The problem is that they're also vouching for lots of non-malicious identities - you can set things up so it's pretty hard to spot the bad guys automatically without also getting good guys, and manually blacklisting a few thousand droids is too much work.
Checksums can be useful for Warez distribution, where a given file has a unique checksum, but they're not useful for compressed audio files, where there isn't a 1-1 mapping between the original and the bitstream you're distributing. Your MP3s, Oggs, etc. depend on the version of compression program you're using, specific parameters you used to compressed with, phase of the moon, or whatever. So all that extra work you're doing with incremental checksums isn't very helpful, because there isn't One True Value for you to compare against, though it can make it easier to do blacklists for some of the more efficient poisoning techniques. The easy way to do poisoning is to take your standard copy of Poison singing "Happy Copyright Violation Lawsuit To You" followed by a small block of serial-number bits to make the checksums different for each copy - this lets you crunch the partial MD5 for the first 99.99% of the file once and only have to do extra work for the last 0.01% of each poison file you're creating, which lets you create ~10000 poison files with only twice as much work as creating one. If you go to the extra work of using your incremental hash techniques for whitelisting files, it doesn't gain you anything, though it will catch this type of blacklisted files after the first block. If you're doing a whitelist-based system, you actually have to have a human listen to the thing (or a robot that does music-to-text, but that isn't going to catch files with the original lyrics and lower resolution or different tunes.) And blacklisting users has other problems, if you're not extremely (and probably unscalably) careful about your web of trust, since Poisoners can do things like give each other good karma and distribute a few real files, then put out lots of Bad Karma blacklist reports about non-Poisoners.
Do go read about BitTorrent, though - it does use a number of the ideas you've mentioned for efficient distrubtion.
If they see you you run out of fingers, you probably won't pass.
If you're having trouble telling how many fingers you're holding up, you probably won't pass either.
If you start counting, and break out in the giggles halfway through, you probably won't pass, unless you can recover and explain what was funny about that time without ratting on your friends.
I've separately posted a discussion about how it's easy to create large numbers of files with different checksums pretending to be different audio rips of the same tune. Not only does this flood the typical index system, but if the Poisoners can create lots of users, they can all rate the poisoned files as good, or rate non-poisoned files as bad, and they can probably give themselves great karma by first sending in lots of reports about having successfully shared lots of good files with each other.
Not much - it's the old "Tentacles Of Medusa" problem. Depending on how complex your rating system is, the Poisoners will probably have to do a little work to give their tentacles lots of really great karma by saying they've shared zillions of files with each other, or whatever else it takes to game the voting system besides voting early and often.
If you don't have a trusted third party, you can't easily create a rule that voting is only allowed after downloads - either the downloader can be a Poisoner's Apprentice who's dishonestly claiming to have downloaded a bad file from you, trashing your karma, or else there's a mechanism for you to claim that he's lying, in which case any Poisoner who you download poisoned files from can use that mechanism to claim that you're lying if you complain.
But if you *do* have Trusted Third Parties, Poisoners will either attack them technically, sue them, or pretend to be them, or all three. And Slashdot MetaModeration isn't directly applicable to this problem, because the disputed event is private, unlike Slashdot postings which third and fourth parties can look at and decide whether they're really Insightful or Trolls.
It's fundamentally a social problem, and public-key crypto webs of trust don't map very well to it. They work badly enough for the social problems they were *designed* to solve (e.g. Phil Zimmermann's anti-nuclear activists try to prevent forgery and eavesdropping by Feds infiltrators - it starts to get weak when one of your friends is gullible about signing keys for his new friends who are really Feds.) And maintaining really broad webs of trust is surprisingly difficult, except when there's a commercial enterprise to sustain it, i.e. either a record company or a lawsuit target.
It's very easy to create a large number of identities in this system, each pretending to be a real person but really just Yet Another Tentacle of the Poisoners. They can all build up great reputations by signing each others's keys, and sending reports into the whoever-archives-reports-about-users system claiming to have done lots of downloads to each other, and they're all listed as having T3 or Ethernet connections so they're very attractive. And they can pump out a large number of files that they've signed, indicating correctly that the checksum on File#12345 is 290384098213 or whatever, for many different files with many different names, all of which are really Poison singing "Happy Copyright Violation Lawsuit To You!" with a different serial-number burst of noise at the end. They can distribute enough non-poisoned songs to create some good genuine reputations, use those to sign peoples' keys and get people to sign their keys, use these reputations to sign the keys of their other tentacles, and start distributing poisoned songs to people who trust them directly or indirectly, using their keys which have been outed as Poisoners to sign the keys of people who aren't tentacles. Even more fun, you can distribute lots of poisoned index data - some P2P systems are much easier to kill that way.
Most E-Bay users are honest, and most of the dishonest ones are afraid of getting caught, and most of the dishonest ones who aren't afraid of getting caught are either too small-time to matter or too stupid to get away with it, and ripping people off takes a certain amount of Real Work and creates a certain level of traceability.
This is different - there's no penalty other than your reputation, the Poisoners have a much stronger legal position than anybody who might complain (Hey - I tried to rip off their music and they gave me a Bad Copy!), identities can be created free by robots, reputations for the identities don't take too much work to forge, and there are lots of creative ways to cheat.
Unlike Warez or some lossless compression systems, this doesn't work for audio or for video applications using lossy compression instead of distributing exact copies. The reason is that different compression runs don't need to have identical checksums, depending on your compression parameters, equipment, etc., so the Poisoners can go create lots of different files all claiming to be a rip of the real thing, and they can have multiple identities all claiming to have a version to share, so even if you burn one file and one identity, they can trivially create more. If they're clever, they can do this with very little extra work - each version has identical data except in the last block (448 bits for MD5, I forget how many for SHA1), which is juggled a bit. Since music files are large, this means they can do 99.99% of the work once and only have to repeat the last 0.01% multiple times. GPG signatures on the files don't help much either - they've provided a genuine signature saying that jack12345 and lars6789 both downloaded this file of "Whoops I Cloned It Again" and got checksum 12903849021834, but when you listen to it, it's just Poison singing "Happy Copyright Violation Lawsuit To You" with a burst of noise in the last few milliseconds.
CRCs aren't the only kind of checksum out there, though they're nice and fast. Cryptographic-quality checksums avoid the problems - if you change one bit of the input, they change about half the bits of the output, and it's nearly impossible to predict what the changes will be. MD5 was the most popular for a long time, though SHA1 has been replacing it for a variety of technical reasons. MD5 is 128 bits long, SHA1 is 160, so you don't need to worry about collisions unless you have more than 2**64 or 2**80 files.
Plover has the clue here - in an open system it's easy to create lots of identities that are just tentacles of yourself. So if you mainly count positive votes, the Poisoners can give their tentacles lots of positive votes. But if you count negative votes heavily, they can go slandering big sites, and unlike Slashdot Karma, it's hard to metamoderate, because the ratings are about private transactions. So it's pretty easy for the Poisoners to create a lot of good-looking sites with good reputations, and use them to spread poisoned files.
What's worse, it's very difficult to identify bad files automatically, because different rips of the same original can have different checksums, so the poisoners can spread lots of versions with different checksums, so you can't tell whether two files claiming to be a 128kbps ogg of "Whoops I Cloned It Again" came from the same original, only that they're not the same, so you have to listen to the thing all the way through to be sure that it doesn't suddenly turn into an FBI/RIAA/KGB warning against copying music, or a commercial for the CD containing the FM version of the track, or that it doesn't have a lot of low-level static in it. (If I were an artist, I might be more annoyed about the latter.)
The census does have some good intentions about privacy, unlike the US census, though I doubt they'll actually do much good. The official procedure is that any name-identifiable material gets shredded after they're done counting, unless you give them permission, in which case they'll keep the name private for 99 years and then release it. However, if they've got your address, employer, ethnic group, etc., it's hard to retain that in a way that doesn't make it easy to tell who you are from correlation. Australia may not do anything bad with the data, but I'd still be wary about trusting them. Does your employer know that two people living at your address is Jewish? How many families on your block in two-bedroom apartments with three children are recent Hong Kong immigrants? Are your papers in order?
The US rules say that the data must be kept private for 75 years, but that rule was blatantly violated during World War 2, when the Army used census data to find people of Japanese ancestry and arrest them all. The current census asks lots of detailed data, especially about national origins of Spanish-speaking people. It's theoretically only kept on a census-tract basis, not an individual address basis, but that can be as small as a block or two. How many Guatemalan couples with three kids aged 2, 4, and 7 live on your block?
The 1996.AU census specified 7 Christian groups, plus an "Other" fill-in-the-blank. The 2001 version adds Islam and Buddhism. I couldn't find any 2001 religion results on the site - I assume they're not done yet, though they've hit high-priority topics such as population and attendance at sporting events.... There weren't any reported Jedi in 1996:-)
Neither the religion nor the language sections explicitly mention Aboriginal religions or languages, though about 7000 people wrote that in on the 1996 form, and a number of other people wrote in "Nature Religions", which may include some aboriginals as well as neo-pagans. The Ancestry section does include "Australian", and there is also an explicit question asking if you're an Aboriginal or Torres Strait Islander, and there are some specific instructions for Australian South Sea Islanders as well.
Sigh - I ask about non-US cable policies, and the only non-US answer talks about a bunch of countries with socialized medicine, gorgeous scenery, and Bad Weather, saying nothing about cable modems, plus Australia which has nice weather but telecom companies who are even more blazingly clueless than most of their politicians:-) Some of the Canadian cable modems are really fast, but I don't know about their server policies. On the US side, thanks to the person who pointed out Earthlink - their DSL policies are about the closest to reasonable among the lower-priced large providers, and I hadn't thought to check their cable modem policies.
Most of the cable companies supported by AT&T Broadband doesn't let you run a server on *your* machine - you can't take that nice In reality, while the cable modem upstream bandwidth is limited, it's not THAT limited. Most of the equipment can limit you to 128kbps upstream, which is a surprisingly large amount of data transfer for any activities other than distributing lots of CDs or movies. You wouldn't want to run a high-volume commercial site on it, not only because it's too slow, but because the cable tv Service Level Agreements say Look, it's just television, if it goes down for a day or two in bad weather, go read a book or help your kids build snowmen and the cable modems get the same quality of repair service. 2.5Mbps upstream is more than a T1 - a surprising number of medium-sized business offices don't need that much for downstream (though it's always nice) for the number of people you're sharing your cable feed with. And most of the newer cable modem systems are running on Hybrid Fiber-Coax - if they run out of bandwidth, it's pretty easy to split the segment, but more importantly, the cost of Packet Shapers has been coming way down - they can stick a box in the upstream that starts throttling individual connections if the total gets too high.
The real reason they banned servers was that the beta-test cities had some equipment problems causing high packet loss (perceived as low throughput due to TCP retransmits) which led to all those bogus but effective Don't Be A Web Hog smear ads from competing telco DSL services - and the equipment they were using couldn't limit individual users below the raw transmit level of 768kbps, so they were worried that they might have worse public relations problems (i.e. even lower sales) if they started having neighborhoods with bad performance because of somebody's p0rn server. By the time Napster came along, they had performance under control, so they had official policies about "Napster is Bad Evil Bandwidth-Eating Copyright Theft" even though half the employees thought "well *duhh*, it's about *time* people had a compelling reason to get broadband besides gamez for their kids":-)
The big US cable modem providers don't allow users to run web servers or other information-producer applications on their sites, which is really much more important than which ISPs deal with which Content Providers and which Cable Modem services and who gets which money. For those of you not under control of the FCC, do your cable modem providers allow you to run servers? Do people in your areas have anything interesting on them, other than simple web servers and maybe webcams? How does this change the dynamics of content that you can access? Do people mostly ignore the issue and put their pictures on terra.es anyway?
It seems that the main location for experimentation is college campuses, which often have high-speed LANs in the dorms and may not be too aggressive about firewalls to the outside world, though there are also some US ISPs and DSL providers that allow servers on their DSL connections.
I've ranted for a long time that the cable modem companies seldom had half a clue about what they were doing, and that the AOL-led "cable openness" nonsense pushed most of them in entirely the wrong directions (stonewalling about technical difficulties and regulatory issues rather than simply negotiating a wholesale discount price, which was one of the two real ISP complaints.) So now they've got a wholesale price, and AOL seems willing to be treated like a premium movie channel on the cable modem bill (or separately billed, or whatever)(rather than the cable modem being a line item on the user's AOL bill, which was one of their major bogus objections.) They seem to have met halfway, which is a Really Good Thing.
The cable modem content-creation efforts failed, partly because the ideas that can be generated by one group of Central Planners are usually much lamer than the ideas created by a large number of different groups (even if they're not well-funded) - the anti-server policies adopted for performance reasons discouraged development of real applications, though some content applications developed by other people in the market succeeded at getting people to want to buy cable modem service (some of the games, and of course Napster). Another big reason, of course, was that there were only so many viable market niches for search engines, and the Excite business model depended on banner advertiser funding at a time that the market was going through rapid discovery of what that was worth (much much less for late arrivals in a crowded market than for the early adopters when web users were also mostly well-paid early adopters in an uncrowded market.) And Blue Mountain Greeting Cards didn't appear to have much business model at all - your mother could send out cutsie MommySpam(tm), but nobody got paid anything:-)
AOL is another example of this - it's content that wanted more bandwidth, but that had become successful without it, though unlike some approaches, it's a combination of user-developed and service-provided content on the service's computers rather than the users'.
So there's a published list, even if it's only published to cops, saying "This person is likely to commit a crime". Leave aside the obvious civil liberties issues for the moment - this seems like simple libel to me. At least for the Usual Suspects who haven't yet been arrested for things, this doesn't sound like investigation of a crime or other legitimate police function that's protected by laws protecting government officials doing their official jobs. Of course, most of the people on the list probably don't have the resources to fight that kind of libel suit, but it'd be fun to get the ACLU or some other pro bono support for it.
The two big economic issues are "What's the cost to deliver a given weight into LEO or GEO?" and "Can you sell enough of these to pay for the development costs?" - they're related. If it's substantially cheaper, they'll find some demand for it, but if it's too similar to its competition, they probably won't. I was recently rereading "The Third Industrial Revolution", a wildly optimistic ~1975 book about how reduced costs of space travel will make it possible to do industry up there which will pay for more launches which amortize development costs and give us economies of scale that will bring down the costs even further which will lead to more industry...., kind of like the Internet boom ca. 1995-1999. While it didn't happen (:-), the Moore's Law effect in the computer industry means that low-weight satellites can do interesting things - at $100/pound, hobbyists, student groups, etc. can launch the occasional CanSat satellite-in-a-coke-can and other Picosats (defined as you can launch picosats just for the fun of it. Remailer-in-the-sky? Radio transmitter playing "Happy Birthday" for your Mom? Whatever - it's not just for OSCAR the Ham Radio Dude any more. But at $1000/pound, you'd need to be a bit more serious about the application (i.e. probably commercial), or *much* better at miniaturization.
The article says Despite Air Force hopes that the Atlas 5 would slash space travel costs, its debut takes place during a prolonged slump in commercial satellite launches. A glut of other new-generation rockets completed or in the works, along with a weak satellite launch market in the coming years, could mean fewer Atlases are built to recoup development costs, according to commercial aerospace officials.
OK, it's off-topic, but the article does start out with it. Can't even find a non-chain restaurant for lunch in Silicon Valley? Where was he? Must have been over on 280 near the airport or something, because just about anywhere else you can find real food if you go more than a block or two off the freeway. Some of it's boring, but it's certainly around.
High-velocity pizza delivery... Is Snow Crash really ten years old? While we really need to have Uncle Enzo arrange to have somebody take care of Ashcroft, if you count software, movies, music, Muzak, videos, books, web pages, and anything else that involves people writing or performing text or songs or other things with original thought, or anybody taking those things and packaging and marketing them, it sounds like it could easily be 5% of the GNP. Ignore whether copyright is actually relevant to the business models (Ashcroft probably does) - since the US joined the Berne Convention, just about anything is born copyrighted.
You need to eat your own dogfood, at least if it's cooked enough, and for most applications you need to develop them to run on trailing-edge user machines, not leading-edge, so it performs well for everybody, unless you're doing things like apps for bleeding-edge graphics boards or some such. On the other hand, there are things you should have overkill capability on - file server space, lots of space hard drives in removable drawers, and if you're doing lots of big compiles, a few fast boxes to run them on so you're not waiting much. The removable drives are a usually-adequate substitute for having lots of extra machines - there are so many times you need a clean environment to build and test in and it's a lot easier to budget for 10 spare disks than for 10 spare computers, plus if you run out you can go spend some petty cash at Fry's for a couple more drives instead of having to get some more computers. Besides, you're running X Windows, so your desktop itself doesn't need to be replaced that often - you can put the bucks into compute servers in some back room instead.
In my case (consultant / systems engineer) my primary environment is a laptop, and we need new ones every ~2 years because they get beaten to splinters hauling them around, plus our Corporate IT Bureaucrats keep giving us new Microsoft Bloated Office and Operating Systems, so we at least need to upgrade RAM and disk space every 1-2 years, while if we were using Unix it wouldn't be such a problem.
Surely you don't mean "Desk-Top Publishing" as something that needs a gigawatt PC... That's a job for low-end Macintoshes with big screens on them, or low-end PCs with somewhat newer graphics card (so you can get enough video memory for enough 2-D resolution.) Other than that, your PII400 is overpowered, or you're running on real bloatware apps and need to switch to three-year-old bloatware that'll be really fast.
Now, video editing, or hi-res graphics, or gamez, those can actually use up horsepower. And I really am happier recompling large applications on my PIII400 than on the P60s, but half of that is just memory size...
Slashdot Mirror
Filk was around as a genre before the typo that gave it its name :-) Consonance is an annual music convention in the Bay Area.
The problem is that they're also vouching for lots of non-malicious identities - you can set things up so it's pretty hard to spot the bad guys automatically without also getting good guys, and manually blacklisting a few thousand droids is too much work.
Do go read about BitTorrent, though - it does use a number of the ideas you've mentioned for efficient distrubtion.
Then there was the line from Night of the Comet "*Daddy* would have bought us *Uzis*!"
If you start counting, and break out in the giggles halfway through, you probably won't pass, unless you can recover and explain what was funny about that time without ratting on your friends.
I've separately posted a discussion about how it's easy to create large numbers of files with different checksums pretending to be different audio rips of the same tune. Not only does this flood the typical index system, but if the Poisoners can create lots of users, they can all rate the poisoned files as good, or rate non-poisoned files as bad, and they can probably give themselves great karma by first sending in lots of reports about having successfully shared lots of good files with each other.
Not much - it's the old "Tentacles Of Medusa" problem. Depending on how complex your rating system is, the Poisoners will probably have to do a little work to give their tentacles lots of really great karma by saying they've shared zillions of files with each other, or whatever else it takes to game the voting system besides voting early and often.
But if you *do* have Trusted Third Parties, Poisoners will either attack them technically, sue them, or pretend to be them, or all three. And Slashdot MetaModeration isn't directly applicable to this problem, because the disputed event is private, unlike Slashdot postings which third and fourth parties can look at and decide whether they're really Insightful or Trolls.
It's very easy to create a large number of identities in this system, each pretending to be a real person but really just Yet Another Tentacle of the Poisoners. They can all build up great reputations by signing each others's keys, and sending reports into the whoever-archives-reports-about-users system claiming to have done lots of downloads to each other, and they're all listed as having T3 or Ethernet connections so they're very attractive. And they can pump out a large number of files that they've signed, indicating correctly that the checksum on File#12345 is 290384098213 or whatever, for many different files with many different names, all of which are really Poison singing "Happy Copyright Violation Lawsuit To You!" with a different serial-number burst of noise at the end. They can distribute enough non-poisoned songs to create some good genuine reputations, use those to sign peoples' keys and get people to sign their keys, use these reputations to sign the keys of their other tentacles, and start distributing poisoned songs to people who trust them directly or indirectly, using their keys which have been outed as Poisoners to sign the keys of people who aren't tentacles. Even more fun, you can distribute lots of poisoned index data - some P2P systems are much easier to kill that way.
This is different - there's no penalty other than your reputation, the Poisoners have a much stronger legal position than anybody who might complain (Hey - I tried to rip off their music and they gave me a Bad Copy!), identities can be created free by robots, reputations for the identities don't take too much work to forge, and there are lots of creative ways to cheat.
Unlike Warez or some lossless compression systems, this doesn't work for audio or for video applications using lossy compression instead of distributing exact copies. The reason is that different compression runs don't need to have identical checksums, depending on your compression parameters, equipment, etc., so the Poisoners can go create lots of different files all claiming to be a rip of the real thing, and they can have multiple identities all claiming to have a version to share, so even if you burn one file and one identity, they can trivially create more. If they're clever, they can do this with very little extra work - each version has identical data except in the last block (448 bits for MD5, I forget how many for SHA1), which is juggled a bit. Since music files are large, this means they can do 99.99% of the work once and only have to repeat the last 0.01% multiple times. GPG signatures on the files don't help much either - they've provided a genuine signature saying that jack12345 and lars6789 both downloaded this file of "Whoops I Cloned It Again" and got checksum 12903849021834, but when you listen to it, it's just Poison singing "Happy Copyright Violation Lawsuit To You" with a burst of noise in the last few milliseconds.
CRCs aren't the only kind of checksum out there, though they're nice and fast. Cryptographic-quality checksums avoid the problems - if you change one bit of the input, they change about half the bits of the output, and it's nearly impossible to predict what the changes will be. MD5 was the most popular for a long time, though SHA1 has been replacing it for a variety of technical reasons. MD5 is 128 bits long, SHA1 is 160, so you don't need to worry about collisions unless you have more than 2**64 or 2**80 files.
What's worse, it's very difficult to identify bad files automatically, because different rips of the same original can have different checksums, so the poisoners can spread lots of versions with different checksums, so you can't tell whether two files claiming to be a 128kbps ogg of "Whoops I Cloned It Again" came from the same original, only that they're not the same, so you have to listen to the thing all the way through to be sure that it doesn't suddenly turn into an FBI/RIAA/KGB warning against copying music, or a commercial for the CD containing the FM version of the track, or that it doesn't have a lot of low-level static in it. (If I were an artist, I might be more annoyed about the latter.)
(please don't bother the folks at san-andreas.com, who for some reason are in New Jersey
The US rules say that the data must be kept private for 75 years, but that rule was blatantly violated during World War 2, when the Army used census data to find people of Japanese ancestry and arrest them all. The current census asks lots of detailed data, especially about national origins of Spanish-speaking people. It's theoretically only kept on a census-tract basis, not an individual address basis, but that can be as small as a block or two. How many Guatemalan couples with three kids aged 2, 4, and 7 live on your block?
Neither the religion nor the language sections explicitly mention Aboriginal religions or languages, though about 7000 people wrote that in on the 1996 form, and a number of other people wrote in "Nature Religions", which may include some aboriginals as well as neo-pagans. The Ancestry section does include "Australian", and there is also an explicit question asking if you're an Aboriginal or Torres Strait Islander, and there are some specific instructions for Australian South Sea Islanders as well.
Most of the cable companies supported by AT&T Broadband doesn't let you run a server on *your* machine - you can't take that nice In reality, while the cable modem upstream bandwidth is limited, it's not THAT limited. Most of the equipment can limit you to 128kbps upstream, which is a surprisingly large amount of data transfer for any activities other than distributing lots of CDs or movies. You wouldn't want to run a high-volume commercial site on it, not only because it's too slow, but because the cable tv Service Level Agreements say Look, it's just television, if it goes down for a day or two in bad weather, go read a book or help your kids build snowmen and the cable modems get the same quality of repair service. 2.5Mbps upstream is more than a T1 - a surprising number of medium-sized business offices don't need that much for downstream (though it's always nice) for the number of people you're sharing your cable feed with. And most of the newer cable modem systems are running on Hybrid Fiber-Coax - if they run out of bandwidth, it's pretty easy to split the segment, but more importantly, the cost of Packet Shapers has been coming way down - they can stick a box in the upstream that starts throttling individual connections if the total gets too high.
The real reason they banned servers was that the beta-test cities had some equipment problems causing high packet loss (perceived as low throughput due to TCP retransmits) which led to all those bogus but effective Don't Be A Web Hog smear ads from competing telco DSL services - and the equipment they were using couldn't limit individual users below the raw transmit level of 768kbps, so they were worried that they might have worse public relations problems (i.e. even lower sales) if they started having neighborhoods with bad performance because of somebody's p0rn server. By the time Napster came along, they had performance under control, so they had official policies about "Napster is Bad Evil Bandwidth-Eating Copyright Theft" even though half the employees thought "well *duhh*, it's about *time* people had a compelling reason to get broadband besides gamez for their kids" :-)
It seems that the main location for experimentation is college campuses, which often have high-speed LANs in the dorms and may not be too aggressive about firewalls to the outside world, though there are also some US ISPs and DSL providers that allow servers on their DSL connections.
The cable modem content-creation efforts failed, partly because the ideas that can be generated by one group of Central Planners are usually much lamer than the ideas created by a large number of different groups (even if they're not well-funded) - the anti-server policies adopted for performance reasons discouraged development of real applications, though some content applications developed by other people in the market succeeded at getting people to want to buy cable modem service (some of the games, and of course Napster). Another big reason, of course, was that there were only so many viable market niches for search engines, and the Excite business model depended on banner advertiser funding at a time that the market was going through rapid discovery of what that was worth (much much less for late arrivals in a crowded market than for the early adopters when web users were also mostly well-paid early adopters in an uncrowded market.) And Blue Mountain Greeting Cards didn't appear to have much business model at all - your mother could send out cutsie MommySpam(tm), but nobody got paid anything
AOL is another example of this - it's content that wanted more bandwidth, but that had become successful without it, though unlike some approaches, it's a combination of user-developed and service-provided content on the service's computers rather than the users'.
So there's a published list, even if it's only published to cops, saying "This person is likely to commit a crime". Leave aside the obvious civil liberties issues for the moment - this seems like simple libel to me. At least for the Usual Suspects who haven't yet been arrested for things, this doesn't sound like investigation of a crime or other legitimate police function that's protected by laws protecting government officials doing their official jobs. Of course, most of the people on the list probably don't have the resources to fight that kind of libel suit, but it'd be fun to get the ACLU or some other pro bono support for it.
The article says Despite Air Force hopes that the Atlas 5 would slash space travel costs, its debut takes place during a prolonged slump in commercial satellite launches. A glut of other new-generation rockets completed or in the works, along with a weak satellite launch market in the coming years, could mean fewer Atlases are built to recoup development costs, according to commercial aerospace officials.
OK, it's off-topic, but the article does start out with it. Can't even find a non-chain restaurant for lunch in Silicon Valley? Where was he? Must have been over on 280 near the airport or something, because just about anywhere else you can find real food if you go more than a block or two off the freeway. Some of it's boring, but it's certainly around.
High-velocity pizza delivery... Is Snow Crash really ten years old? While we really need to have Uncle Enzo arrange to have somebody take care of Ashcroft, if you count software, movies, music, Muzak, videos, books, web pages, and anything else that involves people writing or performing text or songs or other things with original thought, or anybody taking those things and packaging and marketing them, it sounds like it could easily be 5% of the GNP. Ignore whether copyright is actually relevant to the business models (Ashcroft probably does) - since the US joined the Berne Convention, just about anything is born copyrighted.
In my case (consultant / systems engineer) my primary environment is a laptop, and we need new ones every ~2 years because they get beaten to splinters hauling them around, plus our Corporate IT Bureaucrats keep giving us new Microsoft Bloated Office and Operating Systems, so we at least need to upgrade RAM and disk space every 1-2 years, while if we were using Unix it wouldn't be such a problem.
Now, video editing, or hi-res graphics, or gamez, those can actually use up horsepower. And I really am happier recompling large applications on my PIII400 than on the P60s, but half of that is just memory size...