Posts by Bill Gray

Everything is better with pickles... except Windows

Yes, there are many ways to get out of that sort of pickle.

Open source project fools AI scrapers with poisoned font

Re: Grammar rules OK?

And as someone else pointed out, verbing weirds language.

C'mon, just copy this text string and paste it into your macOS Terminal – it'll fix your computer, honest

Re: The Mac needs a built-in cattleprod

...with the Microsoft version adding, "Would you like help with that?"

KDE Plasma users face a dire omen of change: 6.6.6 arrives

Re: "Book of Revelations, the Oxyrhynchus Papyri, page 115 says that the number is 616"

As a child in the US, I was told that if you played a country music record backward, you'd quit drinking, your wife would come back, and your dog would return to life.

Thief posed as Wi-Fi fixing hero, then stole priceless trophy

Re: Really?

I can understand why you're getting downvotes (not from me, though). But I don't think you're entirely without a point here.

I suspect the overwhelmingly greater security threat in most organizations is from the simpler attacks : people whose passwords are 'letmein' or 'qwerty', single-factor authorization, clicking on dodgy links, and so forth. It kinda resembles the way some people worry a lot about incredibly rare risks such as terrorist attacks, then go out and drive like maniacs without their seat belts attached.

I would certainly not ignore this kind of security threat entirely, especially if I were involved with a Fortune 500 company where it might be worthwhile to go to that sort of trouble. But I'd try to keep in mind that the more common attacks represented a much greater risk to the company (and much less of one to the criminal).

Windows 95 detected installers by looking for magic words and hoping for the best

Always be careful when you're making sweeping statements or generalisations.

Well... most of the time, anyway.

Four years into Ukraine invasion, Russia turns influence-ops back to US and Europe

Re: Is AI really helping?

AI is a bullshit generator.

Spot on.

Obviously, the overwhelming majority of commentards are dismissive of AI. That makes sense. We aren't the target audience. We're mostly in roles where we're trying to figure out a Right Answer to a problem. That's not AI's strength.

But in many fields -- politics, sales and marketing, law, scamming (sorry, I repeat myself) -- you have little or no interest in Truth. You start out with an answer and do your best to support it. In that situation, inaccuracies and hallucinations are not as worrisome. Supporting arguments that correspond to reality would be nice, but are not essential.

If you need to send out a lot of junk e-mails or "social" media posts persuading people to vote for a particular candidate, buy a particular product, or send money in hopes of collecting a share of your late dictator spouse's assets, AI would be an extremely useful tool.

Portuguese bank sign's storage is about to cash out

Re: Signs of things to come

I'm confident that as long as we have computers, we'll have bork.

Mythos discovers 'Squidbleed,' a memory leak that's gone undetected since Clinton era

Looks as if they got the order of arguments right :

https://github.com/squid-cache/squid/blob/0b8e893e52b278b87f9ade137c2ddf65195ce028/src/clients/FtpGateway.cc#L624

while (*copyFrom && strchr(w_space, *copyFrom)) /* now, fixed version */

while ( strchr(w_space, *copyFrom)) /* before, broken */

Elsewhere, w_space is #defined to be the string " \t\n\r" (i.e, spaces, tabs, carriage returns, and newlines are whitespace). So, as long as *copyFrom is one of those four characters, increment the pointer.

I must admit that I'd have considered a '\0' line terminator not to be in the string. I'd think of the string as containing the four characters mentioned above. C, apparently, considers it to consist of those four plus the implicit '\0' at its end (which is arguably correct; C strings are `\0` terminated, so that byte is part of the string.) Today I Learned, etc. I didn't really believe it until I wrote a program-let to search for '\0', and sure enough, strchr() finds it:

https://www.projectpluto.com/temp/w_space.c

I have found Valgrind and similar tools to be lovely for debugging this sort of thing. Read or write off the end of an array, and it lets you know about it.

How dare you stop data loss – that's not your job!

Re: Bah!

I have mentioned before that our place used both wooden and metal desks, and the union agreements had it so only carpenters could move the wooden ones and tinsmiths the metal, and how one move revealed a desk made of both that cause a demarcation discussion.

I'm reminded of the (probably apocryphal) story of roofers and siders working on a Quonset hut. The roofers argued that the structure was one big roof, and was therefore their job; the siders argued it was one big wall, and therefore their job.

The eventual decision was that the roofers did as much as they could. When they got far enough down that they couldn't stay on, the siders took over.

Tensordyne makes a big bet on log math to beat Nvidia

Re: I think they just invented the slide rule.

Long ago, I memorized squares of numbers up to 100, partly because it enabled me to (usually) multiply two-digit numbers in my head quickly. If a and b are both even or odd, pull the squares of (a+b)/2 and (a-b)/2 out of wetware memory and subtract. Otherwise, halve the even one, then apply that method and double the result.

It's not always faster than "traditional" methods, even for mental arithmetic. But it helps in a lot of cases.

Microsoft's latest Windows bug belongs in the Recycle Bin

Re: M356

I will take a guess that our fellow commentard knew what you meant, and was making this obscure (to me, at least) reference to an anti-psychotic drug "used to treat major depressive disorder, schizophrenia, or agitation".

It's been about ten years since I really used Mic__oft products much, but I could imagine daily users of its wares being in need of such medication.

Users claimed they'd never seen a spell checker and panicked at the sight of red squiggles

Butterflies.

Consultant mistakenly deleted a ton of data – but reported it as a bug

Re: Works with hardware, too

Circa 1990, in my pimply-faced youth, I inserted an 80387 math coprocessor chip in a '386 box at $WORK. Turned the machine on, and didn't get a speed-up in floating-point math code.

I then learned that

(a) the '387 wasn't oriented the right way;

(b) if you looked pretty carefully, you could see some sort of marking that told you which way it was supposed to go (though there was nothing preventing you from putting it in the wrong way);

(c) putting it back in the right way wouldn't help; the coprocessor had already gone to Silicon Heaven.

Like you, that was enough of an error to make me much more careful around equipment.

Note to self : should I ever submit a story to this august news organ, modify a detail or two to make it sort of, but not quite, point back to me.

(I would guess that most submitters do exactly that. Actually, were I the vulture in charge of this column, I'd suggest to the submitters that they make such alterations.)

Next stop, C:\ ... Paris Metro screen goes off the tracks

Re: the can-can dance, in which participants perform high kicks that reveal their underwear

Hadn't heard of that. But a reliable source confirms :

The dance was considered scandalous, and for a while there were attempts to suppress it. This may have been partly because in the 19th century, women wore pantalettes, which had an open crotch, and the high kicks were intentionally revealing.

Well, it was France in the mid-1800s. You could get away with that sort of thing. Sadly, you couldn't when I was a hormone-driven teenage lad in the US in the 1980s.

Microsoft wants safer C# without turning it into Rust

Re: Am I missing something here?

Having a pointer to the end of an array, and then checking to see that another pointer hasn't gone past that point, is reasonably common. Also, NULL pointers.

Lack of proper indentation is due to unfortunate deficiencies in this fine forum.

void do_stuff_in_array( bogostruct *iptr, const size_t arraysize)

{

bogostruct *endptr = iptr + arraysize;

while( iptr < endptr)

do_stuff_to_bogostruct( iptr++);

}

MyPillow must decide whether to be firm or soft as ransomware crims demand pay

Re: I'll set up a webshop now

Especially in Germany. We can't have gratuitious Saxon violins. -->

"Sex on television is only a problem if you fall off."

Marketing demanded IT add website feature that was already working

On the other hand...

I've occasionally been asked if I could add feature X to my software, started looking into it, and only then realized that I'd added X a decade or two ago.

Some of this is due to having worked on the same software for a few decades. So (a) there's more for me to attempt to keep in my brain and (b) said brain is now a few decades older. As Dan Quayle famously pointed out,

What a waste it is to lose one's mind. Or not to have a mind. How true that is.

Lab worker built a fake PC to nuke his lunch

Re: Radiation hazards are often overblown.

I assume he's talking about the risk commonly found (and quite real) from radon that's the decay product of uranium in granite. In some parts of the world (including mine, in Maine in the northeastern US), where we have lots of granite, this is a regular problem. The stuff tends to accumulate in basements (it's much denser than air), and testing for radon is a regular part of building or buying a house. Abatement usually consists of setting up fans to hoover it up and send it outdoors, and sometimes in trying to seal the basement.

As you say, the risk from just having a lump of uranium around is tiny. Not much of it is going to decay in our lifetimes, still less decay all the way down to radon, and less than that will make its way out of the lump.

Re: Radiation hazards are often overblown.

After the war, it wasn't so much that uranium was more controlled in most places (though some have laws against private possession), it was more that Cold War era governments were buying up all they could...

During the Second Big Mistake, my grandfather was manager of a glass factory in upstate New York, and received an order from the government to turn over any uranium the factory might have "for use in the war effort". He had a degree in physics and had a pretty good guess as to why they might want it.

60% of MD5 password hashes are crackable in under an hour

Re: Timing of MD5, SHA1, SHA256, SHA512

Thank you. I'd tried various things, but your post got me at least headed in the correct direction.

I'm running Linux, so the method for entering a non-breaking space is somewhat different. For starters, a non-breaking space is 0xA0 = 160 in Unicode, and you use Ctrl-Alt-U A 0 to type it. &nbsp; doesn't work. Also, for unknown reasons, I can enter them right after the < code > tag (but it has to be right after that tag; on a new line, the nbsps are ignored. So each line has to have its own code and /code tags.)

      Here's a thoroughly indented line between code and /code tags

      And here's a second indented line,          with spaces in the middle, also between code and /code tags.

Line spacing is a bit large, but this is a good step up from where I was.

Re: How long and complex is safe enough?

do hashes and salts get safer if they are longer and more complex, or does this issue no longer have to do with length and complexity?

The site you referenced gives some specifics of how they computed those times. They assume use of 1024 iterations of bcrypt and twelve "consumer-grade graphics cards", the NVIDIA RTX-5090.

So first off, if your banking site just runs SHA-whatever once and calls it a day, which seems likely (many banks have unimpressive security practices), it won't take nearly as long to crack your password. If they run bcrypt once, they can run 1024 times as many guesses a second (ignoring some per-guess overhead).

However, that will presumably drop the time requirement from, say, trillions of years to "only" billions. Or from millions of years to "only" thousands. I would want the projected crack time to be padded by a few zeroes to allow for the likely stupidity of the banking site.

With a reasonably long password, you at least manage to avoid the problems pointed out by Kaspersky. They're basically saying "if there's salt but the user ID isn't used, we can crack a lot of dumb passwords quickly". That's not news (well, not to those on this site, anyway... admittedly, we aren't their target audience.) It just means you shouldn't use a dumb password.

Timing of MD5, SHA1, SHA256, SHA512

Thank you! That makes generating the following table remarkably easy. The program runs speed tests for blocks of 16, 64, etc. bytes, simulating both password hashing at the small end and file hashing at the large end.

The 'numbers' are in 1000s of bytes per second processed.

blocksize 16 bytes 64 bytes 256 bytes 1024 bytes 8192 bytes 16384 bytes

md5 143135.21k 337991.13k 595097.77k 739155.63k 791857.83k 787152.90k

sha1 163776.57k 375694.87k 755987.71k 957767.34k 1054228.48k 1036894.21k

sha256 90761.17k 204461.03k 359723.43k 457895.59k 490370.39k 495495.85k

sha512 62222.64k 248377.24k 432412.59k 644711.08k 733235.88k 708105.56k

(I can't get the above to display quite as nicely as I'd like. The 'pre', 'code', 'tt', and 'table' HTML tags don't do what you'd really want. Here's a table-ized version of the above that may be easier to read.)

I would think that salt+userid+password would be, mostly, a bit more than sixteen bytes but quite a bit less than 64. So, somewhere between the first two columns of the above table.

Had Kaspersky used SHA-1 instead of MD5, they could actually have claimed even faster password cracking (though still with a deprecated hashing algorithm.) Had they used either of the last two, they'd have cracked fewer passwords, but still enough to make their point. And they wouldn't have every El Reg reader mocking their use of an obsolete hash. (I will still mock Kaspersky's assumption that you'd hash salt+password, and not some form of salt+userid+password. It is, after all, only an error of a factor of 200 million. I am a kind and forgiving person, inclined to assume honest error most of the time. But I'm pretty sure they know better. This looks like a cooked experiment to me.)

True, but that doesn't really matter that much here. (See post above for deeper issues in what Kaspersky did.) They didn't use an MD5-specific attack; they just hashed a lot of possible passwords and checked which ones matched.

Had they used SHA-256 or SHA-512, each hash would have taken longer. I'm not succeeding in finding out how long it takes to hash a short phrase with MD5 vs. SHA1 vs. others.

However, hashing a large file with md5sum, sha1sum, etc. on my machine, MD5 and SHA1 have similar speeds. SHA-256 takes about 2.5 times longer, and SHA-512 about 1.5 times longer. (SHA-512 is a little better optimized for 64-bit systems.) But I'd think that for password cracking, most of your time would be in initialization (not sure of that), so my large-file test is probably misleading.

Still, you get the idea : if SHA-256 or -512 is (say) ten times slower than MD5, it'll take ten minutes instead of one to crack half the passwords. (Assuming they're salted, but the user ID isn't included in the hashing... see post above.)

Plain hashes? or salted-hashes?

I wondered about that as well. The referenced article discusses the various possible attacks, but doesn't say which one was used or whether salts were used.

But if they were plain hashes, you'd use a rainbow table and probably break these in seconds. With salted hashes, you'd have to concatenate the salt with each "likely password" (letmein, 12345, password1, etc.), hash the results, and see which ones match hashes in the leaked data.

So I think they assumed hashes of salt+password. Another clue to this is the following from the referenced article :

It’s worth noting that cracking every password in a dataset doesn’t take much longer than cracking a single one. During each iteration, once the attacker calculates a hash for a specific character combination, they check if that same hash exists anywhere in the dataset — and the larger the dataset, the easier it is to find a match. If a match is found, the corresponding password is flagged as “cracked”, and the algorithm moves along to the next one.

That works if you assume salt+password. In that case, you hash (say) salt+letmein, and can immediately identify the 20% of users whose password was letmein.

If you instead Do Things Correctly and hash salt+userid+password (or similarly mix in the user ID), things are not so easy, because salt+Alice+letmein will hash to something different from salt+Bob+letmein. For a database of 200 million passwords, things become... well... 200 million times more difficult. Instead of cracking nearly half the passwords in under a minute, you require 200 million minutes... call it about six years.

You could reasonably complain, therefore, that Kaspersky has made cracking look easy by saying "if people use an obsolete hash and don't mix the user ID in the hash, passwords are easy to crack." (And I would agree with you.)

But Kaspersky could reply that on the evidence, plenty of sites screw this up. I suspect that when a site limits password length to twenty or fewer characters, and requires no commas or quotation marks, it's because they're storing my password in plain text and have to fit it into a fixed-length record.

BOFH: Nothing says 'business continuity' like a dry wooden broom

Re: That ominous silence

Some days, you get the grue; some days, the grue gets you...

Hobbyist xenomorphs Raspberry Pi into Alien-themed DIY laptop

Re: Ah, progress

In 1979, I was a spotty 14-year old lad in a school that had a PDP-11 with four video terminals hooked up to it, sharing (I think) 16 KBytes. And we were darn lucky to have it, too. 512 MBytes would be... uh... 32000 times what we had at the time. I don't think it would have been more memory than existed on the planet at the time, but it'd have been a significant chunk of it.

(If you've been whooshed, so have I.)

User found the perfect formula to make Excel misbehave

Re: Analogue clocks

Sweden's calendar included a February 30th in 1712. (Part of a mangled effort to switch from the Julian calendar to the Gregorian one, then changing the plan to go back to the Julian.)

Re: musical notation was probably one of the oldest programming languages .....

A sharp comment. One to take note of, and tough to beat.

(Sorry if some of these puns fall flat.)

SpaceX rocket set for unintentional Moon landing – well, a piece of it anyway

Re: Scrapyard moon

That is indeed a problem for objects actually orbiting the moon, but we don't really have very many of those. The much more common case is a flyby, or a mission that goes into orbit but leaves the upper stage flying past the moon. (Obviously, we do have the problem that both the orbiters and the flyby upper stages are increasing in number.)

An upper stage passing by the moon doesn't spend all that much time within the moon's sphere of influence. And, those "lumps" are modelled via a series of spherical harmonics that drop off quickly with distance. Models have been constructed for the gravitational fields of the terrestrial planets and the moon, and even for the much "lumpier" fields around Ceres and Vesta. (For the gas giants, their oblateness matters, and a subset of the spherical harmonics matters. The oblateness of the Sun matters, but it rotates so slowly and is so close to being a sphere that it's barely noticeable.)

Bottom line : if you wanted to plan an impact, you could do so (as was done for, for example, the LCROSS mission in 2009). It will usually be simpler to use the initial lunar flyby to eject the upper stage into heliocentric orbit, though. I think this is what was done with the New Glenn upper stage for the ESCApade mission late last year. (It definitely ended up in an orbit that will put it into heliocentric orbit for a long time, but I don't know if that was a lucky accident or a planned thing.)

I will not be surprised if the eventual "solution" involves throwing the upper stages into heliocentric orbit (because that's easy to do) and being slightly more careful about where satellites that are actually in lunar orbit crash.

Re: Instrumentation

Sky & Telescope article about (six) lunar flashes seen by Artemis 2 crew. Looks to be small meteoroids, somewhat more than were expected to be visible.

"seven times the speed of sound"

(I am the Bill Gray mentioned in the article.)

You can sort of blame me there, I suppose. Here's what I wrote on the subject :

https://www.projectpluto.com/25010d.htm#speed

There is, of course, no air and no sound on the moon [0], so a "Mach number" doesn't really make sense. But if there were air, the speed would be about Mach 7, seven times the speed of sound.

I'll claim some experience here, though. When the (only) previous bit of junk hit the moon unintentionally, the Chang'e-5 T1 upper stage in March 2022, I was frequently asked how fast it would be going when it hit, both in units all of us here would (I hope) understand, such as kilometers/second and km/hour, and in "traditional" units.

Much as heights are often measured in giraffes and volumes in Olympic swimming pools, I was asked for a more reader-friendly way of putting such things. I figured the above would do it.

[0] I managed to resist the temptation to insert "nobody can hear you scream".

NASA boss: make Pluto a planet again

Specifically, a Kansan (and you'll note that Isaacman was replying to the US Senator from that state).

Perhaps I'm wrong, but I think I might cut Isaacman some slack on this one. In his position, I would probably not give a flying (censored) about the is-Pluto-a-planet question. But I might recognize it as being politically important to the senator, and the senator as being politically important to me.

For that matter, it might not even be of great interest to the senator. But it might be of some importance to his constituents.

Re: exoplanets

I'd be happy to invite him into the bar with me, but there isn't mush room.

Does the nomenclature 'planet' serve any actual scientific benefit?

I think you've nailed it.

At the time of the 2006 IAU decision, I was selling a popular desktop planetarium software under the corporate name "Project Pluto". So people frequently asked me about the controversy.

My reply was that it was about as important as the distinction between rocks, boulders, and pebbles. Geologists get along quite well without rigid definitions of those terms; I could not come up with a reason to be interested in a definition of planet, dwarf planet, brown dwarf, etc.

(It occurred to me later that there could be one issue for me. My software would show, for example, a table of data on planets, and/or a dialog in which you picked a planet for the sky chart to be centered on, and so forth. And you could turn display of planets on/off independently of asteroids. I didn't really want to change my software to reflect the IAU's changed definition.

So I didn't. I continued to sell the software for some years. Nobody even asked why Pluto was still treated as a 'planet' in it.)

PowerPoint punishment sent users into an infinite loop after lunch

Re: Re "The law binds none of us unless it binds all of us"

Stop Tolkien out of your arse! I've hobbit with these puns!

Using the password 'admin123' wasn't as bad as sharing it on Slack

Re: I'm surprised...

At the very least, they should have remembered that each extra character adds about six bits of entropy (depending on how restrictive the system is about what characters it will accept.)

Had they gone with 123456789, they'd have been fine.

IT manager approved downtime over lunch, but made a meal of it

Re: Timing

There's probably a named Law

Not exactly on point, but there is (Douglas) Hofstadter's Law : "Everything takes twice as long as you expected, even after taking Hofstadter's Law into account."

Windows Update is a torture chamber for seldom-used PCs

Re: Windows circles of hell

However, for the last few months Zoom has been problematic on Linux

Interesting you say that. I've not had any problems. But I've been running on four or eight-core systems, and it does use two or three cores, which seems rather excessive to me. And I'm running Xubuntu 22.04 and most likely have a different graphics card from you. So it's to be expected that our mileage would vary.

When I read in The Fine Article that the author's temple had a Windows machine used solely for Zoom®, I had the thought (expressed ad nauseam in these fora, I realize) "just install Linux". Your ideal case for so doing is one where you aren't doing anything Windows-specific anyway. In this case, with the laptop being essentially an appliance to run Zoom, I'd install a minimal distro (Lubuntu, maybe) and see what happens. But there'd be some risk they'd run into the same problem you did.

Apple update looks like Czech mate for locked-out iPhone user

Re: Could have been worse...

I'd go with 'octothorpe'.

DARPA looking for battery that could power a laptop for months

I think the difference is that lithium batteries are (mostly) thought of in terms of capacity : total watt-hours, and you're concerned both with how much power they generate and how long they're going to do it. With a radioactive battery, you're concerned with the power generated and (usually) not so concerned about how long it's going to last.

So describe them in watt-hours, and the radioactive batteries look wonderful, and we quietly ignore the fact that most of their watt-hours won't come out until decades after the mission is over. Describe them in watts, and the chemical batteries look wonderful, and we ignore the fact that they die young.

When the Voyager spacecraft were launched, I'm sure a lot of thought went into how much power the RTGs would put out. The Pu-238 has a half-death of 87.7 years; that comes to a whopping number of watt-hours by the time they (for practical purposes) go dead, but most of it would occur long after the mission's expected end (about a dozen years from Voyager 2's launch to the Neptune flyby). If they got lucky enough that it lasted that long... I doubt anybody expected it'd still be running 48 years after launch.

And now I've been nerd-sniped again... the usual source tells me that Voyager 2's RTGs provided 470 watts at launch. Total power generated would be equivalent to running at that level for 87.7/ln(2) years and then stopping, so

470 watts * 87.7 years * (365 days/year) * (24 hours/day) / ln(2) = 6 million watt-hours

Shots fired – literally – over proposal to build datacenter in Indianapolis

> You're also very naïve if you think "either party" are without corruption.

(Sigh) Look, I realize we're deluged with the idea that they're all corrupt. Much of that comes from politicians who are, themselves, corrupt, and figure their best bet is to say "everybody does it; just give up and let me get elected".

It's been quite effective, unfortunately, and has ensured a self-perpetuating cycle : you get some corrupt pols, they smear others whether they're actually corrupt or not. That makes politics a less appealing pursuit for any honest citizen, bringing in more corruption, which...

At this point, your average USAian (other countries as well) has a higher opinion of telemarketers and used car salesmen than they do of politicians.

But it would be naïve to simply accept uncritically the corrupt politician's argument that "they're all corrupt". Yes, a hell of a lot of them are. Many are not corrupt, but are highly ideologically motivated to the point where it interferes with their ability to reason. (Increasingly a problem in a country where most of its citizens are highly polarized. If Congress didn't contain a bunch of loons, it wouldn't be a representative body.)

When a billboard survives the wind, but not the boot

Yank here. I think it would indeed be considered somewhat dated, but would be known to most USAians.

Artemis II countdown begins as NASA prepares for crewed Moon flyby

Re: Communication

Sometime in the early 1960s, there was a paper by Jean Meeus listing transits of the earth and moon as seen from Mars. The next one happened on 1984 March 11. Arthur C. Clarke took the data and used it in a short story, Transit of Earth, about a stranded astronaut on Mars observing the transit before running out of oxygen. In hindsight, the optimism of writing a story in 1971 with a landing on Mars in 1984 is pretty impressive.

As mentioned in the story, the next such transit will be on 2084 November 10. That one might actually get observed, if we don't destroy ourselves before then.

Re: Both?

I'm reminded of the old joke about men having enough blood to operate our brains or our fun bits, but not both at the same time.

Re: Communication

> Even on Mars, you can see Earth, probably need a telescope

Okay, I've been nerd-sniped.

As a lad, I read a Heinlein novel (Red Planet) in which, at one point, our heroes look back and can see earth, and even the moon next to it at fifth magnitude. Fifty years ago, I had no way to check Heinlein's work (though he usually did do his best to reflect the known science of his era). Now, I can. Let's see...

As with Mercury or Venus as seen from Earth, Earth doesn't get all that far from the Sun as seen from Mars, goes through phases, etc. Maximum brightness was last January at around mag -2.1, about that of Jupiter as seen from Earth. The moon, to my surprise, is at mag +2.1. Roughly the ballpark of the stars in the Big Dipper. (Computed using the desktop planetarium software I wrote some years back, then checked against JPL Horizons, then against a back-of-the-envelope calculation.)

The 4.2-magnitude difference puts the earth about fifty times brighter than the moon. Given that we're four times bigger (and therefore sixteen times more area), and that our albedo is a little more than three times that of the moon, that all fits.

I will speculate that Heinlein figured the magnitude for a time when the earth and moon were closer to inferior conjunction, and therefore not at their brightest.

To BSOD or not to BSOD? Only Microsoft knows the answer

Both screens are from that site

Nicely spotted. The second is at https://fakebsod.com/windows-8-and-10/. Both are set up so one can visit the page in question, hit F11, and move the mouse "under" the error message (as described on the site itself).

The wording and/or graphics of both BSODs could probably stand some tinkering by the assembled commentardiat. Anyone have ideas? (Possibly an El Reg contest, with the prize to be a BSOD jumper...)

Engineer sabotaged hardware then complained when it didn't work

Chesterton's fence

G. K. Chesterton wrote something that boils down to : if you see a fence running across a road, you shouldn't tear it down until you figure out why it was put there. Somebody presumably went to the time, trouble, and expense of erecting the fence, and had some reason for doing it.

You may eventually learn that their reason no longer applies, or just doesn't matter as much as it used to, and then you might pull the fence down on a suitably informed basis. But you shouldn't equate "I don't see why that's there" with "there's no good reason for that to be there".

As I recall, he was mostly thinking in terms of politics. The idea is that each generation comes along and assumes its parents were idiots, and that society should be rebuilt on more sensible, modern principles... usually without first considering why the parents did such idiotic things. But it's a good engineering principle as well.

Junior disobeyed orders and tried untested feature during a live robot demo

Re: Short cuts make long delays, but at least nothing shorted out

Well, in theory, theory and practice are the same thing. In practice, they aren't.