Creating a paper record of votes cast electronically is a bit like installing a floor drain to fix a leaking roof.
For one thing, there's always going to be a failure mode in which the paper record does not match the electronic record. That's impossible to do anything about -- you don't know, and can never know, which one is right.
One "proper" design for a vote recorder (assuming you aren't just going to use paper ballots, hand-counted under scrutiny just like people have been doing ever since democracy began) would use one record per possible way of completing a ballot paper, as opposed to one record per voter, to ensure anonymisation happens as soon as a vote is cast; and the recording system would be non-reversible (except possibly by a severe, obvious and deliberate action which would be prevented whilst the machine is in operation). It should also be Universally Comprehensible; that is, it should not employ any technology beyond the understanding of a school leaver with passing grades in all subjects.
I think that those criteria would be satisfied by a simple bank of interlocked, non-resettable, mechanical counters. The housing would be mostly transparent, so that the operation of the interlock mechanism could be seen. Each counter would have a large push-button adjacent to it, labelled for the candidate, and would normally be covered by a shutter. A partial depression of the button would retract the shutter, revealing the counter; and operate the interlock, preventing the depression of any other button. Further depression of the button would advance the counter under the view of the voter; return the shutter to the deployed position; and lock out all buttons until the machine is (remotely, using a Bowden cable) re-primed for another vote by the Presiding Officer or assistant.
The machines could be made available for public scrutiny almost until the commencement of polling. As each one is delivered to a polling station, the Presiding Officer would -- in the presence of representatives of all candidates -- record the present readings on each counter on a paper label on the inside of a hinged flap, underneath which would be a mechanism to retract all the counters' shutters simultaneously, and which would be sealed shut with an Official Seal. More Official Seals would also be used if necessary to lock-out any unused buttons.
At the close of polling the seals are given one final check, then cut open; and the previously-recorded readings are subtracted from the present readings on the counters (allowing for wraparound if necessary) to give the total vote for each candidate. This would be done in the presence of each candidate's representatives.
There is a way to determine partial results before the close of polling, but it would require two groups of as many people as the product of the number of candidates and the number of vote recording machines. Each member of the first group would in succession vote for a different candidate near the beginning; at a later stage each member of the second group would again in succession vote for a different candidate, and subtract their numbers from the first group's numbers. I don't think it's workable in practice.
Yes, but the point I was making is that it looks cheaper just to scrap Windows boxes when they "break" than to employ someone who knows how to fix them. Paying for a good Windows technician (who can actually do all the registry and scripting black magic) comes out of IT's budget. Paying for a brand new computer for Laura in Hay Char because the cheap "Windows technician" you hired doesn't actually know how to fix it properly comes out of HR's budget. The bigger the organisation, the more departments there are, and the thinner the cost of the incompetent Windows "technician" is spread. Since the top brass are computer-clueless (otherwise they'd never have been using Windows in the first place), this is generally accepted as Just The Normal State of Things.
The reason you can bluff your way around Windows is that nine failures out of ten are random, beyond anyone's control (which is fully to be expected when writing low-level software against a closed-source OS with inadequately, and at times downright wrongly, documented APIs). So rebooting will apparently "fix" nine problems out of ten. This doesn't mean that Windows is "easy to fix"; it means it needs fixing more often than it should.
When software is written against an OS with a properly-documented API -- and API documentation doesn't come much more proper than the real, live source code -- it's almost bound to be more stable. The only thing that can break it is a serious foul-up on the application developer's part. If the application is a popular one, developed as Open Source (and thus subject to many levels of independent scrutiny), or the application is Closed Source but the developer is plentifully-resourced (e.g. Sun, Oracle) then there is much less risk of random failures happening.
Fewer than one in ten failures on Unix / Linux systems are random, so nine times out of ten rebooting will not fix anything -- and a bluffer will quickly be found out. Fortunately, there is usually meaningful information in any error messages (which probably were put there by a real, live programmer for their own benefit during the testing phase) that may turn up; so it's possible for anyone with a logical mind, good reasoning ability and the merest smidgen of luck to work out from first principles what's up with the bloody thing. (Someone who has never been a serious Windows user is at a distinct advantage here, from not bringing any baggage of inapplicable preconceptions.) And if anyone else has ever had the same problem before, chances are they will have written it up on some message board, wiki or blog somewhere.
If they're running Windows 3.1, which we can guess is what came on them, they probably don't support network booting. Or maybe they do, but in order for that to work they will need to have a BIOS extension EPROM installed on the network cards (ever wonder what that 28-pin socket on all those old 16-bit network cards was for?) which almost certainly is no longer available; chances are the card maker went out of business and left no trace. Or maybe they still exist (perhaps in name only, having been swallowed up by a larger company) and are still vigorously defending their copyrights.
On the upside, a simple network-booting machine requires only a case and PSU, a modern all-in-one motherboard, a processor and 1GB of RAM -- no drives, CD, hard, floppy or flash; so they're cheap as chips to build (assuming you have a still-serviceable monitor, keyboard and mouse) and have almost nothing to go wrong with them (once you've got your head around setting up a TFTPboot server). You put / in a RAMdisk, and mount the important stuff like/usr,/bin,/lib and/etc read-only over the network.
But, as I've already mentioned, that's very un-Windows-like thinking, and it goes against what a lot of people have already been taught.
At least if it doesn't work out with SUSE, they can jump ship to another vendor. All they need to do is copy the whole of/etc over to whoever's distro they're using</oversimplification>, and it will pretty much work just the same as it did before.
Only Microsoft can sell you Windows.
That's the important difference. It's about the song, not the singer.
"For instance, existing Windows 3.1 public terminals used a program called Deepfreeze that rebooted the system at the end of each session, something that had to be re-engineered for Linux"
"Staff also found that the OS was storing information about the contents of public users' removable media, and for privacy purposes had to develop a script to delete this information"
Or just don't fit public terminals with HDDs -- boot them from CD, or read-only Flash drive, with all writable directories in RAMdisk.
You really do have to think about some things in a different way with Linux. Part of the problem is years of preconditioning to the way Windows has (arbitrarily) chosen to do everything blinding you to the alternatives.
Windows is prone to going Tango Uniform for no good reason, and nine times out of ten Windows cab ne fixed simply by rebooting. And you can train a monkey to reboot a Windows machine -- in fact, it wouldn't surprise me one bit if somebody somewhere has actually done this. This means Windows "techies" are cheap, because nine times out of ten they'll just reboot the errant machine (which the user could have done for themself, were they not scared absolutely shitless by the complexity of anything that plugs in and has more than three buttons on it) and it will work -- and the tenth time they'll reboot it a few times, mutter a few sotto voce expletives, realise it's not having it, give up and buy a new one. This means you end up scrapping repairable machines -- but of course, they ultimately come out of departments' own budgets, not IT's budget.
Unix-like systems don't usually fail without good reason. So anybody working on them really needs to know their arse from a hole in the ground. This means Unix techies are expensive -- because they're good. They have no choice but to be. And there's more transferrability of skills between software: much of what you might learn about Linux can be applied to Solaris and the BSDs, some of what you might learn about MySQL can be applied to PostgreSQL or Firebird, Perl is a bit like PHP, ProFTPD and Apache have similar configuration file syntaxes, and so forth.
Give a man a fish and he'll come back the next day for more fish -- and you can charge him money.
Teach a man to fish and you can sell him expensive, proprietary bait for the rest of his life.
I like SLAX as my general purpose boot-and-go CD, but I'm also getting to like DSL. That will boot fine on older, slower hardware. I find Knoppix is just too big and clumsy anymore {though we should not forget that DSL is based on a trimmed-down Knoppix}.
Some versions of TheOpenCD used to include a bootable, cut-down Ubuntu; but it seems as though they're now concentrating on providing Free software to run on Windows. Which I suppose is better than trying to spread themselves too thin.
Imagine a tax return form in docx format to download, fill in, submit. I'd need Word 2007 to do that. Would it be fair? Do you think this scenario is so far from reality?
No, it would not be fair. And no, I don't think it's at all far from reality.
What is basically happening, through the de facto requirement to use proprietary software, is that ability to do business with almost anyone is contingent upon the use of Microsoft software. That is already a sub-optimal situation with which we should think seriously about dealing. But when Governments start making individuals dependent upon Microsoft to read public documents and submit necessary forms, compliance with the Law of the Land will have been privatised. And that is absolutely unacceptable.
This is a very good point. The company where I work is based almost entirely on in-house-written web applications. The backend is a mix of perl and bash (yes!) CGI-scripting and PHP, with MySQL (when speed is important) and PostgreSQL (when anything else is important) databases. Everything is tailored to the workflow. Most people don't even need to use OpenOffice.org; templates get filled-in and faxed or e-mailed automatically.
We could write our own GTK frontends; but the Mozilla people have already done a fine job, so why reinvent the wheel?
Science is sometimes going to get in the way of politics, or vice-versa. It has been ever since Galileo. Sometimes, political desirability is at odds with scientific soundness.
I have no toaster. I make toast under the grill. Over the years, I have come to appreciate that gas toast tastes noticeably nicer than electric toast (and methane gas toast tastes nicer than propane gas toast).
Of course, if you use store-bought bread and/or artificial "I can't believe anybody thinks this tastes remotely like butter" spreads, you probably won't be able to appreciate the difference. Home-baked bread, toasted evenly on both sides under North Sea Gas and served at once with real dairy butter on a pre-warmed china plate, should be an absolute delight for the palate.
No, the metre has been defined since 1983 in terms of the speed of light (though it used to be defined in terms of the Earth's circumference, which was 40Mm. measured from the North Pole to the South Pole via Paris and back around the other side). The second nowadays is defined by the frequency of the radiation given off by caesium-133 being exactly 9 192 631 770 Hz, and the metre is defined by the speed of light (the previous definition, from 1960, was made in terms of the wavelength of light given off by Krypton-86).
The latest UK Weights and Measures Act demands that goods (except draught beer, which is still sold in 568ml pints) be sold in SI units. The SI measurement is legally binding, and measurements given in "supplementary units", while legally tolerated, need not bear relation to the official measurement.
Yes, but 25.4mm. is the old definition (although it would have been 25.6 if IBM had had their way, or 25.0 if HP had had their way). Today there is no official definition and an inch can be as much or as little as the product manufacturer desires. Which is why, whenever you buy a TV set nowadays, it has to state the screen size in (centi|milli)?metres, and you will often see things like "42 inches (102cm.)"; implying that the manufacturer is defining an inch to consist of 23.2mm. The metre is the official unit of distance, being 1/299 792 458 of the distance travelled through a vacuum by light in a second. As long as a beam of light took 3.40235377ns. to travel from one corner of the screen to the other, Trading Standards would be happy. The manufacturers could be so bold as to call it 51 inches if they liked, as long as the SI measurement was correct.
"Inch" is legally a slang term, without an official meaning in a court of law (unless the measurement is given primarily in official units, with an inch translation in brackets). So my TV set is smaller than 29 inches -- nas long as you consider an inch to be 36mm. or more. Problem solved!
Because the OLPC has lower power requirements, making it better suited to situations where electricity supplies are limited. If the lights dim when you turn on a few of those old clunkers (which will be fine, since they all have switched-mode power supplies and can run off anything from 160 to 300 volts, DC to 1kHz), or a substation fuse blows when you turn on more than one machine at once (those switched-mode supplies can draw tens of amps for a brief instant at power-up), then that might make you unpopular.
Not that it's an inherently bad idea to ship refurbished computers to some people. But the OLPC will be more useful in more situations than used kit.
What's stopping you from taking a year out to work with a programme where you will help the locals sort through the e-waste we're currently dumping in Africa to find any usable parts and assemble working computers (and probably other appliances) which could then be sold? All you'll need are a fine-tipped soldering iron, a digital storage oscilloscope, a known-working computer, a good set of tools, a generator and a few CDs of Open Source software. Be prepared to write the whole lot off if you don't make enough money to replace everything within the first year. You will also have to teach the locals how to do the work after you have gone home. It won't interfere too much with the OLPC project anyway, since OLPC's goals are different.
The whole idea is that once something is in the Public Domain, it should be in the Public Domain for good. The Public Domain is for everyone's benefit, not a resource to be plundered by a profiteering minority. After all, you can't make derivative works of Copyrighted works without permission of the copyright holder, which may well be contingent upon attribution. If you want to copyright something in your name, you should either create it from scratch or modify something else beyond all recognition (to all intents and purposes, the same thing).
Doesn't your written Constitution say something about ex post facto laws? Your copyright extension act is unconstitutional, and therefore Steamboat Willie is in the Public Domain.
This wouldn't affect existing recordings anyway even if it was enacted into law!
If you introduce a new law, it can't be applied to anything that was done before the law was introduced; and if you introduce a new punishment for an offence, anyone who was sentenced to the old punishment can't be given the new one. That's article 11 of the UN Universal Declaration of Human Rights. Article 7 of the European Convention on Human Rights makes similar provisions and is enshrined in UK law as the Human Rights Act 1998.
A copyright term extension could only be applied to brand-new copyrights. Existing copyrights would not be affected. The holders knew -- and presumably agreed to -- the terms when they created their creative works.
I'd like to see two options available under a reformed copyright system:
1. Non-exclusivity. Everyone is granted an implicit licence to reproduce, and to make derivative works, as long as they include proper attribution. Lasts for a term of 20 years, then the work enters the Public Domain. In the case of a computer program, a copy of the Source Code and Build Instructions must be placed in the National Archives (unless ordinarily distributed along with the program).
2. Full exclusivity. Five years from the date the first royalty payment is received, or five years from the date of publication if no royalties are ever paid (face it: if something hasn't made you any money in five years, it's never going to, so quit tying up the legal system with your pettiness and vanity). Renewable for one year at a time. The first year's extension costs 10% of the sum total of all royalties received so far. Each successive year's extension costs twice as much as the previous year's extension (probably exceeding the total amount of royalties received in a few years -- that's the whole idea). In the case of a computer program, a copy of the Source Code and Build Instructions must be placed in the National Archives (unless ordinarily distributed along with the program).
Also: all derivative works based upon a work in the Public Domain, should themselves automatically be in the Public Domain. (Alteration beyond recognition is not a derivative work, but a new work in its own right.)
It's my experience that, unless your fridge is really knackered, food will survive a good week or so beyond the manufacturer's date stamp. And some foods -- the versions of French cheeses that you get in British supermarkets spring to mind -- aren't edible until that date!
Milk is interesting. It goes through a stage where you can taste that it's just starting to go on the turn but it's fine in tea; then a bit later it's no good in tea but OK in coffee. Then it starts to separate into watery and fatty parts. How's a fridge going to know all the subtle stages?
I've even seen expiry dates on eggs! Surely everyone knows how to tell whether an egg is fresh; you place it in a jug of water. If it sinks, it's safe; if it floats, it's foul; and if it stands upright, use it up right away. Also, I think sticking a printer up a chicken's bottom is just cruel.
That's the thing with these computers..... if it knows there's a missing (round bracket|speech mark|posh bracket), why doesn't it just put the bloody thing in for you and be done complaining? Obviously they're not that smart (you can get a bit of a clue from indentation as to where a posh bracket might be missing from).
It's like the way GCC gives you a warning if you omit #include <stdio.h> but then saves you from the worst by automagically including it for you. Well, why can't it go just one teeny step further and -- if the source file is writable and not just because you're being root -- rewrite it with the missing line included, so the next poor sod who compiles it doesn't get a screen full of warnings?
Slashdot Mirror
Creating a paper record of votes cast electronically is a bit like installing a floor drain to fix a leaking roof.
For one thing, there's always going to be a failure mode in which the paper record does not match the electronic record. That's impossible to do anything about -- you don't know, and can never know, which one is right.
One "proper" design for a vote recorder (assuming you aren't just going to use paper ballots, hand-counted under scrutiny just like people have been doing ever since democracy began) would use one record per possible way of completing a ballot paper, as opposed to one record per voter, to ensure anonymisation happens as soon as a vote is cast; and the recording system would be non-reversible (except possibly by a severe, obvious and deliberate action which would be prevented whilst the machine is in operation). It should also be Universally Comprehensible; that is, it should not employ any technology beyond the understanding of a school leaver with passing grades in all subjects.
I think that those criteria would be satisfied by a simple bank of interlocked, non-resettable, mechanical counters. The housing would be mostly transparent, so that the operation of the interlock mechanism could be seen. Each counter would have a large push-button adjacent to it, labelled for the candidate, and would normally be covered by a shutter. A partial depression of the button would retract the shutter, revealing the counter; and operate the interlock, preventing the depression of any other button. Further depression of the button would advance the counter under the view of the voter; return the shutter to the deployed position; and lock out all buttons until the machine is (remotely, using a Bowden cable) re-primed for another vote by the Presiding Officer or assistant.
The machines could be made available for public scrutiny almost until the commencement of polling. As each one is delivered to a polling station, the Presiding Officer would -- in the presence of representatives of all candidates -- record the present readings on each counter on a paper label on the inside of a hinged flap, underneath which would be a mechanism to retract all the counters' shutters simultaneously, and which would be sealed shut with an Official Seal. More Official Seals would also be used if necessary to lock-out any unused buttons.
At the close of polling the seals are given one final check, then cut open; and the previously-recorded readings are subtracted from the present readings on the counters (allowing for wraparound if necessary) to give the total vote for each candidate. This would be done in the presence of each candidate's representatives.
There is a way to determine partial results before the close of polling, but it would require two groups of as many people as the product of the number of candidates and the number of vote recording machines. Each member of the first group would in succession vote for a different candidate near the beginning; at a later stage each member of the second group would again in succession vote for a different candidate, and subtract their numbers from the first group's numbers. I don't think it's workable in practice.
Yes, but the point I was making is that it looks cheaper just to scrap Windows boxes when they "break" than to employ someone who knows how to fix them. Paying for a good Windows technician (who can actually do all the registry and scripting black magic) comes out of IT's budget. Paying for a brand new computer for Laura in Hay Char because the cheap "Windows technician" you hired doesn't actually know how to fix it properly comes out of HR's budget. The bigger the organisation, the more departments there are, and the thinner the cost of the incompetent Windows "technician" is spread. Since the top brass are computer-clueless (otherwise they'd never have been using Windows in the first place), this is generally accepted as Just The Normal State of Things.
The reason you can bluff your way around Windows is that nine failures out of ten are random, beyond anyone's control (which is fully to be expected when writing low-level software against a closed-source OS with inadequately, and at times downright wrongly, documented APIs). So rebooting will apparently "fix" nine problems out of ten. This doesn't mean that Windows is "easy to fix"; it means it needs fixing more often than it should.
When software is written against an OS with a properly-documented API -- and API documentation doesn't come much more proper than the real, live source code -- it's almost bound to be more stable. The only thing that can break it is a serious foul-up on the application developer's part. If the application is a popular one, developed as Open Source (and thus subject to many levels of independent scrutiny), or the application is Closed Source but the developer is plentifully-resourced (e.g. Sun, Oracle) then there is much less risk of random failures happening.
Fewer than one in ten failures on Unix / Linux systems are random, so nine times out of ten rebooting will not fix anything -- and a bluffer will quickly be found out. Fortunately, there is usually meaningful information in any error messages (which probably were put there by a real, live programmer for their own benefit during the testing phase) that may turn up; so it's possible for anyone with a logical mind, good reasoning ability and the merest smidgen of luck to work out from first principles what's up with the bloody thing. (Someone who has never been a serious Windows user is at a distinct advantage here, from not bringing any baggage of inapplicable preconceptions.) And if anyone else has ever had the same problem before, chances are they will have written it up on some message board, wiki or blog somewhere.
If they're running Windows 3.1, which we can guess is what came on them, they probably don't support network booting. Or maybe they do, but in order for that to work they will need to have a BIOS extension EPROM installed on the network cards (ever wonder what that 28-pin socket on all those old 16-bit network cards was for?) which almost certainly is no longer available; chances are the card maker went out of business and left no trace. Or maybe they still exist (perhaps in name only, having been swallowed up by a larger company) and are still vigorously defending their copyrights.
/usr, /bin, /lib and /etc read-only over the network.
On the upside, a simple network-booting machine requires only a case and PSU, a modern all-in-one motherboard, a processor and 1GB of RAM -- no drives, CD, hard, floppy or flash; so they're cheap as chips to build (assuming you have a still-serviceable monitor, keyboard and mouse) and have almost nothing to go wrong with them (once you've got your head around setting up a TFTPboot server). You put / in a RAMdisk, and mount the important stuff like
But, as I've already mentioned, that's very un-Windows-like thinking, and it goes against what a lot of people have already been taught.
At least if it doesn't work out with SUSE, they can jump ship to another vendor. All they need to do is copy the whole of /etc over to whoever's distro they're using</oversimplification>, and it will pretty much work just the same as it did before.
Only Microsoft can sell you Windows.
That's the important difference. It's about the song, not the singer.
You really do have to think about some things in a different way with Linux. Part of the problem is years of preconditioning to the way Windows has (arbitrarily) chosen to do everything blinding you to the alternatives.
Windows is prone to going Tango Uniform for no good reason, and nine times out of ten Windows cab ne fixed simply by rebooting. And you can train a monkey to reboot a Windows machine -- in fact, it wouldn't surprise me one bit if somebody somewhere has actually done this. This means Windows "techies" are cheap, because nine times out of ten they'll just reboot the errant machine (which the user could have done for themself, were they not scared absolutely shitless by the complexity of anything that plugs in and has more than three buttons on it) and it will work -- and the tenth time they'll reboot it a few times, mutter a few sotto voce expletives, realise it's not having it, give up and buy a new one. This means you end up scrapping repairable machines -- but of course, they ultimately come out of departments' own budgets, not IT's budget.
Unix-like systems don't usually fail without good reason. So anybody working on them really needs to know their arse from a hole in the ground. This means Unix techies are expensive -- because they're good. They have no choice but to be. And there's more transferrability of skills between software: much of what you might learn about Linux can be applied to Solaris and the BSDs, some of what you might learn about MySQL can be applied to PostgreSQL or Firebird, Perl is a bit like PHP, ProFTPD and Apache have similar configuration file syntaxes, and so forth.
Basically, if you pay peanuts, you get monkeys.
Give a man a fish and he'll come back the next day for more fish -- and you can charge him money.
Teach a man to fish and you can sell him expensive, proprietary bait for the rest of his life.
I like SLAX as my general purpose boot-and-go CD, but I'm also getting to like DSL. That will boot fine on older, slower hardware. I find Knoppix is just too big and clumsy anymore {though we should not forget that DSL is based on a trimmed-down Knoppix}.
Some versions of TheOpenCD used to include a bootable, cut-down Ubuntu; but it seems as though they're now concentrating on providing Free software to run on Windows. Which I suppose is better than trying to spread themselves too thin.
Not on my PC they don't. I've set up Squid to block all the major advertising providers.
There's actually more chance of me buying a product if I haven't seen it advertised, so don't complain.
What is basically happening, through the de facto requirement to use proprietary software, is that ability to do business with almost anyone is contingent upon the use of Microsoft software. That is already a sub-optimal situation with which we should think seriously about dealing. But when Governments start making individuals dependent upon Microsoft to read public documents and submit necessary forms, compliance with the Law of the Land will have been privatised. And that is absolutely unacceptable.
This is a very good point. The company where I work is based almost entirely on in-house-written web applications. The backend is a mix of perl and bash (yes!) CGI-scripting and PHP, with MySQL (when speed is important) and PostgreSQL (when anything else is important) databases. Everything is tailored to the workflow. Most people don't even need to use OpenOffice.org; templates get filled-in and faxed or e-mailed automatically.
We could write our own GTK frontends; but the Mozilla people have already done a fine job, so why reinvent the wheel?
Science is sometimes going to get in the way of politics, or vice-versa. It has been ever since Galileo. Sometimes, political desirability is at odds with scientific soundness.
Science usually wins in the end, though.
Perhaps the toxicity of the useful chemicals is evidence that God is just a figment of your imagination?
There's little immediate evolutionary benefit in immunity to rare and/or sequestred materials, as compared with more commonly-encountered ones.
I have no toaster. I make toast under the grill. Over the years, I have come to appreciate that gas toast tastes noticeably nicer than electric toast (and methane gas toast tastes nicer than propane gas toast).
Of course, if you use store-bought bread and/or artificial "I can't believe anybody thinks this tastes remotely like butter" spreads, you probably won't be able to appreciate the difference. Home-baked bread, toasted evenly on both sides under North Sea Gas and served at once with real dairy butter on a pre-warmed china plate, should be an absolute delight for the palate.
Or am I just a food snob?
No, the metre has been defined since 1983 in terms of the speed of light (though it used to be defined in terms of the Earth's circumference, which was 40Mm. measured from the North Pole to the South Pole via Paris and back around the other side). The second nowadays is defined by the frequency of the radiation given off by caesium-133 being exactly 9 192 631 770 Hz, and the metre is defined by the speed of light (the previous definition, from 1960, was made in terms of the wavelength of light given off by Krypton-86).
The latest UK Weights and Measures Act demands that goods (except draught beer, which is still sold in 568ml pints) be sold in SI units. The SI measurement is legally binding, and measurements given in "supplementary units", while legally tolerated, need not bear relation to the official measurement.
Yes, but 25.4mm. is the old definition (although it would have been 25.6 if IBM had had their way, or 25.0 if HP had had their way). Today there is no official definition and an inch can be as much or as little as the product manufacturer desires. Which is why, whenever you buy a TV set nowadays, it has to state the screen size in (centi|milli)?metres, and you will often see things like "42 inches (102cm.)"; implying that the manufacturer is defining an inch to consist of 23.2mm. The metre is the official unit of distance, being 1/299 792 458 of the distance travelled through a vacuum by light in a second. As long as a beam of light took 3.40235377ns. to travel from one corner of the screen to the other, Trading Standards would be happy. The manufacturers could be so bold as to call it 51 inches if they liked, as long as the SI measurement was correct.
"Inch" is legally a slang term, without an official meaning in a court of law (unless the measurement is given primarily in official units, with an inch translation in brackets). So my TV set is smaller than 29 inches -- nas long as you consider an inch to be 36mm. or more. Problem solved!
And which countries, do you suppose, they are going to buy those guns from?
Because the OLPC has lower power requirements, making it better suited to situations where electricity supplies are limited. If the lights dim when you turn on a few of those old clunkers (which will be fine, since they all have switched-mode power supplies and can run off anything from 160 to 300 volts, DC to 1kHz), or a substation fuse blows when you turn on more than one machine at once (those switched-mode supplies can draw tens of amps for a brief instant at power-up), then that might make you unpopular.
Not that it's an inherently bad idea to ship refurbished computers to some people. But the OLPC will be more useful in more situations than used kit.
What's stopping you from taking a year out to work with a programme where you will help the locals sort through the e-waste we're currently dumping in Africa to find any usable parts and assemble working computers (and probably other appliances) which could then be sold? All you'll need are a fine-tipped soldering iron, a digital storage oscilloscope, a known-working computer, a good set of tools, a generator and a few CDs of Open Source software. Be prepared to write the whole lot off if you don't make enough money to replace everything within the first year. You will also have to teach the locals how to do the work after you have gone home. It won't interfere too much with the OLPC project anyway, since OLPC's goals are different.
The whole idea is that once something is in the Public Domain, it should be in the Public Domain for good. The Public Domain is for everyone's benefit, not a resource to be plundered by a profiteering minority. After all, you can't make derivative works of Copyrighted works without permission of the copyright holder, which may well be contingent upon attribution. If you want to copyright something in your name, you should either create it from scratch or modify something else beyond all recognition (to all intents and purposes, the same thing).
Doesn't your written Constitution say something about ex post facto laws? Your copyright extension act is unconstitutional, and therefore Steamboat Willie is in the Public Domain.
This wouldn't affect existing recordings anyway even if it was enacted into law!
If you introduce a new law, it can't be applied to anything that was done before the law was introduced; and if you introduce a new punishment for an offence, anyone who was sentenced to the old punishment can't be given the new one. That's article 11 of the UN Universal Declaration of Human Rights. Article 7 of the European Convention on Human Rights makes similar provisions and is enshrined in UK law as the Human Rights Act 1998.
A copyright term extension could only be applied to brand-new copyrights. Existing copyrights would not be affected. The holders knew -- and presumably agreed to -- the terms when they created their creative works.
I'd like to see two options available under a reformed copyright system:
1. Non-exclusivity. Everyone is granted an implicit licence to reproduce, and to make derivative works, as long as they include proper attribution. Lasts for a term of 20 years, then the work enters the Public Domain. In the case of a computer program, a copy of the Source Code and Build Instructions must be placed in the National Archives (unless ordinarily distributed along with the program).
2. Full exclusivity. Five years from the date the first royalty payment is received, or five years from the date of publication if no royalties are ever paid (face it: if something hasn't made you any money in five years, it's never going to, so quit tying up the legal system with your pettiness and vanity). Renewable for one year at a time. The first year's extension costs 10% of the sum total of all royalties received so far. Each successive year's extension costs twice as much as the previous year's extension (probably exceeding the total amount of royalties received in a few years -- that's the whole idea). In the case of a computer program, a copy of the Source Code and Build Instructions must be placed in the National Archives (unless ordinarily distributed along with the program).
Also: all derivative works based upon a work in the Public Domain, should themselves automatically be in the Public Domain. (Alteration beyond recognition is not a derivative work, but a new work in its own right.)
It's my experience that, unless your fridge is really knackered, food will survive a good week or so beyond the manufacturer's date stamp. And some foods -- the versions of French cheeses that you get in British supermarkets spring to mind -- aren't edible until that date!
Milk is interesting. It goes through a stage where you can taste that it's just starting to go on the turn but it's fine in tea; then a bit later it's no good in tea but OK in coffee. Then it starts to separate into watery and fatty parts. How's a fridge going to know all the subtle stages?
I've even seen expiry dates on eggs! Surely everyone knows how to tell whether an egg is fresh; you place it in a jug of water. If it sinks, it's safe; if it floats, it's foul; and if it stands upright, use it up right away. Also, I think sticking a printer up a chicken's bottom is just cruel.
That's the thing with these computers ..... if it knows there's a missing (round bracket|speech mark|posh bracket), why doesn't it just put the bloody thing in for you and be done complaining? Obviously they're not that smart (you can get a bit of a clue from indentation as to where a posh bracket might be missing from).
It's like the way GCC gives you a warning if you omit #include <stdio.h> but then saves you from the worst by automagically including it for you. Well, why can't it go just one teeny step further and -- if the source file is writable and not just because you're being root -- rewrite it with the missing line included, so the next poor sod who compiles it doesn't get a screen full of warnings?