Posts by Nigel 11 3191 publicly visible posts • joined 10 Jun 2009
<paranoia class="professional carefully-nurtured">
The data came from a 3-letter agency. Anyone who has made any use of it will be bumped up their watch list. Like the Reg's journalist (except journalists are probably already a lot higher on the list than John Doe, it goes with the profession).
You'll note I say "up" not "onto". Everyone who has ever made a phone call is on the list already.
</paranoia>
Vendors of devices containing software are allowed far too much latitude to escape any product liability with respect to latent bugs. Contrast the auto industry, where if a serious latent problem is discovered with a car, they have to recall and fix all the cars (or face paying out billion-dollar damages, witness the Ford Pinto). They can't get away with just saying "it's out of warranty" or "it's an old model" or "read the license disclaimers". Similarly, phone manufacturers should be obliged to fix bugs that were present in the device at the time it was sold, or in subsequent versions of its software, use of which is required to provide a fix for day-zero bugs. This for at least five years after sale, preferably ten.
Of course, the result of stricter product liability would either be more expensive phones, or fly-by-night manufacturers of cheap phones whose business plan includes going into liquidation within a year.
If Samsung don't patch my S4, the next phone I buy will be a Google one (the only company that pretty much can't evade its moral responsibilities). That, or there will be a breakthrough for a properly open source phone with Linux-style community support.
Not actually true that there's no daytime household usage. Trouble is, a fridge is high-wattage low-duty-cycle device, whereas a solar panel is low wattage continuous, and the electricity utilities don't have means or incentive to take electricity off your hands now and give you credit for it when you want it back a few minutes later!
Now, here's a sensible use for "smart" appliances (fridges especially). Let the inverter know (via WiFi) when it is running its compressor. Then, an inverter with a one-hour battery (maybe even a 15-minute battery) would make perfect sense. Synch your generation with your fridge's demands. Cookers, dish-wasters, washing machines, might join the party, but the fridge is the appliance that's always in use.
Seems like a complete waste of high-grade energy, but the most cost-effective use for a small solar PV installation here in the UK under current rules is probably to run a low voltage immersion heater in your hot water tank! (No inverter needed, just a thermostat)
If you want to know how big are the files in various folders (and assuming Windows 10 still uses NTFS) ... shut windows down and boot a stand-alone Linux CD. du is probably your first port of call once you've mounted the hard disk under Linux (read-only if you are paranoid).
I'm assuming you can still do an install without obstructive things like secure boot and encrypted filesystem.
Unless there's some thermal contact in that design between the CPU chip and a largish metal plate in the case, there's a fan in there. Also the case design doesn't seem to gave enough holes for passive (convection) cooling. Also convection cooling works best in a "chimney", i.e. a case orientated vertically with air holes bottom and top. (If you stick to SSD you don't have to worry about your data getting trashed when the cat knocks the PC over).
why anyone would put a Celeron in anything for any reason
Because they have nice low TDP figures and can run with passive cooling (without requiring ludicrously expensive heat pipe and case solutions). And yes, because it is Intel, not ARM, for software compatibility.
My home desktop is a Gigabyte J1800 Celeron-based mini-ITX board coupled to an SSD. A PC with absolutely no moving parts, which is therefore totally silent. It's also quite fast enough for everything I use it for. (It's running Linux not Windows, which may be part of the explanation).
Back to this review .. an "exceptionally quiet PC". Does that mean, with an irritating mosquito of a fan in there, that will get louder and louder as the heatsink clogs up with fluff? If I'm wrong, the right word is silent. BTW if you have to have a fan, you want a big one not a small one. The slower it rotates, the quieter it can be.
"Is there any solid, proven evidence for anthropogenic global climate change? Serious question".
Is it?
By the time there is that sort of evidence, it will be far too late to do anything about the changes. we'll have to live with them, or perhaps die because of them.
What is certain is (a) the measureable increase in atmospheric CO2 since the industrial revolution, and (b) the certainty that CO2 is a greenhouse gas. I'd far rather we stopped raising the amount of CO2 in our atmosphere *now*, rather than after it's too late. Especially since we now have the technologies to do without burning stuff for energy, and lack only the will to develop and deploy them. (Taxpayers are still *subsidizing* fossil fuel production, FFS! )
One acronym: ECC. Or one word: professional.
For gaming, you don't give a monkeys that your RAM might drop a bit or two, and corrupt the results that it is generating. But if you are designing bridges or buildings or chemical plants, or even if you are just doing research for a Ph.D or big-money financial planning, you certainly ought to!
You really don't want anything radioactive integrated into a VLSI chip. Each radioactive decay can corrupt a bit of data, just as well as contributing to a bit of random numbers. It all depends on what random direction it is emitted in!
OTOH all motherboards ought to include a noise diode and supporting electronics, and a "gold standard" radioactive-decay noise source on a USB dongle shouldn't be expensive.
Could *you* remember a strong 30-char pw?
Yes - sort of.
Generate a print-out of a lot of entropy. Say a 10 x 10 grid of 5-digit random numbers. Keep it with you. On its own, it can't be used.
What you remember is a hash algorithm for combining the something you have with something you remember to generate a password, that's computable in your head.
For example: Row 8 column 5 reading vertically upwards, six 5-digit numbers. Hash the first by adding d,d,m,y,y of your birthday, each digit modulo 10. The next one, your Mother's birthday. Sister, Spouse, bank sort code (five digits of), bank account # (ditto)
remember: 8, 5, up, self, mum, sister, wife, bank, bank. Not hard.
Better still make the something you have into something(s) with innocent utility. Bank cards. Driving License. Loyalty cards. Torn-off corner of a newspaper stock prices page with a reminder scribbled on it. These don't raise suspicion like a page of explicit random digits does.
Other people will find the surreal imagery trick works better, which is how people have managed to memorize entire telephone directories. Construct a sentence of words that don't normally go together but which do parse correctly. "Green ideas sleep quickly in well-padded spoons". That sort of thing.
Is a pair of normal speakers in a room THAT much of a problem?
Depends on your wallet, your room size, your partner, your landlord, and your masonry.
Speakers: large doesn't always imply good (there are crap big ones for poseurs with defective hearing). However, given the best at any particular price, the smaller it is the more restricted its bass output will be.
If you're willing to accept studio monitor speakers, you can get them out of your normal living space by wall-mounting them several feet up. Except, the location on the wall has considerable effect on the sound. You may have to try several different locations for your speaker brackets, each requring the drilling of several holes into the masonry, plugs, big screws .... So allow a day for audio experiments, and another one to fill and paint over the holes that were in the wrong places.
If your walls are single layer plasterboard, they may not be strong enough ....
( If you don't understand why simple corner shelves from B&Q aren't the answer, and can't hear the reason after installing them, then there's something wrong with your ears! )
Except that I doubt whether a 200 mg Ibuprofen tablet is actually within half a milligram of that dose. There wouldn't be any point given that adult human body masses vary by a factor of two or more (depending on your definition of normal adult human). 200mg +/- 10% would be just fine. (Manufacturing accuracy is probably better than that without adding any cost). Scientists quote the amount without gratuitous extra digits, and the error separately. 1 gram or 1000 mg +/- 0.03g or 30mg is fine. 1000.0 mg +/- 30mg would be silly. 1000mg equals one gram and shouldn't imply anything about the error when no error is quoted.
Drugs for which a 50% extra dose is dangerous are unusual, and require individual prescriptions based on a patient's body mass and/or metabolic variables, also careful ongoing monitoring for toxicity. An ideal drug hits its maximum therapeutic effect at one dose and doesn't become toxic until a much higher dose. Ibuprofen (which I know about) maxes out at around 2400mg per day (a doctor-prescribable dose) and doesn't usually have serious toxicity issues at considerably higher doses ... it's just useless to take more. A greater danger is the long-term effect on your body of the drug doing exactly what it is supposed to do (ie suppressing inflammation).
Incidentally with many drugs (including Ibuprofen) it's safe to take a doubled first dose for a more rapid effect. That first dose finds your body "empty". When it's time for the next, half the original dose is still not present. The third dose finds a quarter of the first and half of the second ... Don't try this trick ("front-loading") without asking a doctor, or at least carefully checking the literature. There are exceptions.
But just how many do you have to 3D print each day to make the printing process profitable? (If the answer is a large number, increase the price.)
Most pharmaceuticals have extremely high margin over manufacturing cost during their period on patent. On the flip side, they also have huge development costs to recoup, and the profit has to cover the development cost of a large number of other drugs which fail their clinical trials and never make it to market.
Why is a gram of this drug packaged as a conventional pill so much larger than a 1 gram vitamin C or fish oil capsule? (I take these daily with no difficulty at all swallowing them).
A good use for 3D printing might be time-release medication. Print a complex microstructure of cells with walls of different thicknesses containing a drug, that will release as a constant rate over 12 or 24 hours inside one's digestive system.
OfficeJet 8000 models and predecessors - replacing a print-head is as easy as replacing a cartridge. But buying two print-heads costs more than a new printer, and if one has just failed you have to consider that there's a high risk of the other one following it soon.
They don't change the head design very often. ISTR that there have been only three iterations in the last two decades: the type 10/11, the type 88, and the type 940 (ink cartridge numbers, I think printheads use the same numbers).
Universal drawback for Laser printing.
If you leave a stack of laser printer output bound under pressure for years (such as in lever-arch binders or just a big heap), the ink will tend to adhere to the opposite page. The pages peel apart when needed, but there's some ink transfer which at best makes the single-sided pages look grubby, and at worst makes the double-sided pages illegible. Related to this is the (theoretical? ) possibility of someone malign lifting selected characters off the pages. Lawyers will advise printing your will using an ink printer for this reason.
Colour Laser printers are either expensive to buy or expensive to run or both. (The quoted running cost never seems to include replacing the drum, which can last as little as 10K prints).
There are two different methods of providing ink, typified by HP on the one hand, and the older Epson printers on the other.
To be fair, it' s the cheaper HP printers that use this model.
The HP OfficeJets use ink-only cartridges, and print-heads that are also user-replaceable. The running cost of these comes out pretty competitive in the field of printers in that price bracket (broadly £70 to £200).
The problem with expensive printers and cheap ink is seen when some idiot loser wrecks the printer by poking his fingers somewhere he shouldn't, or by yanking out a jammed sheet of paper gorilla-style leaving bits of paper jamming up the works ( or bits of printer scattered on the carpet). This is one reason I like the OfficeJets. They're just about cheap enough to replace out of the consumables budget when the losers wreck them.
My one puzzle is why are the replaceable print-heads so expensive, that it's cheaper to replace the whole printer shoud a print-head fail out of warranty? How does HP make more money out of shipping a vast lump of plastic and metal, than print-heads sold at half the price that they currently charge?
No reason I can think of why you can't run two of them in a Linux software RAID-1 set.
What I wonder about, is whether SSDs of any sort can be trusted to report data that's gone bad in the same way as hard drives can. Checksums and ECC codes are utterly fundamental to spinning rust drives, but even with these it's possible for a controller failure to allow undetected data corruption. And RAID-1 normally assumes that if a write succeeded without error, a read may be satisfied with data off either disk in the RAID array without checking that it's (still) the same on the other device.
For the paranoid, these devices might be fast enough that a new RAID class could be defined and used without crippling loss of performance. Minimum three mirrored members. On read, get the data from all three members and check for equality. If one differs from the other two, it can be assumed to be the bad one. Two members wouldn't let you know which was the good one (assuming that the bad drive wasn't detecting its own failure state).
Like navigating with one chronometer or three chronometers, but never two.
Wasn't in the review? Anyway, Googled it around £180 for anyone wondering.
Yes, you can buy a SATA SSD for less than half the price. But if you need the speed of a PCIe drive, SATA is useless however much cheaper it is! Hopefully competition will rapidly drive PCIe drive prices down to SATA SSD levels.
I want one, but don't need one!
Its about time some of the reliability enhancing technologies used in servers migrated down into 'user' devices.
You can't have been around for long. It was only a decade or so ago that Intel stopped selling Desktop chipsets with ECC support. (Cure 2 Duo CPU era). ISTR that AMD chipsets still do support ECC, although for some reason most MoBo manufacturers disabled it at the board level. Certainly it wasn't many years ago I constructed a microserver around an AMD Sempron rather than anything Intel for this reason. It didn't need speed but it really did need not to corrupt data without anyone knowing that it was failing.
Doesn't happen often, but I have seen filesystems trashed by an undetected RAM failure. I also wonder why O/Ses don't try to protect themselves by running a memory tester on unallocated memory pages, maybe one minute per hour whenever the system is lightly loaded or idle. Nowhere as near as good as ECC, but better than doing nothing and hoping for the best.
Re Core 2 duo desktops
My last desktop "upgrade" was to complete silence, and in CPU terms was a downgrade possibly even from Core 2 Duo. SSD + fanless Celeron J1900 Mobo in mini-ITX case may well be the last desktop PC I ever need to buy, apart from replacements after hardware failure. The are good reasons for full-spec Core-i5/i7 systems, but none that make me want one at home.
I have a tablet as well, it does all I want it to (not very much!) And a £50 Kindle with paper display, which I find the most "magical" of my devices, and a perfect replacement for a daily paper made of mashed tree.
Actually there are a few primitive sorts of connectivity that are OK. One day I will get around to connecting my central heating to my home PC ... with a single signal to a relay, so that the PC can turn the heating on and off. With a manual override switch, for if/when the PC doesn't do what it's supposed to.
I get to come home to a warm house, and/or tell the computer not to waste heating when I'll be home late.
Someone hacks me, worst they can do is waste some fuel (and lay waste my PC which would be far worse).
A few ...? I'm having trouble thinking of any others that would actually be useful.
given all that ...
(a) software will still have remotely exploitable bugs
(b) the number of IOT systems will reduce to two or three (cf Apple, Android or MS for your mobile; MS or Linux for your desktop)
(c) So sooner or later the very bad guys will exploit a bug in a high percentage of the national /global IOT-enabled households to do something horrible. Like crash the nation's power grids. I hope that my hypothetical prankster gets to a lot of fridges first, to deliver a nonlethal warning by stink and lawyers that couldn't be ignored.
I have my doubts if it can be avoided.
I'm sure it can and will. I just hope that the number of fatalities is no higher than single digits before the penny drops. It'll probably be assassination by car-hacking that causes the public mood to change. If not that, the big stink when some joker turns off every domestic IOT-enabled refrigerator in the USA. Shortly after which a terrorist will turn on every kettle and heater in the USA all at once, and the power grids will crash.
Mars (or just about anywhere in the solar system) would be a lot easier with a nuclear-powered ion drive interplanetary stage. Assemble (fuel) it in Earth orbit, so there's never a critical mass or anything more radioactive than Uranium at risk to a rocket failure in Earth's atmosphere. Of course, you'd still need to lug along a conventionally powered lander.
Pity about the mindless politics concerning "nuclear".
There's going to be some reticence around loading up your multi-million dollar satellite on a slightly used rocket.
The same reticence there is about loading it on a rocket with a short track record, or one that's suffered a recent failure. Which will be balanced against the reduced launch fee for early flights or multiply-reused launchers. Market forces really do work!
Also one would hope that a bathtub curve might emerge, and that a second, third, ... launch might actually have a lower failure rate than the first one. But it's never been done before, and it'll take a good while to accumulate anything like useful statistics. So to start with it'll involve a lot of guesswork by all parties.
My view is that we should continue to oppose any changes to the law that make it legal to conduct surveillance on a greater scale. Once surveillance is legalized, it will be used by those in power against the people who were supposed to be protected.
On the other hand, I'm more relaxed with GCHQ etc. breaking the law on surveillance in order to keep us safe from serious evil-doers. That's how it always worked in the past. The intelligence agencies operate in a very grey zone where the secret breaking of small laws is justified by the seriousness of the illegality being contemplated by those they are supposed to be spying on. The fact that they are operating outside the law means that they can be stamped on by the courts if they start acting outside their remit. Which in turn means they will tell the police what they know about a terror group planning to blow up a shopping centre, but not what they have (hopefully accidentally) learned about a non-violent protest group planning to dump a ton of manure outside some errant council's offices, or a million other almost-harmless illegalities that reach their ears.
I wonder whether it is covered by this DMCA exemption
a person who has lawfully obtained use of a computer program accesses a particular portion of the program solely to identify and study elements of the program that are necessary for interoperability and that have not been previously available to him or her
IANAL but ... You've lawfully obtained use when you buy the car. You require inter-operability with yourself in order to drive the car. Where does it say inter-operability with another program excludes inter-operability with a human being? In any case, there must be a lawful use for inter-operating your car so that you can remotely control it from your laptop. I think I've just invented the remote-control demolition derby ....
I won't be buying any car without an air gap.
That's the air gap between the engine and the mechanically operated clutch. I'd far rather that there's also no possible control of the steering, gearbox or brakes by the computer. Failing which, they must at most be servo-assisted with mechanical override possible via the major controls, not "drive by wire" through a computer.
There's going to be a Ford Pinto moment for someone in the auto industry in the near future.
A bit negative aren't we? I wonder what the performance of the world's first germanium alloy-junction transistor was like back in Bell labs?
I find it hard to imagine that in a few decades, they'll be able to integrate maybe 20 to 200 billion of these on one small chip, along with all the wiring, and sell that chip for $100 or less. But on the other hand if you'd forseen the billion-transistor CPU back in 1960, few would have believed it was even a theoretical possibility, let alone reality in the late naughties.
I'd call advocating, threatening or committing crimes against humanity a completely non-stupid definition. Many of the other (wider) definitions are also not stupid, although failing to bear in mind that one man's terrorist is another man's freedom-fighter certainly *is* stupid.
"My masturbation practices performed in the privacy of my own home aren't fodder for blackmail."
Priest: embarassing. (Is it less or more of a sin than actual sex)?
Doctor: recommended by recent medical advice discussed here in the Reg. Reduces your likelyhood of getting prostate cancer, if you're not in a relationship.
School teacher: that's one heck of a leap isn't it? I guess there are places in the USA it could be a problem.
At least in the UK, marital infidelity is not a crime, and is very rarely grounds for being dismissed by an employer. Your main worry is that your spouse might refuse to believe your denials. That aside, being on that list should merely be an embarassment. Also our libel laws are the strictest in the world. Anyone publishing the allegation that you were a signed-up cheater when you weren't, could be in expensive trouble.
(It occurs to me that getting yourself hacked into that list might be a route to simultaneous divorce and profit for you and your spouse alike ...)
Cheaper than spinning rust? That's not true. Neither is it particularly cheap compared to other solid-state options.
Of course, there's "enterprise grade" snake-oil or wizardry included in that price. Without knowing a lot more about it, price/Gb comparisons are a bit pointless.
And I recently spotted via a serverfault link to another site, a worried intern asking what he should do because he thought he'd spotted a showstopper bug in a project very close to release date. He feared that telling management about it would be a career-limiting move.
Managers never learn, do they? (Except, how to justify ever-more-obscene salaries for themselves).
His parting words were "don't have any parties until it is all fixed".
Which is of course another reason for over-engineering buildings. The architects can't be sure that someone won't overload a floor way past design specifications, be it with 200 people, or a medium-sized library of mashed-tree books, or a large sculpture, or several tons of stolen silver bullion.
(That last item pushed the floor loading past all tolerances. The bullion spontaneously descended from the third floor to the basement, fortunately while the residents of the lower floors were away. The police were waiting when the crims came back to collect their loot. Note for crims: stash your loot around the *edges* of the room, or better still rent a basement flat.
In passing I was once told about the "jump test" by an old-school surveyor. You can get a pretty good idea of the overall soundness of a suspended timber floor under carpet etc., by jumping up and down on it. (Engineering explanation: field-expedient impulse response assessment). Just be sure that it's not *completely* rotten before you employ this test!
Redundancy is always a good idea, isn't it?
No.
Nature decided that redundant kidneys were a good idea, and put a lot of low-level redundancy (or over-capacity) into our livers and brains. But when it came to the big pump, we got just one of them.
There's also the joke about the right number of chronometers for a ship to navigate by. One is OK, if it's a good one. Three is overkill. Two is a very silly idea.
The DC10 cargo door problems surely came a close second to the Fort Pinto in terms of corporate misbehaviour in the face of an engineering defect that was killing people.
Those of you too young to have ever experienced a DC10 should also know:
It was designed with a 2-5-2 seating plan, rather than 3-3-3. I once had to fly the Atlantic in seat position 3/5. Thereafter, I asked the airline if the flight was a DC10 and if so, I found another flight or airline.
It was designed with overhead baggage racks too shallow to take any normal carry-on bag, so you had to put your bag under the seat in front of you with your feet jammed in on either side.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
