred is also private. A citizen’s right to privacy already prevents the state discovering if, for example, they had had treatment for severe mental illness, sexually-transmitted disease, alcohol or drug addiction, etc. (you cannot charge someone with use of illegal drugs by compelling their doctor to reveal that they were treated for heroin addiction, for example). Roe v. Wade admitted that the fact of an abortion is also protected, and thus the state cannot compel or induce records of abortions.

Consider also that the vast majority of abortions are early-term and pharmaceutically-induced, and are indistinguishable from a miscarriage. So to prosecute someone for having an abortion rather than victimising someone who has just lost a pregnancy, it would be necessary to discover proof of supply of the necessary tablets, or proof or having had the procedure clinically; and the right of privacy asserted by Roe v. Wade prevents a state from doing this. You cannot make something a crime unless you can prove it without violating the rights of the parties involved, and so abortion becomes un-prohibitable. Not “legal”, just “impossible to make illegal”, and so the States could only regulate it, not prohibit it.

But the overturning of Roe v. Wade has other implications for citizens’ rights, because Roe v. Wade is based on the assertion that the US Constitution actually confers a right to privacy; the judgement that overturned Roe v. Wade says that no, the US Constitution does not always grant citizens a right to privacy, and the States can basically decide what is and is not private. But don’t worry: I’m sure it’s only guilty people who will have their private medical information seized by overreaching government prosecutors...

Techie wasn't being paid, until he taught HR a lesson

Kristian Walsh

Re: Unique keys

I had a similar problem in a US multinational with a dev shop in India, and little interest in learning about other countries’ conventions (Ironically, I was employed in the then-new Internationalisation department to try fix those attitudes).

The problem was exactly what you said: you’d get an email summary of an issue from a Bala or Giri, and because the US IT dept had just dumped people’s informal names into the Exchange directory, you had no clue which user that actually was when you went to the issue tracker, which (being run by the Indian operation) used official names for users.

Eventually, you figured out who was who by domain responsibility, but the whole thing would have been easier if the “also known as” fields had been filled in the US email directory. Apparently it was a long-running battle from the Indian side to get this done, as it had led to a couple of accidental mis-emailings of sensitive information, but as these problems didn't happen in Santa Clara, they were obviously not important..