Re: Pedant Alert Incoming...
"Who, Me?" is always on Mondays; it's the blurb that's wrong, not the publication date.
Then again, it was Monday morning. Give him a break...
685 publicly visible posts • joined 1 Aug 2020
$14.8 billion in share buybacks since the beginning of 2025. Then $15 billion in new share issue (now increased to $20b).
They'd have had most of the money already if they'd, you know, kept it in the business instead of spending it to try and juice their stock price.
So effectively a payout to existing shareholders (above and beyond the traditional dividends), then sell new shares to subsidize it. Another thing to add to their "not exactly confidence-inspiring" scorecard of late.
"$80 million a year over 50 years"
Its annual budget now is an order of magnitude less. The best I've been able to determine is about $5 million/yr (from two unreliable sources, but they agree with each other).
Not disputing your average, just pointing out that the ongoing costs of the program are a lot lower.
"I'm sure these sort of arguments have been used throughout history every time technology has become common place e.g. *pocket calculators*, computers, or even the internet." (emphasis added)
See, for example, Isaac Asimov's short story The Feeling of Power. [1]
Socrates criticised writing itself on that basis. (That page's final paragraph contains the actual criticism.). [2]
The Luddites had a different objection to the tech of their day. They didn't object to the new machines per se, but to what they meant for employment, pay, quality of output, etc. Sounds familiar, no?
[1] I love that in 1958, a plausible science-fiction "pocket computer" had to have a mechanical display whose "numbers whirl to a halt".
[2] The thing is, because Socrates didn't write any texts, all we know of his thought is what others wrote about him.
Headlines have, as I see it, two primary requirements: clarity and brevity. Alas, these are often in direct competition.
I suppose that conflict has bedevilled headline writers since the form was invented.
(I went looking for examples on bad-headlines lists, but didn't find any. Instead, here's a completely different flavour of Oops for your amusement. It's the juxtaposition that makes it. Chef's kiss!)
"IMHO, it is long past time for AI's to be subject to modern Salem Witch Trials"
Um, you do know that those were innocent people, hanged on the basis of false accusations, right?
(Hanged, not burned. One man, Giles Corey, was tortured to death. I'm not disputing that "witch" burnings took place; just not in Massachusetts.)
"you cannot stop billing your clients waiting for [a software fix]"
Indeed.
I was once out TGIW'ing with coworkers (for reasons, the end of our work week was Wednesdays) when the Accounts Receivable clerk came to our usual watering hole to find me because that week's invoices were all printing with Invoice Numbers of "*****".
I knew what was wrong before I was out of my chair (as I'm sure you do now). Fixing it, on the other hand, was a royal PITA. I had to pull an almost-all-nighter extending the field to six digits [1] so that the A/R guy could reprint the invoice run first thing in the morning.
The point of the story is that waiting till the morning to fix the code simply wasn't an option; the invoices *had* to go out in Thursday's mail.
[1] The sales-order system was a mess of cut'n'pasted'n'hacked-on code (in SCO FoxPlus, a dBase sort-of-clone). I had to do a lot of tracing the invoice-number database field through many code paths, to find all the places where the display format needed to be adjusted.
It was a good thing the A/R guy caught me early enough in the drinking cycle that I was in shape to do the work!
---> for what I had to leave unfinished that night...
There used to be, and maybe still are, time-based MFA implementations that used a dedicated key fob or the like, which displayed a code that changed every N seconds.
ISTR that at least some of those implementations would actually accept not just the currently visible code, but also the one on each side of it -- including the future code that the user's fob was going to switch to next (I.e. each code was "live" for 3*N seconds). The server would then analyze which of the three codes any given user tended to provide, and use that to compensate for any clock drift in that user's fob.
Very clever.
I don't identify exits on entering any building, but you're right -- it's a new habit worth cultivating.
I've always done it when first checking into a hotel room, though: Where are the fire extinguishers? Which way(s) to the exit(s) -- and how many doors away? (Because if the hall is filled with smoke I might not be able to see the signage, either due to the smoke itself or to watering eyes. How many doors to crawl past on hands and knees, should it come to that -- that's a good thing to know in advance.)
At school when I was about 10yo. Two kids were fighting and slammed into a wall, hard enough to activate the manual fire alarm that was mounted on it. (IIRC the alarm switch was one of these then-standard units.)
I no longer recall, if I ever knew, what punishment was meted out.
And the formerly clean, straightforward (on the mobile version of the site, anyway) home page has succumbed to clutter. The last holdout for good design has fallen.
Worse than all that, the text has shrunk and I can't fix it -- my browser's Zoom feature zooms the top-of-page banner, but leaves the actual article text blissfully untouched, i.e. at the limits of readability for me.
(The comments pages are still OK for now, but I'm sure it's only a matter of time.)
Really, The Register, you of all people? Do you really think that you know better than I do what works best for my eyes?
It's really quite simple: Don't. Override. The user's. Preference. Let them decide how big they want their text.
"[...] you count the 3 joints on the 4 fingers (not the thumb) to get your 12."
I do that, using my thumb as a pointer. I learned it from a book.[1]
"And for 60 you basically fold down a finger (and the thumb) on the other hand after each count to 12, in order to count to 60 for the number of minutes in an hour."
Or just run the same technique on your other hand, assuming both are free.
Why do I, a reasonably intelligent adult, count on my hands in the first place? Not for simple cases, but for the same reason they make these: if each of the items being counted takes some thought (e.g. to decide whether to include or exclude it), keeping the running tally on my hands frees up brainspace to concentrate on each individual item.
Tallying with digits instead of whole fingers, I can get up to 99 if using both hands. The only challenge, since I'm half-distracted (because see above), is remembering to do the carry at 10, instead of counting on to 12 as the low-level muscle-control part of my brain wants to do.
If you're OK counting in duodecimal and converting at the end, you can count up to 143 on your hands.
[1] "The Universal History of Numbers", by Georges Ifrah. What I have is a three-volume set: "The World's First Number-Systems", "The Modern Number-System", and "The Computer and the Information Revolution". I've barely dipped my toes into it, but can see myself getting lost for weeks...
I had a girlfriend once whose surname was unique to her extended family. It seems that two or three generations back, someone had misspelled a somewhat common name in a baptismal register, and all the way down the subtree, that's just what their name has been. It wasn't "Smith" -> "Smih", but you get the idea.
"They were expecting a 'Sieve of Eratosthenes' implementation [but I cribbed the list from an external source]"
I went the other way. Being familiar with the Unix factor program, which produces a number's prime factors, I wanted an Amiga version. I wrote it in C (only for 32-bit ints; beyond that was out of scope), because that was what I had available, and it was my main programming language then anyway. It brute-forced the problem by the obvious "keep dividing by this prime until it doesn't go anymore, then move on to the next one".
But it needed the primes to test against. Not having the Internet available (this was the 80s), I wrote a Sieve of Eratosthenes to generate the primes up to 2^16 and spit them out as a C array initializer into a .h file, which then got #include'd into the main factor program.
Probably overengineered but, well, it was a way to avoid doing real work...
"I had to explain to the user that to get billable time in hours you must divide by 60 not 100," Albert recalled. And that's why the numbers were out by about a third!"
GCOS-3 (Honeywell's mainframe OS) used to display the time on its operator console in decimal hours. It would have displayed 15:20:00 as "15.333".
And batch jobs were accounted for in "centihours" -- their word, not mine.
All to save a convert-to-HH:MM:SS routine in the kernel. I mean, memory was super tight back then, but geez!
"I remember explaining to an accountant that if a tax rate had gone up by 5% and had then come back down again you couldn't get back to where you started by reducing your price by 5%."
I'd expect an accountant to know better!
That said, though, it's not an intuitive insight. I remember being confused by it the first time I encountered it as a youngster, and having to think it through to really "get" it.
Which stood me in good stead, years later, when a friend commented about the apparent unfairness that (at the time) the Canadian dollar was worth about US$0.70, but the US$ was worth about C$1.40. (A small part of that is the buy/sell spread, of course, but it's the larger part of the "30 cents vs. 40 cents" WTF'ness that I'm talking about here.)
I explained to my friend that you don't reverse it by subtraction, but by division. To which he replied, "Why?"
Here's how I explained it: "A Canadian dollar is worth 70 US cents ... Which means it's worth 30 US cents less than a US dollar ... Which means that a US dollar is worth 30 US cents more than a Canadian dollar ... Which is 40 Canadian cents more than a Canadian dollar."
I'm especially proud of that because I came up with it on the spur of the moment, in the middle of a somewhat Fear-and-Loathing-esque party weekend in Montreal when we both mostly had other things on our minds...
... including, in abundance, ---->
The worst one I ever saw was for a time-tracking web app. Besides the usual annoying complexity requirements and an onerous expiry schedule, it prohibited double letters.
I'm sure that rule was meant to prevent arrant stupidity like "aabbcc", but it also would have rejected "correct horse battery staple" -- and not because it's become a cliché. Among other fixes, amusingly, one would have had to deliberately misspell "corect".
There's a quarry not too many miles from where I grew up. Back in the 70s there were signs on the road that ran past it, warning people to turn off any radio transmitters. It seems there was some risk of a CB radio or whatever detonating the explosives the quarry used, before the workers were a safe distance away!
(I've never understood how that might happen, TBH. Would they have been using radio-controlled detonators, perhaps?)
Agreed. Hugh can't claim credit for sniffing out the problem with the drive, but can't reasonably be blamed for it either. The specific failure mode sounds like one of those random shit-happens situations.
In general, one might criticize his choice of when to do the cleanup, and I'd definitely agree -- and I presume it's a lesson he learned from this experience.
But suppose he had done the Right Thing and sacrificed part of his weekend. The drive would still have been cooked, the manufacturer's support line would presumably have been closed (seeing as they weren't even keeping the lines staffed over lunch), and he'd still have been faced with production-hours downtime, just time-shifted till Monday morning.
Once that part failed, laying a trap for the next power-down, things were bound to get ugly; it was just a question of when.
The only way out would have been to have the RAID setup already in place -- and that sounds to have been beyond Hugh's discretionary budget.
We're seeing the pair of dots centered over the "n" -- including in your own comment.
So it seems your machine can generate n-umlaut [1], but not render it properly. Can it handle more traditional two-dotted characters: ä ë ï ö ü?
[1] The phone on which I'm typing this has the opposite issue -- renders it OK but the on-screen keyboard doesn't offer it -- which is why I spelled it out.
It was a regular office building with fully functional HVAC. Problem #1: as is often the case, the thermostat was set too low for my liking. I think I'm part cat -- I like things warm, dammit.
Problem #2: one of my co-workers found the temperature too high for his liking. And he had access to the machine room. So he tended to block the door open, so the cold machine-room air would grant him relief.
Problem #3: The machine-room door was just across from my desk, which meant that I -- who was already chilly -- got most of the "benefit" of his efforts, while he, on the far side of the office, got fairly little.
And as a lowly contractor, I wasn't in a position to ask to swap desks with him.
As for what his workaround meant for the servers' security (and temperature)? Well, I grumbled to myself, but as low man on the totem pole, it wasn't my place to raise a fuss...