Back when I had a lab, there was a time that the A/C system in the ceiling croaked and they had a free-standing A/C unit as a stopgap until they got it fixed. It had a bunch of tubes in the back and an opening in the front where the cold air came out. And my department had a sales guy who was a wine expert, had a small Napa Valley winery with a couple of friends, and did occasional evening winetastings after work. So there was obviously one thing we had to do with the A/C unit, which was to chill a couple of bottles of interesting white wine, as a change from our usual reds. Worked real fine.
Ok, so the fixes are in 1.5.0.5. Does the 2.0 release candidate alpha/beta/etc. have the same vulnerabilities, and are they fixed, and (less likely) does Mozilla Suite have them?
It's getting to be time to update my Mozilla Suite anyway - is 2.0xx cooked enough to use, or is it better to go to 1.5.0.5 and wait for 2.0 final to update again?
Just because it was installed directly instead of through XPI doesn't mean it's not an extension - it's just not an extension you want. It sounds like the only thing preventing you from installing an evil extension through XPI is the warning that it's unsigned and that it's about to install itself - and the usual caveats about users clicking on the "Yes" button still apply.
David Brin's book "The Transparent Society" (1998, online excerpts) talks about the effects of cheap computing/camera/database technology on privacy. It's a pre-9/11/2001 look at what societies and governments can do with Moore's Law kicking technology. A major point was that either society forces the government to be open about what it's doing and allow the public to watch it, or else the government will use all the same technology _without_ anybody watching it. Now, of course, we've got the Bush Administration, so all the happy 90s-boom speculation about "Well, what if another president as evil as Nixon got elected?" "No way, the public wouldn't let that happen again!" is moot, and we didn't get the transparency locked down when we could.
License plate reading has been relatively possible for a long time - the big innovation here is that it's a mobile reader you can use from a moving car as opposed to a fixed-location mounted reader. Back in the mid 90s, San Francisco was going to be tearing down a highway for earthquake repair, and wanted to prevent traffic problems. So they scanned all the license plates of the cars that took the highway, looked up the owners' addresses, and sent them all postcards saying "We'll be closing the highway starting next month, please find an alternate route", and while not everybody got the hint, enough people did that there wasn't a problem. At the time, the most cost-effective reader technology for that quality of image wasn't OCR - it was prison convicts reading license plates instead of making them.
These days the automated systems are fast and cheap, and don't be surprised if tollbooths and similar locations have license-plate readers. And don't be surprised if they start correlating cellphone numbers with license plates. San Francisco Bay Area currently has traffic cameras mounted on most of the major highways, and new automated signs saying things like "Time to downtown: 17 minutes". I don't know if they're just looking at average traffic speed at multiple points, or tracking license plates as they go from Point A to Point B, or reading cellphone signals, but all of them are fairly practical methods for getting timing.
Bitrates are usually measured in powers of 10, with kilo=1000, but they're actually rounded values in practice. typically a multiple of a power of 2 * 1000 for telco circuits. A "2 Mbps" E1 line has a raw speed of 2048 kbps, but there are framing channels that often reduce that to 1984 or sometimes 1920 kbps. A "1.5 Mbps" T1 line is really 1544000 bps, or 1536000 after framing. For Ethernets, it's 10/100/1000 Mbps, where those are straight 1000000 bps Megabits per second.
Yes, that's common over here too, but I'm suggesting the opposite transaction - buying the business-rated service for residential use. You might get overcharged, but you won't get kicked.
Unless I've dropped some zeros or 8s, 100GBytes/month is only about 256 kilobits per second fulltime (ok, 277 for a 30-day month, but 256kbps is a nice round size that telcos sell.) A business that bought a 2 Mbps E1 line and got spanked for using it more than 1/8 of the time would quickly find another ISP. I don't know the prices for an E1 in the UK; a 1.5 Mbps T1 in the US is typically under $500 including access. Fractional-speed service is more expensive per bit, of course, but you may still be able to find a good price for the box you hang your P2P service on, and then use cable or DSL for web browsing.
Different molecules have different energy levels in them - you extract energy by combining or splitting them. Different reactions also require a certain amount of energy to make them happen - you can have a reaction that will produce net energy but needs a certain temperature or amount of energy to get it started, and what catalysts do is provide alternate paths for the reaction to happen with less starting energy or lower starting temperatures.
Combining Carbon with Oxygen or Hydrogen with Oxygen produces energy - but splitting up a chain of carbon and hydrogen to get the individual atoms to do that with requires some energy, though it's a lot less than burning the C and H will provide. Catalytic Converters on cars take the unburned hydrocarbons in the exhaust, split them and burn them before they get out the exhaust pipes, and take partially burned carbon monoxide and finish burning it. It's a waste of energy, but it was going to be wasted anyway - the reason to do this is that hydrocarbons and CO lead to air-pollution problems including smog. (They also split various nitrogen oxides to give nitrogen and oxygen; I don't know if this is exothermic or if it's using heat generated by the other reactions.)
You can't split the CO2 up into C and O2 without putting back the energy you got out of that reaction, so a catalytic converter won't help you. You could do things like combine it with calcium oxide to make calcium carbonate, and store that, but the usual way to make calcium oxide is by heating calcium carbonate to get rid of the CO2, so that's really no help.
SMTP senders are one obvious thing to measure (either by number of addresses or by message volume), though of course relaying obscures the real sources. URLs for responses are another - for a while those were largely in China, but now zombies are starting to provide those services. Following the money is really the fun part, but that one's hard, and of course that's easy for a spammer to obfuscate (e.g. open a small corporation in a tax-haven country to receive the loot, and launder the profits by buying things from the real spammer at high prices or selling them below-market.)
But identifying whatever characteristics of the spammer that you can helps you figure out tools to deal with the problem. For instance, my ISP lets me block or filter email by sending country - I don't know anybody in China, Korea, or Nigeria, so I block all SMTP sent from there, and I don't know many people from Japan, so I accept email from there but put it in the extra-filtering bucket. It cut down my spam significantly when I started, though of course spammers continue to be clever. ["Rule 2" says that they're stupid, but that doesn't mean that they're not also clever....]
Sorting by countries does provide some suggestions about where or whether regulation might be useful - though in practice it's usually not helpful, e.g. the US You-CAN-SPAM law, and China's Great Firewall. But it also lets you look at different markets and different technologies - for instance, even though adding more oppressive regulations in China would be a Bad Thing, getting the China Netcom and China Telecom duopolists to provide details about specific spammers or respond to large volumes of complaints can be good. Knowing that Cable Modems in the US or Elementary Schools in Korea tend to have infected machines can help you know what DNSBLs you might want to use, or which ISPs to send complaints to.
Insert standard "You're proposing the following [x] well-known solutions, and they won't work for the following [x] well-known reasons" checklist here....
If you're proposing charging for email, you need to think about who's charging whom for doing what - if you get it wrong, then it's doomed to fail, but if you identify the economic actors and actions correctly, then people may or may not use your system but at least they won't hate you.
The fundamental transaction is that the reader is charging the sender for reading the mail and setting the price based on the sender, and the market will determine whether any senders are sufficiently interested in installing the payment software and paying the price to reach a given user, and the market will also determine whether a typical reader alienates too many of his friends and acquaintences to keep using it. The price might not be cash - it might be a Captcha image recognition, or even a simple "click here to acknowledge that you received my handshake mail", or it might be some token or micropayment system that might or might not succeed in the market.
You can deploy this kind of system for *your* *own* email without forcing everybody else in the world to adopt it. If the price and required effort to send you mail is non-zero, you'll eliminate most spammers right away, but you'll also eliminate some people you did want to hear from; it's your choice and theirs. And you're providing a valuable service to the world by testing the various kinds of software like this that are already out there, and by helping determine the market price for your attention.
The ISP already *has* a box connecting to the user, whether it's a DSL router or a cable modem, and they've got another box just upstream of that which can do anything the customer's-home box can't do. DSL routers and such vary significantly in their capabilities, but almost all of them have far more features than the ISPs or end users actually use. The real difficulty is management - you need to build some kind of web-based admin system so the users who want to turn things on and off can do so, and validation systems so users don't lock themselves out.
Teergrube is the original German Tarpit system. It does run correct SMTP, but vvvv....eeeee......rrrr......yyyyy.....s...llll... ooooooo....wwwww....llll....yyyyy ; it doesn't take much to keep a TCP connection busy, and it doesn't take much to keep a correct SMTP implementation busy (if the spammer is using the zombie's own mail server, though sometimes the spammer is using customer spamware SMTP senders that don't pay enough attention to responses for traps to work.)
Tarpits are a fine thing to do with a spare IP address or a machine that wasn't going to run SMTP (unless your ISP blocks port 25, of course:-) - set up a domain name or subdomain that points to the tarpit, splatter some email addresses around the net for the harvesters to find, and start tracking IP addresses. It's more fun if the spammer thinks you're an open relay and starts pumping lots of their other spam into your tarpit, but that's *so* five years ago; you can still be a good target address.
I get really tired of the "I've only got two broadband providers available" complaints by US residents. Sometimes you've only got one good choice - cable modem - if you're too far from your telco CO to get DSL, and the wireless providers in your area don't have high-speed data yet (or you don't want to pay their cluelessly high prices) and you think satellite latency is too slow for your applications and where you live is too hilly to get an 802.11 directional antenna to hit anything useful.
But if you've got DSL, almost anywhere in the US, you've got lots of providers who will sell you service, and they're the ones who set open/closed/AUP policies even though many of them are selling telco-provided connectivity underneath their service. Some, like Covad, are renting dry copper and running their own DSLAMs; some are using telco-provided Layer 3 IP with PPPoE, and many are using telco-dslam Layer 2 ATM connections. They may charge a bit more than the latest sale price from your telco (though those deals usually seem to be 3-months-cheap customer-acquisitiion scams rather than long-term low prices), but if you want an ISP that allows you to run your own Linux mail server, lets you share your wireless with your guests or neighbors, lets you run your own web server, lets you use multiple computers in your house, doesn't censor what kind of content you send (except for spam, of course), doesn't interfere with VOIP, gives you a range of services like mailboxes, shell, web+ftp space - all of those things are commonly available. I use Sonic.net, and Speakeasy is similar and better known, but hundreds of small ISPs have this as part of their business model. Most of them aren't supporting ToS or DSCP traffic prioritization yet, but some of them probably are figuring out that there's a market for it.
And if that's not flexible enough for you, you can get a cheap cable modem or telco DSL and get some hosting space from thousands of places around the net. Some of them are selling cheap virtual web hosting ($1-10/month), some are doing Linux Virtual Machines (typically $20-30/month), some are fancier. If your cable/DSL access provider limits the services you can run, you can run an IPSEC or other tunnel to your colo server.
The original Orion proposals were nukes from the ground up, and hope there wasn't too much fallout; the revisionist idea of using conventional rocketry to get the building materials into LEO and then firing the nukes where fallout wouldn't matter would be horrendously impractical. Maybe you could build an interplanetary Orion on a Moon base if you had one of those, though of course hauling thousands of nukes to the moon has its own risks of catastrophic failure.
And of course that doesn't even *begin* to count the *serious* risks, like what happens if you develop nice convenient little Mr. Fusion Hand Grenades and an assembly line to produce them by the tens of thousands, or the risks that doing enough nuclear explosives research to get the right size Project Orion fuel charge means the Weapons Of Mass Destruction people get to reuse any test design work for whatever other applications they can think of.
I want to have my data encrypted, but I want my guests to be able to access the net, and if passersby want to freeload a bit, that's fine too. If I set my SSID to "password==guest" and let people log in with "guest" as the authentication system password, will my data channel still be usefully encrypted, or does knowing the access password let people eavesdrop on my connections?
A friend of mine used to leave his wireless open, figuring it was his civic duty to let the neighbors' teenagers have uncensored net access. It was fine for a while, but eventually somebody started doing too much P2P traffic and his wireless performance got unusable, so he closed it down.
In my building, there are at least two unencrypted wireless connections. One's mine; the other is a neighbor with a Belkin node. I can see 3-4 other encrypted connections. Occasionally something goes wrong with my connection (DSL flakiness or whatever) and I'll piggyback off my neighbor for a few days, and of course occasionally my laptop decides it would rather use my neighbor's signal instead of mine for no particular reason, which tends to disconnect my work VPN.
Maybe you've forgotten, but the French-Vietnamese war in 1954 wasn't "France defending the world's freedom from Communist Domino-Tippers" - it was a colonialist empire getting kicked out by nationalists, just as it got kicked out of Algeria a few years later. The nationalists found they got a better deal from the Communists so they became Commies, though of course part of the deal was "having the huge country of traditional enemies who live next door being on your side". The US went in to "help" stomp out the freedom fighters, without any success either, supported some dictators and assassinated others, and of course down in the south, the Commies were anti-freedom fighters, not that the dictators were all that pro-freedom, and a few decades of war in neighboring Cambodia led to a group of psychotic killers who made Mao's Cultural Revolutionaries look like nice guys.
At least the French had the sense to get out when it was hopeless, rather than continuing to escalate because backing down would send a message that we were wimps and ensuring that Communism would continue to oppress people for decades afterwords, though it's mostly gone now. All in all, it was another fine moment in militarist history.
My initial impression of "hordes of animals at De Gaulle" was the crowds of people waiting in various lines to buy tickets or go through customs or whatever. But you're probably talking about all the rabbits hanging around the outside green spaces? CDG also had *lots* of plane-spotters hanging around when I was last there - kind of a weird obsession, but it was plane-spotters who provided a lot of the evidence about CIA planes hauling kidnapped people around to secret prisons.
At least when the French were taken over by the Germans, they had the excuse that a large foreign army had found the flaws in their defenses and stomped on them. When the Republican Party and then the US were taken over by an organization aggressively opposed to the traditional values of liberty, small government, fiscal responsibility, and common sense, it was done by propaganda and money, except for a small attack by foreign enemies and some recycled military anthrax, and the large army, police, and spying organizations that are now attacking the US were all our own.
There has been a lot of excellent French engineering, though there's also a tradition of central planning by "experts", which led to things like the Maginot Line, attacking Russia in winter, suppression of regional languages, and a lot of the 19th-century mercantilism. Hopefully the robotics engineering will more resemble the social structure of French painting...
I was out in Maui about 2 years ago, and the restaurant by the aquarium in Maalea had "Freedom Fries". Maybe some of the US mainlander tourists liked that sort of thing, but presumably the rest of the tourists were annoyed at the stupid American jingoists. And Hawaii is so thoroughly multicultural that it's much more ridiculous there that it would be in Iowa or something, though most of the European-ancestry people in Hawaii are either Portugese or Anglos (or hippies.)
My mother was appalled - she's a politically active liberal Republican who didn't like either George Bush, and while she'd heard the politicians doing their anti-French rant, she hadn't seen anyone actually believing it back on the East Coast. I hadn't seen any Free Dumb Fries here in Silicon Valley either, but it's a sufficiently blue state that I wouldn't expect to find them much closer than maybe Fresno.
Fine, they're cool toys, but that's about it. Humanoid shapes are useful for non-specialized robots that need to be able to do a wide variety of things, and two-foot-tall robots aren't able to do many of the things a larger robot could.
For instance, if you want the robot to fetch you a beer, it needs to be able to reach into the refrigerator.
If you want a robot to vacuum your floors in your uncluttered house, you can get a specialized robot like Roomba, or you can have a general-purpose robot push a vaccuum cleaner that's designed for humans to push.
If you want to vacuum the floors in a cluttered house, the humanoid robots's got some advantages, because it can move things around, but it needs to be strong enough to do the job, not tip over much, and able to see things on top of tables if it's moving them there, so again a 2-foot-tall robot isn't useful.
If you want a robot to change the channels on the TV after it fetches your beer, humanoid robots aren't really that helpful - a Tivo-shaped robot is better at the job, or you might want voice recognition in your TV or Tivo to tell it to change channels; maybe the humanoid robot can fetch DVDs and put them into the DVD player, but it's probably not worth the trouble, and certainly not compared to a large hard disk...
Newsforge has no technical information, and Freeswitch is largely Slashdotted, but there's one sentence that says that they're using the Speex Wideband Codec as their 16kbps codec. One reason Speex is using 16khz sampling is because it's relatively available on PC sound cards, but another reason is that they do a cute sub-band coding technique - instead of representing the 8kHz analog waveform by directly encoding the 16k samples/second, they split the information into two bands - 0-4kHz which they encode using the same encoding they use for their 8kbps codec, and 4-8kHz which they encode (somewhat differently, for complex technical reasons:-) to provide additional depth for receivers that support the wideband format. So if you've got a wideband codec encoding the speech at 16kbps, you can play it on an 8kbps player if that's all you've got. For a live conversation between two software-phones, that's not particularly useful (except for a bit of code reuse), but if you're playing recorded files or setting up a multipoint conference between some 8kbps phones and some 16kbps phones, it's easy to send each phone what it wants.
Don't know where you got your information, but it's unfortunately incorrect. There are some of the early ADPCM codecs (e.g. 32kbps) that do the same transformations every time, so you can convert from 64 to 32 to 64 to 32 to 64 again without additional damage, but most of the newer high-density codecs take a significant hit if you do them two or more times, e.g. 64kbps to 8 to 64 to 8 to 64. I've forgotten the precise MOS scores, but the standard G.729 codec family goes from "better than a cellphone with a decent headset" at the initial compression to "worse than a cellphone in bad traffic" with two 8kbps compressions to "cellphone in a tunnel full of trucks during a lightning storm" with three.
The one case where several conversions are safe is converting from the original G.711 64kbps format down to almost any 8kbps or 6.3kbps format and back to 64kbps. The reason that mobile-to-mobile calls can work better than that is that most of them use the same codecs, so they can usually avoid conversion if you're going from one GSM phone to another. I don't know if this works when connecting a GSM carrier to a CDMA carrier or not. Also, of course, mobile phones usually have lousy little microphones and tinny little speakers, so much of the audio damage is done at the ends rather than by the codec itself, especially if you're using it in traffic.
Slashdot Mirror
Back when I had a lab, there was a time that the A/C system in the ceiling croaked and they had a free-standing A/C unit as a stopgap until they got it fixed. It had a bunch of tubes in the back and an opening in the front where the cold air came out. And my department had a sales guy who was a wine expert, had a small Napa Valley winery with a couple of friends, and did occasional evening winetastings after work. So there was obviously one thing we had to do with the A/C unit, which was to chill a couple of bottles of interesting white wine, as a change from our usual reds. Worked real fine.
It's getting to be time to update my Mozilla Suite anyway - is 2.0xx cooked enough to use, or is it better to go to 1.5.0.5 and wait for 2.0 final to update again?
Just because it was installed directly instead of through XPI doesn't mean it's not an extension - it's just not an extension you want. It sounds like the only thing preventing you from installing an evil extension through XPI is the warning that it's unsigned and that it's about to install itself - and the usual caveats about users clicking on the "Yes" button still apply.
Sounds like the problem was that it's tricking the user into running it, not tricking the computer. Hard to fix that sometimes.
David Brin's book "The Transparent Society" (1998, online excerpts) talks about the effects of cheap computing/camera/database technology on privacy. It's a pre-9/11/2001 look at what societies and governments can do with Moore's Law kicking technology. A major point was that either society forces the government to be open about what it's doing and allow the public to watch it, or else the government will use all the same technology _without_ anybody watching it. Now, of course, we've got the Bush Administration, so all the happy 90s-boom speculation about "Well, what if another president as evil as Nixon got elected?" "No way, the public wouldn't let that happen again!" is moot, and we didn't get the transparency locked down when we could.
These days the automated systems are fast and cheap, and don't be surprised if tollbooths and similar locations have license-plate readers. And don't be surprised if they start correlating cellphone numbers with license plates. San Francisco Bay Area currently has traffic cameras mounted on most of the major highways, and new automated signs saying things like "Time to downtown: 17 minutes". I don't know if they're just looking at average traffic speed at multiple points, or tracking license plates as they go from Point A to Point B, or reading cellphone signals, but all of them are fairly practical methods for getting timing.
Bitrates are usually measured in powers of 10, with kilo=1000, but they're actually rounded values in practice. typically a multiple of a power of 2 * 1000 for telco circuits. A "2 Mbps" E1 line has a raw speed of 2048 kbps, but there are framing channels that often reduce that to 1984 or sometimes 1920 kbps. A "1.5 Mbps" T1 line is really 1544000 bps, or 1536000 after framing. For Ethernets, it's 10/100/1000 Mbps, where those are straight 1000000 bps Megabits per second.
Yes, that's common over here too, but I'm suggesting the opposite transaction - buying the business-rated service for residential use. You might get overcharged, but you won't get kicked.
Unless I've dropped some zeros or 8s, 100GBytes/month is only about 256 kilobits per second fulltime (ok, 277 for a 30-day month, but 256kbps is a nice round size that telcos sell.) A business that bought a 2 Mbps E1 line and got spanked for using it more than 1/8 of the time would quickly find another ISP. I don't know the prices for an E1 in the UK; a 1.5 Mbps T1 in the US is typically under $500 including access. Fractional-speed service is more expensive per bit, of course, but you may still be able to find a good price for the box you hang your P2P service on, and then use cable or DSL for web browsing.
Combining Carbon with Oxygen or Hydrogen with Oxygen produces energy - but splitting up a chain of carbon and hydrogen to get the individual atoms to do that with requires some energy, though it's a lot less than burning the C and H will provide. Catalytic Converters on cars take the unburned hydrocarbons in the exhaust, split them and burn them before they get out the exhaust pipes, and take partially burned carbon monoxide and finish burning it. It's a waste of energy, but it was going to be wasted anyway - the reason to do this is that hydrocarbons and CO lead to air-pollution problems including smog. (They also split various nitrogen oxides to give nitrogen and oxygen; I don't know if this is exothermic or if it's using heat generated by the other reactions.)
You can't split the CO2 up into C and O2 without putting back the energy you got out of that reaction, so a catalytic converter won't help you. You could do things like combine it with calcium oxide to make calcium carbonate, and store that, but the usual way to make calcium oxide is by heating calcium carbonate to get rid of the CO2, so that's really no help.
URLs for responses are another - for a while those were largely in China, but now zombies are starting to provide those services.
Following the money is really the fun part, but that one's hard, and of course that's easy for a spammer to obfuscate (e.g. open a small corporation in a tax-haven country to receive the loot, and launder the profits by buying things from the real spammer at high prices or selling them below-market.)
But identifying whatever characteristics of the spammer that you can helps you figure out tools to deal with the problem. For instance, my ISP lets me block or filter email by sending country - I don't know anybody in China, Korea, or Nigeria, so I block all SMTP sent from there, and I don't know many people from Japan, so I accept email from there but put it in the extra-filtering bucket. It cut down my spam significantly when I started, though of course spammers continue to be clever. ["Rule 2" says that they're stupid, but that doesn't mean that they're not also clever....]
Sorting by countries does provide some suggestions about where or whether regulation might be useful - though in practice it's usually not helpful, e.g. the US You-CAN-SPAM law, and China's Great Firewall. But it also lets you look at different markets and different technologies - for instance, even though adding more oppressive regulations in China would be a Bad Thing, getting the China Netcom and China Telecom duopolists to provide details about specific spammers or respond to large volumes of complaints can be good. Knowing that Cable Modems in the US or Elementary Schools in Korea tend to have infected machines can help you know what DNSBLs you might want to use, or which ISPs to send complaints to.
If you're proposing charging for email, you need to think about who's charging whom for doing what - if you get it wrong, then it's doomed to fail, but if you identify the economic actors and actions correctly, then people may or may not use your system but at least they won't hate you.
The fundamental transaction is that the reader is charging the sender for reading the mail and setting the price based on the sender, and the market will determine whether any senders are sufficiently interested in installing the payment software and paying the price to reach a given user, and the market will also determine whether a typical reader alienates too many of his friends and acquaintences to keep using it. The price might not be cash - it might be a Captcha image recognition, or even a simple "click here to acknowledge that you received my handshake mail", or it might be some token or micropayment system that might or might not succeed in the market.
You can deploy this kind of system for *your* *own* email without forcing everybody else in the world to adopt it. If the price and required effort to send you mail is non-zero, you'll eliminate most spammers right away, but you'll also eliminate some people you did want to hear from; it's your choice and theirs. And you're providing a valuable service to the world by testing the various kinds of software like this that are already out there, and by helping determine the market price for your attention.
Tarpits are a fine thing to do with a spare IP address or a machine that wasn't going to run SMTP (unless your ISP blocks port 25, of course :-) - set up a domain name or subdomain that points to the tarpit, splatter some email addresses around the net for the harvesters to find, and start tracking IP addresses. It's more fun if the spammer thinks you're an open relay and starts pumping lots of their other spam into your tarpit, but that's *so* five years ago; you can still be a good target address.
But if you've got DSL, almost anywhere in the US, you've got lots of providers who will sell you service, and they're the ones who set open/closed/AUP policies even though many of them are selling telco-provided connectivity underneath their service. Some, like Covad, are renting dry copper and running their own DSLAMs; some are using telco-provided Layer 3 IP with PPPoE, and many are using telco-dslam Layer 2 ATM connections. They may charge a bit more than the latest sale price from your telco (though those deals usually seem to be 3-months-cheap customer-acquisitiion scams rather than long-term low prices), but if you want an ISP that allows you to run your own Linux mail server, lets you share your wireless with your guests or neighbors, lets you run your own web server, lets you use multiple computers in your house, doesn't censor what kind of content you send (except for spam, of course), doesn't interfere with VOIP, gives you a range of services like mailboxes, shell, web+ftp space - all of those things are commonly available. I use Sonic.net, and Speakeasy is similar and better known, but hundreds of small ISPs have this as part of their business model. Most of them aren't supporting ToS or DSCP traffic prioritization yet, but some of them probably are figuring out that there's a market for it.
And if that's not flexible enough for you, you can get a cheap cable modem or telco DSL and get some hosting space from thousands of places around the net. Some of them are selling cheap virtual web hosting ($1-10/month), some are doing Linux Virtual Machines (typically $20-30/month), some are fancier. If your cable/DSL access provider limits the services you can run, you can run an IPSEC or other tunnel to your colo server.
And of course that doesn't even *begin* to count the *serious* risks, like what happens if you develop nice convenient little Mr. Fusion Hand Grenades and an assembly line to produce them by the tens of thousands, or the risks that doing enough nuclear explosives research to get the right size Project Orion fuel charge means the Weapons Of Mass Destruction people get to reuse any test design work for whatever other applications they can think of.
Nonetheless, it was *way* *fscking8 *cool*.
I want to have my data encrypted, but I want my guests to be able to access the net, and if passersby want to freeload a bit, that's fine too. If I set my SSID to "password==guest" and let people log in with "guest" as the authentication system password, will my data channel still be usefully encrypted, or does knowing the access password let people eavesdrop on my connections?
In my building, there are at least two unencrypted wireless connections. One's mine; the other is a neighbor with a Belkin node. I can see 3-4 other encrypted connections. Occasionally something goes wrong with my connection (DSL flakiness or whatever) and I'll piggyback off my neighbor for a few days, and of course occasionally my laptop decides it would rather use my neighbor's signal instead of mine for no particular reason, which tends to disconnect my work VPN.
At least the French had the sense to get out when it was hopeless, rather than continuing to escalate because backing down would send a message that we were wimps and ensuring that Communism would continue to oppress people for decades afterwords, though it's mostly gone now. All in all, it was another fine moment in militarist history.
My initial impression of "hordes of animals at De Gaulle" was the crowds of people waiting in various lines to buy tickets or go through customs or whatever. But you're probably talking about all the rabbits hanging around the outside green spaces? CDG also had *lots* of plane-spotters hanging around when I was last there - kind of a weird obsession, but it was plane-spotters who provided a lot of the evidence about CIA planes hauling kidnapped people around to secret prisons.
There has been a lot of excellent French engineering, though there's also a tradition of central planning by "experts", which led to things like the Maginot Line, attacking Russia in winter, suppression of regional languages, and a lot of the 19th-century mercantilism. Hopefully the robotics engineering will more resemble the social structure of French painting...
My mother was appalled - she's a politically active liberal Republican who didn't like either George Bush, and while she'd heard the politicians doing their anti-French rant, she hadn't seen anyone actually believing it back on the East Coast. I hadn't seen any Free Dumb Fries here in Silicon Valley either, but it's a sufficiently blue state that I wouldn't expect to find them much closer than maybe Fresno.
Newsforge has no technical information, and Freeswitch is largely Slashdotted, but there's one sentence that says that they're using the Speex Wideband Codec as their 16kbps codec. One reason Speex is using 16khz sampling is because it's relatively available on PC sound cards, but another reason is that they do a cute sub-band coding technique - instead of representing the 8kHz analog waveform by directly encoding the 16k samples/second, they split the information into two bands - 0-4kHz which they encode using the same encoding they use for their 8kbps codec, and 4-8kHz which they encode (somewhat differently, for complex technical reasons :-) to provide additional depth for receivers that support the wideband format. So if you've got a wideband codec encoding the speech at 16kbps, you can play it on an 8kbps player if that's all you've got. For a live conversation between two software-phones, that's not particularly useful (except for a bit of code reuse), but if you're playing recorded files or setting up a multipoint conference between some 8kbps phones and some 16kbps phones, it's easy to send each phone what it wants.
The one case where several conversions are safe is converting from the original G.711 64kbps format down to almost any 8kbps or 6.3kbps format and back to 64kbps. The reason that mobile-to-mobile calls can work better than that is that most of them use the same codecs, so they can usually avoid conversion if you're going from one GSM phone to another. I don't know if this works when connecting a GSM carrier to a CDMA carrier or not. Also, of course, mobile phones usually have lousy little microphones and tinny little speakers, so much of the audio damage is done at the ends rather than by the codec itself, especially if you're using it in traffic.