I learned two rules for avoiding charges of cheating in my CS department, both of which strike me as pretty sound:
1 - You can't be punished for citing references.
If you get code from some source and you document it, you say so when you turn it in, it isn't cheating. You may not get a very good grade and your prof may not be pleased, but you can't be brought up on charges of cheating.
2 - The Gilligan's Island rule
If you look at someone else's code, then go watch an episode of Gilligan's Island. Anything you can still remember afterwards is fair use. This was treated as a way of defining the line between copying someone else's code and learning from it. As rule, it won't save you in a court of law, but unless you have superhuman memory, you will be hard pressed to remember enough detail about someone else's work to be demonstrably cheating from it.
I have never heard of a student using these policies ever having been charged with academic improprieties. They are much, much easier to follow than the rules for when you can and can't sleep with profs and TA's.
They don't seem to have very accurate speech recognition technology. The article claims to reduce transcription time by a factor of about nine. That's a lot less unreasonable than believing in good speech recognition technology.
My guess is that it's really fairly poor speaker independent stuff. It probably does a quick, low quality word recognition algorithm - quite a few of those are around - and then some sort of Bayesian network to correct the transcription using lexical context. I know that ARPA was openly funding people doing exactly that a few years ago, and I'll bet their papers are on the web. It doesn't shock me greatly that someone has had some measure of success with it.
If it was 100% accurate transcription, then I wouldn't believe it. But as a time saving device for transcribers... that I find credible.
DARPA also funds a lot of automatic topic spotting research. One of my ex-profs received grants from them under just such a rubric and her papers are publicly available on the web. I'll bet whatever technology they are using, it was developed by a prof at an open university who publishes freely.
As for multilingual text searching and summarisation, the best technology of its kind known to me is Latent Semantic Analysis - the brain child of Thomas Landauer. It's a fairly recent, but hardly secret or obscure, indexing technique that's gaining ground commercially for data mining applications. It can certainly do the the small number of things being claimed by this article. All the relevant papers are on the web.
In short, this doesn't sound like super-secret spy stuff. I'll give long odds the real work is in journals and webpages that are publicly available. Having a couple billion dollars to speed up testing and implementation probably helps, but none of this sounds revolutionary or years ahead of the curve.
Here's a few related books I'd like to see reviewed here:
Manuel Castells' The Information Age: Economy, Society, and Culture
Geoffrey Hodgson's Economics and Utopia: Why the Learning Economy is Not the End of History
Paul Ormerod's Butterfly Economics
Bruce Sterling's Zeitgeist
Bonnie Nardi and Vicki O'Day's Information Ecologies: Using Technology with Heart or Context and Consciousness: Activity Theory and Human-Computer Interaction edited by Nardi.
These kinds of books are a lot more relevant to higher geek culture than the latest Python book, and I'm really glad to see some of this kind of thing on/. these days. Since I've only read two of those and have a long reading list of other stuff before I get to the others, I'd really like to see someone else here take a few of them on.
I could cite Kuhn here and just blow this idea away, but since I'm usually the one on the baracades railing away against Kuhn abuse in the social sciences, it'd look bad for me.
So, let me take an example. Since the early 19th century, it was known that the planet Merucry did not orbit the sun in the manner Newtonian mechanics prescribes for it. So, here's my question. Did this 'disprove' Newtonian mechanics? If so, were the scientists of the time right or wrong in sticking with Newtonian mechanics until the 1920's? If they were wrong, what programme do you suggest the physicists of the 19th century should have followed that would have produced the same massive advancement in human knowledge and power?
Let me add another example from the social sciences: Child development theory takes as a given that children under the age of 5 are not able to judge whether two quantities are of different magnitude by counting them. This is the most consistent result found in child development acccording to authorities like Gallistel and Gelman and a wide variety of child development theories rest on this result among others. They include Piaget's theory of general development and more nativist theories like Pinker's, which seek to show that counting (as opposed to other behaviours like language) are not the product of biologically driven forces.
Now, I can assure you (although you need not take my word for it in order to use this example, just assume I'm telling the truth) that my little brother could make set comparisons by counting at the age of three years and two months and was able to do so for sets of up 70 items. My brother is a particularly gifted mathematician (a strange form of mental illness that explains why he's still unemployed.) Does this fact render null all of the child development research done in the last 50 years? Are we now compelled to say that no one knows anything about child development because a single child exists who defies the empirical results on which those theories are based?
Alternatively, can we view existing child development theories as incomplete but still viable bodies of thought?
Now, let me propose an alternative version of what a theory is. Theories are tools which mediate human interaction with the world. They are semiotic tools, rather than physical tools, but they work in much the same way.
Humans behave in goal-directed ways. For example, when you want to build a house, you have a goal: to have a house. To do this, you must interact with other bits of the universe: land, wood, nails, etc. To do this, you use tools. Furthermore, the kind of house you make - the structure, the composition, the design, even the uses - are in part determined by the tools you have on hand. With a cheap nails and a strong hammer, you build a very different kind of house than the way pre-industrialised people build their homes. (Go look at old homes in Europe or colonial era dwellings on the East Coast.) In fact, you see the problem of building a house very differently with modern tools than you do with other tools.
Furthermore, you judge one tool to be better than another tool by using it. If you buy a nail gun, it is because it makes it easier to build houses. If the nail gun was too heavy or bulky or was constatantly breaking and you couldn't depend on it, you would go back to using the old manual hammer and nails. In fact, the very existence of nail guns is predicated on people having certain tools, like automated, precision nail-making machines so that nails are uniform. Even tools are the products of tools.
And if you find something you can't build because you don't have the right tools for it, you don't abandon your tools and go back to making things with your bare hands.
Friedrich Dürrenmatt and Sean Penn
on
The Pledge
·
· Score: 2
I'm a fan of Dürrenmatt, and I have been even since I was introduced to his still - as far as I know - untranslated play Die Wiedertäufer. His best known play in English is Die Physiker, about a group of residents of an insane asylum who believe that they are various famous physicists.
Dürrenmatt is exactly the kind of thing I would never have expected to see on/. Among anglophones, only snobby, elite literati types are likely ever to have even heard of him, much less ever have read his plays. I applaud Katz for doing something as audaciously anti-social as bringing up literary culture in/.'s technofetishist discourse and I await his analysis of Brecht's Leben des Galilei or perhaps Borges' La Biblioteca de Babel.
However, what really blows me away is that Sean Penn would read Dürrenmatt, and that by all accounts he seems to have made a decent art house film of it. Wasn't he the guy who married Madonna and got a reputation for beating people up? I seem to recall him as the butt of jokes back in the 80's as the archtype of the Hollywood ruffian actor with poor impulse control.
I guess this just goes to show that everybody can grow up.
Before he was writer, Heinlein ran for the California assembly on Upton Sinclair's EPIC ticket. EPIC was strongly socialist, with links to radicals, communists, unions, and anti-poverty groups. Heinlein was considered such a radical leftist at the time that the Republican candidate was able to cross-file as a Democrat and win their primaries, leaving him unchallenged on the ballot.
Heinlein certainly takes a strong stance against any kind of interference in markets in his later books, but the positions of EPIC were diametrically opposed to those he espoused in Expanded Universe. They were interventionist, they demanded government action against the Depression and they had nothing but contempt for California's wealthy industrialists and land owners who did nothing in the face of such vast public misery.
Certainly, Heinlein's politics changed a lot between 1938 and 1980. Frankly, I liked the younger Heinlein a lot more. The authoritarian, Chicago-school, Reaganite Heinlein of the 70's and 80's was half the writter of the witty, liberal, socially and culturally conscious Heinlein of the 40's and 50's.
The late Heinlein was a far cry from any anarcho-syndicalist that I know of, although the young one had his moments. He takes a near Randian position on the virtues of the market in all his books after Time Enough for Love (for example, his utopian portrayal of Hell in Job), while taking a truly unlibertarian position on the importance of a powerful central state in Friday. The later Heinlein appears to place little value in democracy and collective action (which are perhaps the most central values of traditional anarcho-syndicalism) and advocates a sort of minimalist dictatorship as the ideal form of government, or at least he does in all of the Lazarus Long books.
Perhaps those weren't his "true" politcal feelings. I have no way to know, authors are allowed to play with ideas. But at the very least, the opinions he lays out in Expanded Universe, a polemic by his own admission, are quite remote from anarcho-syndicalism and even more remote from his political roots.
Robotron manufactured most of the old Warsaw Pact's computer technology. They built a lot of PDP clones and some more original designs, but they relied a lot on reappropriated Western technology.
Robotron produced an 8-bit chip called the U880 that was basically a copy of the ZX80 chip until reunification in 1990. It was generally believed that the chip design had been stolen from its British manufacturer and reimplimented unchanged.
The KC-85 and KC-87 were the only things that could be called a "personal computer" in the East. It had 64k of RAM and could run CP/M. It was roughly equal to early 80's Commodores. It also had a printer, a disk drive and some other peripherals.
A more interesting factoid: Vladimir Putin was directly invovled with Robotron's operations in the 80's. He was the KGB liason in Dresden and Leipzig who controlled Soviet technology transfers (or to put it another way, he coordinated stealing Western computer designs.)
Half of eastern Germany's senior computer scientists seem to have worked there at one time or another.
Robotron is still around, although it is a shadow of its former self. Nowadays its a database services company. They are now "Robotron Datenbank-Software GmbH" and they have a website at http://www.robotron.de/.
There are still hobbyists playing with old eastern block computers. For the German-compatible, try: http://www.robotron-net.de/, the KC-Club at http://www.iee.et.tu-dresden.de/~kc-club/, and http://robotron.informatik.hu-berlin.de/.
According to the artivle, they're working on a substitution scheme so ASCII only users can still type in the URL's. Does this mean that ASCII equivalets will be arbitrary and unintuitive? If so, that's a problem. Let me propose something slightly different:
Unicode is not supposed to over-unify characters, so the ASCII fallback for Japanese could be the romanji transcription - and therefor registering a Japanese domain name automatically registers the romanji equivalent, except that some kanji have more than one possible romanji transcription.
However, some kanji are unified with Chinese characters, which have a different pinyin trasncription.
Chinese is another problem. The logical ASCII equivalent is pinyin stripped of its diacritical marks. But then, many different characters may have the same transcription.
All Cyrillic languages also have an ASCII trasncription scheme too, but it isn't unified. One character may be trasncribed one way in Russian and another way in Bulgarian. Is there a unified transcription scheme for all Cyrillic languages, and is it truely one-to-one? I don't think so. Look at the character usually transcribed as "j" in Russian, and the one usually transcribed that way in Serbian.
ISO-Latin-1 and -2 fallbacks: For ISO-Latin-1, the fallbacks are pretty obvious: "Champs-Élysée" ==> "Champs-Elysee" or in German "Düsseldorf" ==> "Duesseldorf", but in Czech it's a little less obvious. Does "C hacek" map to "Cz" or "Ch" or "Cs"?
So, here is a possible solution: devise unified ASCII transcritption schemes for each language, admitting whatever ambiguities exist in Japanese or similar languages. Then, when you register a non-ASCII name, you are asked on the form to fill out the transcribed ASCII name that corresponds to it and it is also automatically registered to you.
There is some potential for conflict here, if the ASCII transcription corresponds to an existing registered domain or, as in the case of Chinese more than one foreign name corresponds to the same transcription, but I think the problem is manageable.
I want to be able to register domain names in French, German and Russian too. If they are going to support all three zillion kanji and Chinese characters, they need to at least support the various Cyrillic and eastern European Roman alphabets, and the rest of ISO-Latin-1 (which covers all the major and most of the minor Western European languages.) The Persian-based alphabets (Arabic, Farsi, Urdu, etc), Hebrew and Thai are written right-to-left, so I suppose that won't be implimented right away, but it needs to be on the drawing board.
If all those other languages are accounted for, I view this as a good thing. If this is part of an overall shift to Unicode on the web, then all these languages are automatically supported, and I would think it an even better thing.
Electronic voting will never be fully trusted. Much of the confusion in the current election is simply because computers tabulate paper ballots.
Do the opposite. Go to paper balloting with big print and boxes that have to be marked with a big 'X'. Then, these ballots need to be counted by hand in each ward. This is what is done in the overwhelming majority of countries.
An even better reform: stop holding elections for everything on the same day. It's not genuinely convenient to anyone. Compel state and local government to hold elections on a day other than the second Tuesday in November of even numbered years. This would shrink the size of the ballot enough to make paper balloting and manual counting easier. In Federal elections, there would never be more than 3 offices to vote for: a Presdient, a Senator and a Rep.
Even better, have three different election days, all at different times during the 2-year election cycle. One for the feds, one for states, one for local government. That way, there are only three things to vote for in Federal election, three in state election (except for Nebraska) plus state referenda, if any, and one day for local government, which usually means one or two candidates in county races, one or two in municipal races, a school board election, in some places a hospital and/or public transit board election plus county and municipal referenda.
Furthermore, make the FEC final arbiter of all elections. Take local government out of the process of deciding on voting methods. I think this would minimise corruption rather than make it more likely.
And, if you really want to bring American voting into the modern world, use Condorcet voting and/or proportional representation.
Here would be my reform - if I had the dictatorial power to impose it:
1) Austrailian-style manditory voting. No more griping about people who didn't register, or registered but didn't vote. It costs more, but it's worth it.
2) A paid half day off on election day. Give everyone a chance to get to the polls.
3) Condorcet voting for Presdents, Senators and Governors.
4) Allocate seats in the House to each state rather than drawing districts. If a state has only one House Rep, use Condorcet voting. If it has two, divide the state into two electoral districts and use Condorcet voting. For more than that, use party proportional representation to allocate seats, but also guarantee that any party or independent that gets at least the fraction of votes in the state equal to the number of votes divided by the number of seats in the House gets one seat.
That way, all Reps still represent a state, rather than being 'at-large' national Reps as the Germans have, but the number of seats in the House is still apportioned more reasonably according to voters demands.
5) Move all states to unicameral legislatures like Nebraska. There is no need for state government to replicate the silliness of the Federal government. This way, state elections are for a Governor, one Rep and whatever referenda are going on, and judges in those states where state judges are elected. Also, make state legislatures mixed district/proportional voting on the German model. States are small enough to support 'at-large' representation.
6) Elect a single board for county government by at-large voting for multiple candidates. This means your ballot lists all the qualified candidates and asks you to vote for as many as their are seats on the county board. County supervisors are chosen by the elected board.
7) Do the same for school boards, hospital boards and public transit boards, where such things are elected.
8) Do the same for municipal governments, unless they are elected on the "New Plan", where city commissioners are elected instead of appointed by the municipal government and there is no city council. In that case, go back to Condorcet voting.
9) Stop electing every damn office under the sun, especially judges!!! Elections for judges force judges to be biased. It is a travesty of law to do it this way. In California, we elect offices like state treasurer and insurance commissioner and this is stupid. These offices were less corrupt when they were appointed. I haven't seen anything brilliant come out of elected hospital boards or public transit boards either, and the first thing I would do to reform education is get rid of local school boards.
These reforms would bring the US in line with - in fact ahead of - most other countries in terms of sane, modern, reliable, unambiguous voting systems.
Canada is a monarchy. Mexico is a federation. China is a "People's Republic." They're just names, and have little or nothing to do with democracy.
There are huge differences between different schools of thought on what exactly constitutes a fair vote and a fair campaign. Most countries that run regular multi-party elections limit candiadtes access to funds and media, and as far as I know, all of them have some centre of power not elected by simple one-man-one-vote formulas.
In Canada and the UK, leadership is decided based on a first-past-the-post system in each constituency, where the party with the most seats in parliament leads. This can - and in fact usually does - lead to representation out of proportion ot the actual vote count. In France, in addition to a British-style parliament, there is a Senate with limited powers elected by methods that heavily favour conservative rural districts over liberal urban ones. Canada has an appointed Senate, and the UK has the House of Lords. Germany has proportional representation in the Bundestag. The UK has regional governments in Scotland, Wales and Northern Ireland that distort power relationships to favour those regions.
Does any of this disqualify them as democracies? Not by any sensible current definition.
As for me, I think that Condorcet voting is the fairest way to handle single office elections, and proportional representation is the only realistically fair way to handle electing legislative assemblies. I understand that political realities in most nations require some from of disproportionate regional representation, and I'm for them wherever truly necessary, but not otherwise.
I don't think the US needs the electoral college anymore, not in an age of rightly diminished state power and loyalty. As for this election, it seems to me that considering the Nader vote, the majority of the American electorate has expressed a liberal tendency and that Gore's mandate is better than Bushes. Condorcet voting would almost certainly have elected Gore, although still not by any huge margin.
It's mostly the same as generic mozzarella - raw farmer's cheese with the standard set of industrial additives and conservatives, uncoloured and prepared in whatever manner most cost effective.
Poutine cheese is not for cheese conoisseurs. It's cheap and nasty.
Look, people eat bacon on hamburgers with cheese because this paragon of unkosher cholesterol nightmares tastes damn good. Our bodies are programmed to love fat, and poutine delivers. Trust me, if you love bacon cheeseburgers, you'll love poutine.
Now, having said that, poutine is probably worse for your health than smoking and ought to come with a label from Health Canada just like Export A's. You can gain weight just watching someone else eat it.
But, after a night of hard drinking and pot smoking (did I inhale? Hell, yes!), nothing hits the spot like hitting Chez Lafleur at the corner of rue St-Denis and Carreé St-Louis (in Montreal's Plateau district next to St-Denis Metro - you can't miss it, it's open 24/7) and ordering a big ol' poutine.
I've never been asked about past criminal activity on crossing the border. However, my mother-in-law looks as poor as she is, and her paperwork was a bit of a mess. We had paid for her airline ticket so she could come to our wedding. I suspect that Canada Customs uses some form of profiling, either formally or informally, in deciding who gets asked what questions.
The law actually is clear. She needed to obtain a dispensation to get into Canada. It would no doubt have been issued, but I had no idea that she had a record or I would have asked the consulate.
My mother-in-law was denied entry into Canada becasue she (stupidly enough) admitted to the customs inspector that she had had a drunk driving conviction 14 years earlier.
We had to wait two hours and pay a $300 fine to clear this up. It was stupid and embarassing, especially when a simple little self-serving lie would have eliminated the problem.
Of course a President usually travels by invitation - which is equivalent to a diplomatic passport for most purposes - so it makes no difference while in office. But, I wonder if Bush has travalled to Canada in the last 20 years and if he admitted his past crimes at the time. If not, he is in violation of the Immigration Act and can be permanently barred from entering Canada.
Well, obviously you won't be voting for the Alliance, a group that might be called the fiscal Democrats and social Republicans.
I come from a long line of unionised NDP party hacks and moderate Trotskyites, and now I live in the US, where I work for the Man, have a 401k and worry about the capital gains tax. I'm hoping to get out of the USA before whoever's in office next has been in for long enough to start screwing things up, because America's a bad place to live if you still remember what it's like to be poor.
Whoever gets elected in the US will be a disaster. There will be an economic downturn - either soft or hard - in the next four years. Someone will have to take the blame, and it's usually whoever's in the White House. So, if I could vote in the American elections, I'd vote Nader. Harry Browne turns my stomach - corporations already control too much of my life, thank you, and compare any Canadian hospital to, say, LA county's public hospitals and you'll begin to appreciate socialised medicine. Nader at least is campaigning on actually improving people's lives, instead of the slow erosion of wages under the status quo. It won't make much difference if Bush or Gore is elected. Unless there's a Republican President, Republican Senate and a Republican House, no one will have the power to interefere with the status quo anyway. But a vote for Nader is a vote for the possibility of a real, social democratic third party.
In Canada, a vote for a party that can't win is still a vote that makes a difference. 10-20 NDP MP's moves parliament well to the left, just as 10-20 PC MP's would to the right. Better still would be proportional representation.
But in the US, if you live in a state where the election is a foregone conclusion, or you just can't abide the two candidates with real chances, your vote is wasted.
Stanislaw Lem and American SF
on
Solaris
·
· Score: 2
I love Lem's work. It derives from such a completely different tradition from Anglo SF and remains so beautifully written. Only Phillip K Dick wrote anything like it in Enlgish. My only criticism is the English translation, which wasn't from the Polish, but translated from the French translation.
I read it in French, and for those capable of doing so, I recommend it over the English version.
One of the things I see in this book is the question of whether we could ever identify a non-human intelligence if we found one. Just how alien is alien? Is Solaris intelligent? It certainly isn't human. This question is never answered, and it remains an open question today. Can we build an artificial intelligence without basically building an artificial human? Is any definition of intelligence possible without making reference to purely human abilities?
You won't find the answers in Lem's work, but you will find the question repeated over and over.
I think Lanier makes some good points. Yes, this is a rant rather than an well thought out refutation, but as the Communist Manifesto shows, you can go a long way with a rant.
The case against AI is pretty good these days. We are beginning to understadn how intelligence emerges from systems, and one of the conclusions we can make is that human-like intelligence seems unlikely to emerge from anything that isn't human. Johnson and Lakoff's exposé of embodied intelligence makes that case convincingly.
The real clue here should be our inability to define intelligence in a way that distinguishes it from being human. Quite possibly we will never be able to identify anything non-human as intelligent.
Belief #3 is interesting. I though only the looney corner of the libertarian movement still stuck to objective semantics. Putnam convincingly dismantled objectivism in the 70's by showing that it is a contradictory system which is incapable of providing a theory of meaning.
Embodied semantics are the rage these days. Perhaps that won't last forever, but I don't see any other solutions that makes sense in light of what we know about biology and computers.
Belief #4 is true of too many of today's technocrats. However, I think the power of darwinian thinking is very strong and very real. What I reject are the conclusions many amateur darwinists draw from it. Cyberselfish, for example, takes the bionomics people to task for their poor economic thinking. Darwinism is increasingly used to justify the worst aspects of capitalism, ignoring the obvious Darwinian analysis of government. The great failure of Darwinian social thinkers is to view their domain in isolation. Just as no responsible ecologist describes the lifecycle of a single animal without considering the other organisms that share its environment, it is stupid to try to understand economics without considering government, society, culture and environment.
A good Darwinian analysis would show that regulatory action and environmental effects form a part of the economic ecosystem. It's never just survival of the fittest organism, even in nature.
Belief #5 is an admission that faster CPU's won't solve any problems. I'm glad to hear it. Commercial computing is still rooted in the 1970's and we need to move on.
And that brings us to belief #6. Millenialist thinking. Things have changed before, and they'll change in the future. But I have seen nothing to lead me to believe that nanotechnology or any other kind of technology will solve any fundammental human problems. Computers have done a great deal for us, but have they really brought any kind of millenial change? No, not really. We had machines before to help do things, we have machines now. Sometimes they still don't work. We still can't take the human out of the loop. We don't appear to be on the verge of uncovering immortality, we can't transfer our brains into computers and may never be able to. Computers can do a lot for us, including making us more intelligent and more adaptable, but this is the same trend that's been in place for millenia.
The year 2000 has come and will soon be gone. And nothing big happened. It's nice to see the technical class getting back to reality.
The cost of printing and distributing textbooks is fairly small. Admittedly, it is less profitable to do a small run of some huge graduate physics text than a mass market bestseller, but still, it isn't the cost of printing and distribution that makes textbooks expensive.
I've been trying for years to find out what the problem is, and as far as I can see, it's a case of simple market failure. A big textbook publisher offers a large royalty to an author to write a text, spends a lot of money marketing it, uses various forms of kickbacks and influence to get universities and school boards to buy it, and then has to charge a fortune to compensate for these costs. Smaller price publishers could enter the market, but without high pressure marketing or big names doing the writing, or claims of US Dept. of Education approval or something, they stand little chance of selling, much less making a profit on short runs.
In the end costs are high and neither publishers nor authors are seeing large profits. Important texts often go out of print. My textbook for phonetics was Ladefoged's "Course in Phonetics" - far and away the best text in the English (and several other languages) on the subject. Yet, it was out of print for 15 years because the publisher held the rights and didn't think it was profitable to print up. I used photocopies. (BTW, the book is now back in print and selling well. It's available from Amazon in paperback for $57.50 - a still obscenely high price for such a fundammental text.)
I have a proposal, but no one is going to like since it invovles spending tax money.
Government could commission textbooks from worthy authors, print them up and sell them at cost, and release them as some form of open content. (I prefer the kind of model that allows unlimited republishing, but not the kinds of modifications open source software allows. Authors want books with their names on them to say what they want them to, and not something else.) If other publishers want to reprint them and sell them, they can, but they have to compete on an open market for their editions.
I think this would have the effect of reducing prices in general. Good text writers could receive a reasonable cash payment up front, and the government has no vested interest in uselessly rereleasing modified versions, but has no motive not to update the text when it genuinely calls for updating.
Those who do write genuinely better texts and don't want to participate in this scheme can still write and publish independently. They do, however, have to have content that is really better because they have to compete with these low cost texts.
In the end, everybody gets at least a bit of money and good texts stay available. No one ever got rich writing textbooks anyway, so I think my half-loaf is better than none for everyone involved.
As for the cost to government, remember that tax money already pays for high priced texts. Most students get grants or student loans to go to college, and public school textbooks are fully paid for out of public school budgets. I think my system would cost less in the long run.
These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it.
I would imagine that program modules that are shipped in separate files constitute "identifiable sections." Furthermore, shipping separately doesn't mean on separate CD's, or else Red Hat would have to GPL all the non-GPL'ed programs that ship on their Linux CD's. Yes, if Sun took GPL'ed code (like the TCP/IP stack) and stuffed it into their monolithic kernal binary, that would strike me as a violation, but any part that can be optionally used would seem to me to be a separate work, and rightly so.
...that the problem is not that Sun is distributing drivers without distributing their source, but rather that Sun is using GPL'ed drivers without distributing the source for the Solaris OS?
I'm not sure this is a problem that needs to be fixed. I always understood the GPL as meaning that if I integrate GPL'ed code into a program in a way that isn't transparently separable from my new code, then I must GPL my new code along with the GPL'ed older code. But, if my code is transparently separable, like header files or object files, even if they are all compiled into a single binary, then I need only distribute the source for changes I have made to files that were already under GPL. I always thought the idea was that you could use GPL'ed code in your non-open projects, as long as the GPL'ed part is transparently separable and distributed whith the program.
Now, I don't see why a device driver's code would be integrated into Sun's source tree in a non-transparent way. It almost certainly resides in separate files, and is either loaded seperately at compile time or at run time. How, then, would this be a GPL violation, even if Sun changed their kernel to make it compatible with formerly Linux drivers?
If I wrote a program (hypothetically) that turned Linux device drivers into Windows device drivers, I couldn't turn around and sue Microsoft for not GPL'ing their source. I fail to see how this is different.
When Chomsky is dead and his linguistcs is forgotten (soon I hope), he'll be remembered for some fairly important contributions to the theory of formal grammars.
He's been at MIT since the 1950's. What leads you to think a liberal (or even a real Marxist, which Chomsky certainly isn't) can't be a tech? Serious, hardcore biology is still dominated by Marxists and far leftists. There is still no such thing as a conservative linguistics department.
Don't mix his crappy linguistics with his crappy politics. Chomsky is careful not to, and those among his detractors who have mixed them generally make asses of themselves.
Japanese telegraphs and their "morris code" equivalent uses exclusively katakana and is strictly phonetic, just like Japanese Braille.
Radio and telegraphic code would not have carried any of the ambiguity of the written language.
Written orders used the whole character set, but I doubt it was genuinely that ambiguous. I might understand a few errors made that way, but not systematically.
Controlling people is something that can be done in any language
Which is preceisely why I find the Churchill argument doubtful. If the langauge is good enough to give orders on factory floors, I doubt military work requires less ambiguity.
While many Japanese industrial processes were adopted from abroad, borrowing words has nothing to do with it. Most of the English words found in Japanese non-technical dictionaries were borrowed before WWII. Furthermore, adopting English words does nothing to change ambiguity. Anything that can be accomplished using an English word could be accomplished using a Japanese one, so long as the speakers agree to that code.
But the fixed order of arguments in mathematics comes from the Greek tradition, which enjoyed much freer word order than English does. For statements of a mathematical or logical nature, the shortest way of expressing them involves fixing the order of arguments, unless all the functions are transitive.
I suspect that there might be more use of quasi-morphology in programming languages if a more morphologically rich language like Russian or Japanese dominated things, although there's plenty of pseudomorphological constructs in C. For example, the distinction between printf, fprintf, sprintf, and wsprintf could be viewed as a matter of morphology.
Slashdot Mirror
I learned two rules for avoiding charges of cheating in my CS department, both of which strike me as pretty sound:
1 - You can't be punished for citing references.
If you get code from some source and you document it, you say so when you turn it in, it isn't cheating. You may not get a very good grade and your prof may not be pleased, but you can't be brought up on charges of cheating.
2 - The Gilligan's Island rule
If you look at someone else's code, then go watch an episode of Gilligan's Island. Anything you can still remember afterwards is fair use. This was treated as a way of defining the line between copying someone else's code and learning from it. As rule, it won't save you in a court of law, but unless you have superhuman memory, you will be hard pressed to remember enough detail about someone else's work to be demonstrably cheating from it.
I have never heard of a student using these policies ever having been charged with academic improprieties. They are much, much easier to follow than the rules for when you can and can't sleep with profs and TA's.
They don't seem to have very accurate speech recognition technology. The article claims to reduce transcription time by a factor of about nine. That's a lot less unreasonable than believing in good speech recognition technology.
My guess is that it's really fairly poor speaker independent stuff. It probably does a quick, low quality word recognition algorithm - quite a few of those are around - and then some sort of Bayesian network to correct the transcription using lexical context. I know that ARPA was openly funding people doing exactly that a few years ago, and I'll bet their papers are on the web. It doesn't shock me greatly that someone has had some measure of success with it.
If it was 100% accurate transcription, then I wouldn't believe it. But as a time saving device for transcribers... that I find credible.
DARPA also funds a lot of automatic topic spotting research. One of my ex-profs received grants from them under just such a rubric and her papers are publicly available on the web. I'll bet whatever technology they are using, it was developed by a prof at an open university who publishes freely.
As for multilingual text searching and summarisation, the best technology of its kind known to me is Latent Semantic Analysis - the brain child of Thomas Landauer. It's a fairly recent, but hardly secret or obscure, indexing technique that's gaining ground commercially for data mining applications. It can certainly do the the small number of things being claimed by this article. All the relevant papers are on the web.
In short, this doesn't sound like super-secret spy stuff. I'll give long odds the real work is in journals and webpages that are publicly available. Having a couple billion dollars to speed up testing and implementation probably helps, but none of this sounds revolutionary or years ahead of the curve.
Here's a few related books I'd like to see reviewed here:
/. these days. Since I've only read two of those and have a long reading list of other stuff before I get to the others, I'd really like to see someone else here take a few of them on.
Manuel Castells' The Information Age: Economy, Society, and Culture
Geoffrey Hodgson's Economics and Utopia: Why the Learning Economy is Not the End of History
Paul Ormerod's Butterfly Economics
Bruce Sterling's Zeitgeist
Bonnie Nardi and Vicki O'Day's Information Ecologies: Using Technology with Heart or Context and Consciousness: Activity Theory and Human-Computer Interaction edited by Nardi.
These kinds of books are a lot more relevant to higher geek culture than the latest Python book, and I'm really glad to see some of this kind of thing on
I could cite Kuhn here and just blow this idea away, but since I'm usually the one on the baracades railing away against Kuhn abuse in the social sciences, it'd look bad for me.
So, let me take an example. Since the early 19th century, it was known that the planet Merucry did not orbit the sun in the manner Newtonian mechanics prescribes for it. So, here's my question. Did this 'disprove' Newtonian mechanics? If so, were the scientists of the time right or wrong in sticking with Newtonian mechanics until the 1920's? If they were wrong, what programme do you suggest the physicists of the 19th century should have followed that would have produced the same massive advancement in human knowledge and power?
Let me add another example from the social sciences: Child development theory takes as a given that children under the age of 5 are not able to judge whether two quantities are of different magnitude by counting them. This is the most consistent result found in child development acccording to authorities like Gallistel and Gelman and a wide variety of child development theories rest on this result among others. They include Piaget's theory of general development and more nativist theories like Pinker's, which seek to show that counting (as opposed to other behaviours like language) are not the product of biologically driven forces.
Now, I can assure you (although you need not take my word for it in order to use this example, just assume I'm telling the truth) that my little brother could make set comparisons by counting at the age of three years and two months and was able to do so for sets of up 70 items. My brother is a particularly gifted mathematician (a strange form of mental illness that explains why he's still unemployed.) Does this fact render null all of the child development research done in the last 50 years? Are we now compelled to say that no one knows anything about child development because a single child exists who defies the empirical results on which those theories are based?
Alternatively, can we view existing child development theories as incomplete but still viable bodies of thought?
Now, let me propose an alternative version of what a theory is. Theories are tools which mediate human interaction with the world. They are semiotic tools, rather than physical tools, but they work in much the same way.
Humans behave in goal-directed ways. For example, when you want to build a house, you have a goal: to have a house. To do this, you must interact with other bits of the universe: land, wood, nails, etc. To do this, you use tools. Furthermore, the kind of house you make - the structure, the composition, the design, even the uses - are in part determined by the tools you have on hand. With a cheap nails and a strong hammer, you build a very different kind of house than the way pre-industrialised people build their homes. (Go look at old homes in Europe or colonial era dwellings on the East Coast.) In fact, you see the problem of building a house very differently with modern tools than you do with other tools.
Furthermore, you judge one tool to be better than another tool by using it. If you buy a nail gun, it is because it makes it easier to build houses. If the nail gun was too heavy or bulky or was constatantly breaking and you couldn't depend on it, you would go back to using the old manual hammer and nails. In fact, the very existence of nail guns is predicated on people having certain tools, like automated, precision nail-making machines so that nails are uniform. Even tools are the products of tools.
And if you find something you can't build because you don't have the right tools for it, you don't abandon your tools and go back to making things with your bare hands.
I'm a fan of Dürrenmatt, and I have been even since I was introduced to his still - as far as I know - untranslated play Die Wiedertäufer. His best known play in English is Die Physiker, about a group of residents of an insane asylum who believe that they are various famous physicists.
/. Among anglophones, only snobby, elite literati types are likely ever to have even heard of him, much less ever have read his plays. I applaud Katz for doing something as audaciously anti-social as bringing up literary culture in /.'s technofetishist discourse and I await his analysis of Brecht's Leben des Galilei or perhaps Borges' La Biblioteca de Babel.
Dürrenmatt is exactly the kind of thing I would never have expected to see on
However, what really blows me away is that Sean Penn would read Dürrenmatt, and that by all accounts he seems to have made a decent art house film of it. Wasn't he the guy who married Madonna and got a reputation for beating people up? I seem to recall him as the butt of jokes back in the 80's as the archtype of the Hollywood ruffian actor with poor impulse control.
I guess this just goes to show that everybody can grow up.
Before he was writer, Heinlein ran for the California assembly on Upton Sinclair's EPIC ticket. EPIC was strongly socialist, with links to radicals, communists, unions, and anti-poverty groups. Heinlein was considered such a radical leftist at the time that the Republican candidate was able to cross-file as a Democrat and win their primaries, leaving him unchallenged on the ballot.
Heinlein certainly takes a strong stance against any kind of interference in markets in his later books, but the positions of EPIC were diametrically opposed to those he espoused in Expanded Universe. They were interventionist, they demanded government action against the Depression and they had nothing but contempt for California's wealthy industrialists and land owners who did nothing in the face of such vast public misery.
Certainly, Heinlein's politics changed a lot between 1938 and 1980. Frankly, I liked the younger Heinlein a lot more. The authoritarian, Chicago-school, Reaganite Heinlein of the 70's and 80's was half the writter of the witty, liberal, socially and culturally conscious Heinlein of the 40's and 50's.
The late Heinlein was a far cry from any anarcho-syndicalist that I know of, although the young one had his moments. He takes a near Randian position on the virtues of the market in all his books after Time Enough for Love (for example, his utopian portrayal of Hell in Job), while taking a truly unlibertarian position on the importance of a powerful central state in Friday. The later Heinlein appears to place little value in democracy and collective action (which are perhaps the most central values of traditional anarcho-syndicalism) and advocates a sort of minimalist dictatorship as the ideal form of government, or at least he does in all of the Lazarus Long books.
Perhaps those weren't his "true" politcal feelings. I have no way to know, authors are allowed to play with ideas. But at the very least, the opinions he lays out in Expanded Universe, a polemic by his own admission, are quite remote from anarcho-syndicalism and even more remote from his political roots.
Robotron manufactured most of the old Warsaw Pact's computer technology. They built a lot of PDP clones and some more original designs, but they relied a lot on reappropriated Western technology.
Robotron produced an 8-bit chip called the U880 that was basically a copy of the ZX80 chip until reunification in 1990. It was generally believed that the chip design had been stolen from its British manufacturer and reimplimented unchanged.
The KC-85 and KC-87 were the only things that could be called a "personal computer" in the East. It had 64k of RAM and could run CP/M. It was roughly equal to early 80's Commodores. It also had a printer, a disk drive and some other peripherals.
A more interesting factoid: Vladimir Putin was directly invovled with Robotron's operations in the 80's. He was the KGB liason in Dresden and Leipzig who controlled Soviet technology transfers (or to put it another way, he coordinated stealing Western computer designs.)
Half of eastern Germany's senior computer scientists seem to have worked there at one time or another.
Robotron is still around, although it is a shadow of its former self. Nowadays its a database services company. They are now "Robotron Datenbank-Software GmbH" and they have a website at http://www.robotron.de/.
There are still hobbyists playing with old eastern block computers. For the German-compatible, try: http://www.robotron-net.de/, the KC-Club at http://www.iee.et.tu-dresden.de/~kc-club/, and http://robotron.informatik.hu-berlin.de/.
According to the artivle, they're working on a substitution scheme so ASCII only users can still type in the URL's. Does this mean that ASCII equivalets will be arbitrary and unintuitive? If so, that's a problem. Let me propose something slightly different:
Unicode is not supposed to over-unify characters, so the ASCII fallback for Japanese could be the romanji transcription - and therefor registering a Japanese domain name automatically registers the romanji equivalent, except that some kanji have more than one possible romanji transcription.
However, some kanji are unified with Chinese characters, which have a different pinyin trasncription.
Chinese is another problem. The logical ASCII equivalent is pinyin stripped of its diacritical marks. But then, many different characters may have the same transcription.
All Cyrillic languages also have an ASCII trasncription scheme too, but it isn't unified. One character may be trasncribed one way in Russian and another way in Bulgarian. Is there a unified transcription scheme for all Cyrillic languages, and is it truely one-to-one? I don't think so. Look at the character usually transcribed as "j" in Russian, and the one usually transcribed that way in Serbian.
ISO-Latin-1 and -2 fallbacks: For ISO-Latin-1, the fallbacks are pretty obvious: "Champs-Élysée" ==> "Champs-Elysee" or in German "Düsseldorf" ==> "Duesseldorf", but in Czech it's a little less obvious. Does "C hacek" map to "Cz" or "Ch" or "Cs"?
So, here is a possible solution: devise unified ASCII transcritption schemes for each language, admitting whatever ambiguities exist in Japanese or similar languages. Then, when you register a non-ASCII name, you are asked on the form to fill out the transcribed ASCII name that corresponds to it and it is also automatically registered to you.
There is some potential for conflict here, if the ASCII transcription corresponds to an existing registered domain or, as in the case of Chinese more than one foreign name corresponds to the same transcription, but I think the problem is manageable.
I want to be able to register domain names in French, German and Russian too. If they are going to support all three zillion kanji and Chinese characters, they need to at least support the various Cyrillic and eastern European Roman alphabets, and the rest of ISO-Latin-1 (which covers all the major and most of the minor Western European languages.) The Persian-based alphabets (Arabic, Farsi, Urdu, etc), Hebrew and Thai are written right-to-left, so I suppose that won't be implimented right away, but it needs to be on the drawing board.
If all those other languages are accounted for, I view this as a good thing. If this is part of an overall shift to Unicode on the web, then all these languages are automatically supported, and I would think it an even better thing.
Electronic voting will never be fully trusted. Much of the confusion in the current election is simply because computers tabulate paper ballots.
Do the opposite. Go to paper balloting with big print and boxes that have to be marked with a big 'X'. Then, these ballots need to be counted by hand in each ward. This is what is done in the overwhelming majority of countries.
An even better reform: stop holding elections for everything on the same day. It's not genuinely convenient to anyone. Compel state and local government to hold elections on a day other than the second Tuesday in November of even numbered years. This would shrink the size of the ballot enough to make paper balloting and manual counting easier. In Federal elections, there would never be more than 3 offices to vote for: a Presdient, a Senator and a Rep.
Even better, have three different election days, all at different times during the 2-year election cycle. One for the feds, one for states, one for local government. That way, there are only three things to vote for in Federal election, three in state election (except for Nebraska) plus state referenda, if any, and one day for local government, which usually means one or two candidates in county races, one or two in municipal races, a school board election, in some places a hospital and/or public transit board election plus county and municipal referenda.
Furthermore, make the FEC final arbiter of all elections. Take local government out of the process of deciding on voting methods. I think this would minimise corruption rather than make it more likely.
And, if you really want to bring American voting into the modern world, use Condorcet voting and/or proportional representation.
Here would be my reform - if I had the dictatorial power to impose it:
1) Austrailian-style manditory voting. No more griping about people who didn't register, or registered but didn't vote. It costs more, but it's worth it.
2) A paid half day off on election day. Give everyone a chance to get to the polls.
3) Condorcet voting for Presdents, Senators and Governors.
4) Allocate seats in the House to each state rather than drawing districts. If a state has only one House Rep, use Condorcet voting. If it has two, divide the state into two electoral districts and use Condorcet voting. For more than that, use party proportional representation to allocate seats, but also guarantee that any party or independent that gets at least the fraction of votes in the state equal to the number of votes divided by the number of seats in the House gets one seat.
That way, all Reps still represent a state, rather than being 'at-large' national Reps as the Germans have, but the number of seats in the House is still apportioned more reasonably according to voters demands.
5) Move all states to unicameral legislatures like Nebraska. There is no need for state government to replicate the silliness of the Federal government. This way, state elections are for a Governor, one Rep and whatever referenda are going on, and judges in those states where state judges are elected. Also, make state legislatures mixed district/proportional voting on the German model. States are small enough to support 'at-large' representation.
6) Elect a single board for county government by at-large voting for multiple candidates. This means your ballot lists all the qualified candidates and asks you to vote for as many as their are seats on the county board. County supervisors are chosen by the elected board.
7) Do the same for school boards, hospital boards and public transit boards, where such things are elected.
8) Do the same for municipal governments, unless they are elected on the "New Plan", where city commissioners are elected instead of appointed by the municipal government and there is no city council. In that case, go back to Condorcet voting.
9) Stop electing every damn office under the sun, especially judges!!! Elections for judges force judges to be biased. It is a travesty of law to do it this way. In California, we elect offices like state treasurer and insurance commissioner and this is stupid. These offices were less corrupt when they were appointed. I haven't seen anything brilliant come out of elected hospital boards or public transit boards either, and the first thing I would do to reform education is get rid of local school boards.
These reforms would bring the US in line with - in fact ahead of - most other countries in terms of sane, modern, reliable, unambiguous voting systems.
Canada is a monarchy. Mexico is a federation. China is a "People's Republic." They're just names, and have little or nothing to do with democracy.
There are huge differences between different schools of thought on what exactly constitutes a fair vote and a fair campaign. Most countries that run regular multi-party elections limit candiadtes access to funds and media, and as far as I know, all of them have some centre of power not elected by simple one-man-one-vote formulas.
In Canada and the UK, leadership is decided based on a first-past-the-post system in each constituency, where the party with the most seats in parliament leads. This can - and in fact usually does - lead to representation out of proportion ot the actual vote count. In France, in addition to a British-style parliament, there is a Senate with limited powers elected by methods that heavily favour conservative rural districts over liberal urban ones. Canada has an appointed Senate, and the UK has the House of Lords. Germany has proportional representation in the Bundestag. The UK has regional governments in Scotland, Wales and Northern Ireland that distort power relationships to favour those regions.
Does any of this disqualify them as democracies? Not by any sensible current definition.
As for me, I think that Condorcet voting is the fairest way to handle single office elections, and proportional representation is the only realistically fair way to handle electing legislative assemblies. I understand that political realities in most nations require some from of disproportionate regional representation, and I'm for them wherever truly necessary, but not otherwise.
I don't think the US needs the electoral college anymore, not in an age of rightly diminished state power and loyalty. As for this election, it seems to me that considering the Nader vote, the majority of the American electorate has expressed a liberal tendency and that Gore's mandate is better than Bushes. Condorcet voting would almost certainly have elected Gore, although still not by any huge margin.
It's mostly the same as generic mozzarella - raw farmer's cheese with the standard set of industrial additives and conservatives, uncoloured and prepared in whatever manner most cost effective.
Poutine cheese is not for cheese conoisseurs. It's cheap and nasty.
It's like eating Tenderflake. :^)
Look, people eat bacon on hamburgers with cheese because this paragon of unkosher cholesterol nightmares tastes damn good. Our bodies are programmed to love fat, and poutine delivers. Trust me, if you love bacon cheeseburgers, you'll love poutine.
Now, having said that, poutine is probably worse for your health than smoking and ought to come with a label from Health Canada just like Export A's. You can gain weight just watching someone else eat it.
But, after a night of hard drinking and pot smoking (did I inhale? Hell, yes!), nothing hits the spot like hitting Chez Lafleur at the corner of rue St-Denis and Carreé St-Louis (in Montreal's Plateau district next to St-Denis Metro - you can't miss it, it's open 24/7) and ordering a big ol' poutine.
I've never been asked about past criminal activity on crossing the border. However, my mother-in-law looks as poor as she is, and her paperwork was a bit of a mess. We had paid for her airline ticket so she could come to our wedding. I suspect that Canada Customs uses some form of profiling, either formally or informally, in deciding who gets asked what questions.
The law actually is clear. She needed to obtain a dispensation to get into Canada. It would no doubt have been issued, but I had no idea that she had a record or I would have asked the consulate.
My mother-in-law was denied entry into Canada becasue she (stupidly enough) admitted to the customs inspector that she had had a drunk driving conviction 14 years earlier.
We had to wait two hours and pay a $300 fine to clear this up. It was stupid and embarassing, especially when a simple little self-serving lie would have eliminated the problem.
Of course a President usually travels by invitation - which is equivalent to a diplomatic passport for most purposes - so it makes no difference while in office. But, I wonder if Bush has travalled to Canada in the last 20 years and if he admitted his past crimes at the time. If not, he is in violation of the Immigration Act and can be permanently barred from entering Canada.
Well, obviously you won't be voting for the Alliance, a group that might be called the fiscal Democrats and social Republicans.
I come from a long line of unionised NDP party hacks and moderate Trotskyites, and now I live in the US, where I work for the Man, have a 401k and worry about the capital gains tax. I'm hoping to get out of the USA before whoever's in office next has been in for long enough to start screwing things up, because America's a bad place to live if you still remember what it's like to be poor.
Whoever gets elected in the US will be a disaster. There will be an economic downturn - either soft or hard - in the next four years. Someone will have to take the blame, and it's usually whoever's in the White House. So, if I could vote in the American elections, I'd vote Nader. Harry Browne turns my stomach - corporations already control too much of my life, thank you, and compare any Canadian hospital to, say, LA county's public hospitals and you'll begin to appreciate socialised medicine. Nader at least is campaigning on actually improving people's lives, instead of the slow erosion of wages under the status quo. It won't make much difference if Bush or Gore is elected. Unless there's a Republican President, Republican Senate and a Republican House, no one will have the power to interefere with the status quo anyway. But a vote for Nader is a vote for the possibility of a real, social democratic third party.
In Canada, a vote for a party that can't win is still a vote that makes a difference. 10-20 NDP MP's moves parliament well to the left, just as 10-20 PC MP's would to the right. Better still would be proportional representation.
But in the US, if you live in a state where the election is a foregone conclusion, or you just can't abide the two candidates with real chances, your vote is wasted.
There is an interesting article by Bruce Sterling about Stanislaw Lem at http://www.well.com:70/0/Publications/authors/Ster ling/Catscan_Stuff/catscan_two.txt& lt;/a> . It is trenchant in just the manner that Sterling's catscan columns and "Cheap Truth" newsletters were.
I love Lem's work. It derives from such a completely different tradition from Anglo SF and remains so beautifully written. Only Phillip K Dick wrote anything like it in Enlgish. My only criticism is the English translation, which wasn't from the Polish, but translated from the French translation.
I read it in French, and for those capable of doing so, I recommend it over the English version.
One of the things I see in this book is the question of whether we could ever identify a non-human intelligence if we found one. Just how alien is alien? Is Solaris intelligent? It certainly isn't human. This question is never answered, and it remains an open question today. Can we build an artificial intelligence without basically building an artificial human? Is any definition of intelligence possible without making reference to purely human abilities?
You won't find the answers in Lem's work, but you will find the question repeated over and over.
I think Lanier makes some good points. Yes, this is a rant rather than an well thought out refutation, but as the Communist Manifesto shows, you can go a long way with a rant.
The case against AI is pretty good these days. We are beginning to understadn how intelligence emerges from systems, and one of the conclusions we can make is that human-like intelligence seems unlikely to emerge from anything that isn't human. Johnson and Lakoff's exposé of embodied intelligence makes that case convincingly.
The real clue here should be our inability to define intelligence in a way that distinguishes it from being human. Quite possibly we will never be able to identify anything non-human as intelligent.
Belief #3 is interesting. I though only the looney corner of the libertarian movement still stuck to objective semantics. Putnam convincingly dismantled objectivism in the 70's by showing that it is a contradictory system which is incapable of providing a theory of meaning.
Embodied semantics are the rage these days. Perhaps that won't last forever, but I don't see any other solutions that makes sense in light of what we know about biology and computers.
Belief #4 is true of too many of today's technocrats. However, I think the power of darwinian thinking is very strong and very real. What I reject are the conclusions many amateur darwinists draw from it. Cyberselfish, for example, takes the bionomics people to task for their poor economic thinking. Darwinism is increasingly used to justify the worst aspects of capitalism, ignoring the obvious Darwinian analysis of government. The great failure of Darwinian social thinkers is to view their domain in isolation. Just as no responsible ecologist describes the lifecycle of a single animal without considering the other organisms that share its environment, it is stupid to try to understand economics without considering government, society, culture and environment.
A good Darwinian analysis would show that regulatory action and environmental effects form a part of the economic ecosystem. It's never just survival of the fittest organism, even in nature.
Belief #5 is an admission that faster CPU's won't solve any problems. I'm glad to hear it. Commercial computing is still rooted in the 1970's and we need to move on.
And that brings us to belief #6. Millenialist thinking. Things have changed before, and they'll change in the future. But I have seen nothing to lead me to believe that nanotechnology or any other kind of technology will solve any fundammental human problems. Computers have done a great deal for us, but have they really brought any kind of millenial change? No, not really. We had machines before to help do things, we have machines now. Sometimes they still don't work. We still can't take the human out of the loop. We don't appear to be on the verge of uncovering immortality, we can't transfer our brains into computers and may never be able to. Computers can do a lot for us, including making us more intelligent and more adaptable, but this is the same trend that's been in place for millenia.
The year 2000 has come and will soon be gone. And nothing big happened. It's nice to see the technical class getting back to reality.
The cost of printing and distributing textbooks is fairly small. Admittedly, it is less profitable to do a small run of some huge graduate physics text than a mass market bestseller, but still, it isn't the cost of printing and distribution that makes textbooks expensive.
I've been trying for years to find out what the problem is, and as far as I can see, it's a case of simple market failure. A big textbook publisher offers a large royalty to an author to write a text, spends a lot of money marketing it, uses various forms of kickbacks and influence to get universities and school boards to buy it, and then has to charge a fortune to compensate for these costs. Smaller price publishers could enter the market, but without high pressure marketing or big names doing the writing, or claims of US Dept. of Education approval or something, they stand little chance of selling, much less making a profit on short runs.
In the end costs are high and neither publishers nor authors are seeing large profits. Important texts often go out of print. My textbook for phonetics was Ladefoged's "Course in Phonetics" - far and away the best text in the English (and several other languages) on the subject. Yet, it was out of print for 15 years because the publisher held the rights and didn't think it was profitable to print up. I used photocopies. (BTW, the book is now back in print and selling well. It's available from Amazon in paperback for $57.50 - a still obscenely high price for such a fundammental text.)
I have a proposal, but no one is going to like since it invovles spending tax money.
Government could commission textbooks from worthy authors, print them up and sell them at cost, and release them as some form of open content. (I prefer the kind of model that allows unlimited republishing, but not the kinds of modifications open source software allows. Authors want books with their names on them to say what they want them to, and not something else.) If other publishers want to reprint them and sell them, they can, but they have to compete on an open market for their editions.
I think this would have the effect of reducing prices in general. Good text writers could receive a reasonable cash payment up front, and the government has no vested interest in uselessly rereleasing modified versions, but has no motive not to update the text when it genuinely calls for updating.
Those who do write genuinely better texts and don't want to participate in this scheme can still write and publish independently. They do, however, have to have content that is really better because they have to compete with these low cost texts.
In the end, everybody gets at least a bit of money and good texts stay available. No one ever got rich writing textbooks anyway, so I think my half-loaf is better than none for everyone involved.
As for the cost to government, remember that tax money already pays for high priced texts. Most students get grants or student loans to go to college, and public school textbooks are fully paid for out of public school budgets. I think my system would cost less in the long run.
From the GPL license:
These requirements apply to the modified work as a whole. If identifiable sections of that work are not derived from the Program, and can be reasonably considered independent and separate works in themselves, then this License, and its terms, do not apply to those sections when you distribute them as separate works. But when you distribute the same sections as part of a whole which is a work based on the Program, the distribution of the whole must be on the terms of this License, whose permissions for other licensees extend to the entire whole, and thus to each and every part regardless of who wrote it.
I would imagine that program modules that are shipped in separate files constitute "identifiable sections." Furthermore, shipping separately doesn't mean on separate CD's, or else Red Hat would have to GPL all the non-GPL'ed programs that ship on their Linux CD's. Yes, if Sun took GPL'ed code (like the TCP/IP stack) and stuffed it into their monolithic kernal binary, that would strike me as a violation, but any part that can be optionally used would seem to me to be a separate work, and rightly so.
...that the problem is not that Sun is distributing drivers without distributing their source, but rather that Sun is using GPL'ed drivers without distributing the source for the Solaris OS?
I'm not sure this is a problem that needs to be fixed. I always understood the GPL as meaning that if I integrate GPL'ed code into a program in a way that isn't transparently separable from my new code, then I must GPL my new code along with the GPL'ed older code. But, if my code is transparently separable, like header files or object files, even if they are all compiled into a single binary, then I need only distribute the source for changes I have made to files that were already under GPL. I always thought the idea was that you could use GPL'ed code in your non-open projects, as long as the GPL'ed part is transparently separable and distributed whith the program.
Now, I don't see why a device driver's code would be integrated into Sun's source tree in a non-transparent way. It almost certainly resides in separate files, and is either loaded seperately at compile time or at run time. How, then, would this be a GPL violation, even if Sun changed their kernel to make it compatible with formerly Linux drivers?
If I wrote a program (hypothetically) that turned Linux device drivers into Windows device drivers, I couldn't turn around and sue Microsoft for not GPL'ing their source. I fail to see how this is different.
When Chomsky is dead and his linguistcs is forgotten (soon I hope), he'll be remembered for some fairly important contributions to the theory of formal grammars.
He's been at MIT since the 1950's. What leads you to think a liberal (or even a real Marxist, which Chomsky certainly isn't) can't be a tech? Serious, hardcore biology is still dominated by Marxists and far leftists. There is still no such thing as a conservative linguistics department.
Don't mix his crappy linguistics with his crappy politics. Chomsky is careful not to, and those among his detractors who have mixed them generally make asses of themselves.
Japanese telegraphs and their "morris code" equivalent uses exclusively katakana and is strictly phonetic, just like Japanese Braille.
Radio and telegraphic code would not have carried any of the ambiguity of the written language.
Written orders used the whole character set, but I doubt it was genuinely that ambiguous. I might understand a few errors made that way, but not systematically.
Controlling people is something that can be done in any language
Which is preceisely why I find the Churchill argument doubtful. If the langauge is good enough to give orders on factory floors, I doubt military work requires less ambiguity.
While many Japanese industrial processes were adopted from abroad, borrowing words has nothing to do with it. Most of the English words found in Japanese non-technical dictionaries were borrowed before WWII. Furthermore, adopting English words does nothing to change ambiguity. Anything that can be accomplished using an English word could be accomplished using a Japanese one, so long as the speakers agree to that code.
But the fixed order of arguments in mathematics comes from the Greek tradition, which enjoyed much freer word order than English does. For statements of a mathematical or logical nature, the shortest way of expressing them involves fixing the order of arguments, unless all the functions are transitive.
I suspect that there might be more use of quasi-morphology in programming languages if a more morphologically rich language like Russian or Japanese dominated things, although there's plenty of pseudomorphological constructs in C. For example, the distinction between printf, fprintf, sprintf, and wsprintf could be viewed as a matter of morphology.