Cable Modem vs. DSL vs. Business Connections
on
Wireless Freenets
·
· Score: 3
You'd be surprised how many people can share a megabit of bandwidth if they're only doing email and web traffic, not file system sharing or good-quality video or serving popular web sites. The only places there are serious bandwidth issue are modems (duh) upstream bandwidth on cable modems, and the cable companies have responded to this by serious Fear, Uncertainty, and Doubt policies (IMHO, they're stupid and self-defeating - they need to find every approach they can to get users to want always-on fast bandwidth.) Some DSL companies feel the same way, but many have more of a clue. The FCC's basically not involved in this, and there's no reason they will be. Cable companies can do nastier FUD without FCC help, and they really want the FCC to stay out of their game, especially as they try to roll out telephony ([Expletive deleted]CALEA FBI wiretap rules seem to apply to them, unfortunately.) And telcos only care about connect time on modems, not traffic levels - some of them are playing PR games discouraging connect time, especially if they also sell DSL, while others have figured out that they make more money selling second phone lines to modem users.
It's true that the FCC are a bunch of spectrum thieves, who nationalized spectrum in the Roosevelt days to protect the big-money communications companies from competition (even if they made lots of hype about protecting the public's interest in the public's airwaves), and US and European spectrum regulators figured out that the hype about Next Generation Wireless Services could be used to put a big hidden tax on wireless telephony and low-speed data services, which is showing up as huge debts by cellphone companies, just in case any of them weren't getting themselves into debt trouble investing in the fiber bandwidth glut:-) In some countries that will play out as higher wireless costs; in other countries the wireless costs will be enough higher that they'll kill off any of the overhyped high-speed wireless data services and the spectrum buyer will go bankrupt or give up and resell their spectrum at a big loss.
Fortunately, there's enough unlicensed spectrum to build some reasonable collections of services, but it'll take a lot of coordination. Metricom / Ricochet tried for a while; unfortunately they couldn't make enough money at it, but maybe another generation of providers will succeed, using faster commodity equipment. It's possible to do freenets if you can find a way to coordinate them (tough). But Starbucks is starting to offer commmercial wireless services for $X/month, and so are a few other companies, targeting either the coffeeshop market or the airport market, where there are enough business users with laptops and possibly wireless networks. Not much help if you live in the burbs, but here in San Francisco there's a law requiring a Starbucks on every other block, so if you live in the dot-com live-work loft district, that may be an attractive way to get service. For urban residential areas, where there's enough density for wireless nets to work, it's hard to say whether freenets or businesses will be more successful.
The 5GHz stuff I've seen descriptions of generally has shorter range and higher price than 2.4GHz. Some of that's just new-equipment pricing, and will come down in price as there's more of it on the market and development gets amortized more, but the bandwidth is partly a tradeoff with distance and number of simultaneous users. Besides, if you're feeding the thing from a DSL line or 10Mbs Ethernet or in-between cable modem, it doesn't make much difference whether you've got 11Mbps or 22, 54, 70, or 108 Mbps.
Software is also an issue - 802.11b tries very hard to look like an Ethernet card with a few extra hooks, but the 802.11a stuff wants more of the complex processing done by the CPU, in a WinModem-ish fashion, which says there'll be some major issues getting Linux drivers done for a while unless one of the two maufacturers decides it's worth helping out a lot
Moving the Huge Aircrash factories out of LA would be a highly expensive proposition - it's not just cubicles and people. But there's a much easier way to fix the problem, which is to create a non-California subsidiary corporation to own the satellites and transfer the satellites to them. At that point, there's no longer any movable property involved, just paper. Sure, it requires moving a few lawyers and managers around, but that's pretty simple, and it's far less bizarre than some of the other corporate tricks used by a non-profit medical foundation that builds satellites. I don't know if they can do this with just a Delaware or Nevada corporation, or if they need to go to some Caribbean tax haven to solve the tax problems (since that affects US corporate taxes as well as the California problems.)
Basically, if the LA tax goons push too hard, they'll just lose, and make themselves look stupid as opposed to just greedy.
PDF files are usually even wronger than MSWord - they capture the printed output of a file while losing the structure and editability, so any work put into creating a PDF form is pretty much lost.
HTML and its relatives retain the structure of the document, they're editable and cut&paste-able, and even though MSWord generates ugly clueless HTML code, you can still reuse stuff.
The social engineering process you use is worthwhile, though.
Wait a minute here! Zinc Oxide is the stuff you use to keep damaging UltraViolet sunlight off your face. Now they're using it to turn it into UV Lasers? I can feel my nose burning already. Ouch!
If you want to use Microsoft's TCP/IP stack and a Winsock interface, building a spoofing DDOS is probably difficult. But that's the wrong way to do it, and a guy who writes assembler and reads RFCs should realize it. A good TCP/IP stack gives you good performance for two-way communications. Many DDOS programs don't need that, unless they're pretty sophisticated. They just need to build credible-looking packets and send them out the Ethernet or PPP driver, which is fundamentally Not That Hard. SYN-flood attack packets don't require you to read the ACK response - that's part of the point. Dumb UDP attacks don't need to either. Fancy attacks which react adaptively to the responses of the target program do need to read the responses, so they're tough to implement except by using the MS stack - things like attacking web servers using TCP/IP. But dumb flooding attacks don't care.
Disc-lamer: I *am* making much of this up. I haven't tried to turn this into working code. Take it with several grains of salt. But the principles are pretty much correct.
One of the crackers that Bill Cheswick and Steve Bellovin caught trying to break into the AT&T Bell Labs firewall was a kid from the Netherlands. Apparently with Netherlandish law at the time, he wasn't doing anything illegal, so they did something more direct and effective to resolve the problem:
When the guy said it would take a thousand computers over 10 years to decrypt, it's excessively unlikely that he was trying to be scientifically precise in a way that you can calculate the real encryption strength from. He was making up hype numbers for a press release that were intended to give the general public a feel for how hard the problem is. So don't try calculating whether it's really 10**17 keys vs. 10**38. He said "it's really really hard to crack", and his hype numbers happen to be low rather than high.
I usually give crypto-cracking speeds (for adequately strong algorithms) in terms of planet-sized computers and billions of years, because that's obviously infeasible to crack, and if it's not, you should have made the keys a few bits longer. For RC4, that doesn't even cost you anything:-) Since you know how to calculate using exponentials, keep in mind that given good algorithms, it's trivial to make things that take that long to crack, and are so far out of reach of intelligence agencies that you should be worrying about other threats, like keyboard sniffers planted in your phone or passwords on yellow sticky notes. Single-DES can be brute-forced - John Gilmore proved that with the EFF Deep Crack machine, and the distributed crackers also showed they can do it. But Triple-DES isn't just 3 times as hard - it's 2**56 times as hard (total strength is only 112 bits, not 168, because there's a meet-in-the-middle attack that uses 2**56 pieces of memory, which is currently impractical.) RC4 is adjustable from near-0 to 255 bits of key length, with much less work per key brute-forced, but 128 bits is enough. The new US NIST Advanced Encryption Standard (contest won by Rijndael from Belgium) has modes for 128, 192, and 256 bits, if I remember correctly - even the weakest mode is strong enough for Earth-bound attackers.
The hard part of the crypto isn't the symmetric algorithm - it's the public-key part. I suppose they *could* have used 128-bit algorithms for that, but Elliptic Curve isn't strong enough at that length, and they'd be expected to know it. If you're not worried about traffic analysis, you could build a Kerberos-like system using 3DES or AES that fits in 128-bit keys.
There's plenty of horsepower handy - most of the work is compressing the voice, and once you've done that, encrypting ~6.5kbps or 13kbps is pretty trivial. The trick is to get the cell phone system to complete handset-to-handset calls using the compressed digital voice stream, which would let you pass encrypted compressed voice between the phones instead of unencrypted, and prevent it from converting the compressed voice to conventional uncompressed voice and back, which would obviously trash the encrypted data. I think GSM may let you do that - I know some of the digital versions don't, and obviously analog also has its limitations. All you'd need would be a sufficiently programmable phone to make it happen. Otherwise, any encryption in the phone happens between the phone handset and the cell site, which is useless for end-to-end because you can't modify the cell site.
The other obvious approach is to add a cellular modem to the cellphone, as long as it can get at least ~6.5kbps of throughput (one of the tighter compressions used in US digital cellphones) and set up a modem call. This needs a bit more hardware, but modems can be pretty compact, and again you've already done the compression in an ASIC. If you can't get fast enough modem speeds, you either need a tighter but nastier-sounding codec, e.g. 4800 baud or (gak!) 2400 baud or 1200 baud LPC (Speak-And-Spell is a trademark of somebody or other.) Or you can cheat and make a double-sized cellphone that's doing two simultaneous calls - klugey, but if you can afford DM6000 for a phone, you should be able to live with a much cheaper phone that burns minutes twice as fast.
Another approach is to wait for those 3G phones that the EU governments scammed their phone companies into paying giga-Euros of debt money for in the license auctions. Shouldn't cost any DM6K for one of those.
Real Algorithms can be executed in software just as well as hardware, though some things are more efficient on specially-tuned hardware than on general-purpose computers. DES, for instance, uses a lot of ugly bit-twiddling which is annoying to do on typical hardware, so it gains a lot by running on special gate-array designs, but you can still keep a 10Mbps Ethernet or a T1 line pretty full on cheap Pentiums. Voice, on the other hand, only requires about 6-10 kbps for most cell-phone voice compression algorithms, so the load from using DES is much less than the computation used in the voice compression itself. Some of the public-key algorithms can benefit from special hardware designed to do bignum multiplies, which can benefit from a lot of pipelining and parallel computation, so there's some market for accelerator boards to do that for web servers.
But the main reason you'd do crypto in hardware in a cellphone is that callphones tend to do the heavy lifting in ASICs and not have a lot of general-purpose computing horsepower or memory - it's easier to put the crypto into the ASIC than find somewhere else to wedge it.
Check out the Cypherpunks archives on the net.
GSM doesn't use ECC - it uses a couple of algorithms called A5, A8, etc. which look something like a fast fourier transform. Ian Goldberg, a Berkeley grad student, cracked them over lunch one day (he's not Israeli, just Canadian.) The authentication is a bit stronger than the message encryption. One of the entertaining results of the crack was the discovery that, while the keys are too short to start with, most of them have 10 bits set to 0, so they're even easier to crack, which is a strong argument that there was government pressure on the development process.
They're just pod bay doors, Hal, sure you can close them...
...
...
You probably have an easement allowing cable co.
on
Make Way for Fiber
·
· Score: 2
It's highly unlikely that the cable company is doing this using the city's eminent domain powers. Almost always, if utilities are in your back yard, one of the utility companies has an easement to use your property - check your deeds to find out where it is, if you don't remember from when you bought your house. It'll typically say that they've got the right to put stuff along whatever chunk of your land they're using with some width like 6 feet. If you didn't sell them the easement directly, probably the builder who you bought your house from did (since you're in a new neighborhood.) If they're in the front of your house, either they've got an easement there, or perhaps the city actually owns the street and the chunk of grass between the street and sidewalk, or has an easement for that area. Again, it's usually not eminent domain, it's a deal the builder did.
And wiretapping your cable company to steal internet access is likely to be difficult, depending on which protocols your cable provider is using - even if the transmissions aren't encrypted, the address assignment is usually tied into the billing mechanism somehow. You're better off paying your neighbor to put in an 802.11b wireless LAN instead.
There are a lot of interesting bodywork techniques out there from people who are usually very much unlike most hackers I know. (We and they both tend to be intuitive, but in much different directions.) A friend of mine teaches Feldenkrais, and I've found it very interesting for giving me insight into how I move, particularly the aspects beyond the obvious large motions that are more intentional. A lot of that stuff is subconscious and habitual. Rolfing is somewhat of a dual to Feldenkrais - it deals more with alignment and position rather than motion, and was at least originally a more forceful technique than Alexander or Feldenkrais which are gentler and more exploratory.
I'll probably go back to a chiropractor, since some of that insight is how my spine and hips move together, and that's some thing chiropractors mess with a bit more directly that may help me benefit from some of the other techniques. Also, now that I'm getting my lazy butt into the gym more often, I'm finding that alignment issues affect my running, and the hip bone really *is* connected to the leg bone, etc.
Beds come in really wide varieties, and what matters isn't some absolute that some sales person is trying to push, but what's good for you, with your body and your current condition. Good beds are hard to define, though really bad beds are pretty easy to recognize:-) Sometimes just variety makes a difference - staying in a hotel or bed&breakfast with good beds (as opposed to staying in one with bad beds, or sleeping on airplane seats, or those beds your parents have had since you were a kid when you're home visiting them... When I visit my mother-in-law, I usually sleep on the floor because it beats her sofa-bed.)
Some of the newer foam technology is really nice, and helped a lot for some of my problems. I don't know if waterbeds are a good idea long-term - some people swear by them, some people view them as leftover hippie stuff (some of us view them nostalgicly as old hippie stuff:-) The newer versions give you ways to cut down on sloshing around and make the things easier to handle. I've never been that impressed with futons, but they work for some people. Some people do well with extra padding, though too much usually means you're not getting enough support, which is bad for your lower back, though it may be just great for your shoulders.
Sounds suspicious to me.... Depending on whether the "centralized" box is really a centralized box run by his company or only a centralized-per-customer firewall-like-thing, it could be a golden opportunity for wiretapping the paranoid, or it could be just watered-down explanations given to the non-technical press by the Corporate Speaker-To-Publicists.
The US government* already has extensive regulations about privacy for US citizens. With few exceptions, they're the main problem, and adding Band-Aids to the system without addressing them is at best a pretense of a solution. The requirements that cause the problems are the ones that create common identifying numbers and databases and require businesses and individuals to use them, which provide the tools for privacy-problematic activities by business and make it economically necessary for businesses to use those tools to be competitive. Some examples:
Same Social Security Number for All Tax Records - By forcing everybody to collect your SSN, everybody who pays you wages or dividends or bank interest has to have that number, and has a relatively strong belief in its uniqueness and long-term persistence, so they can use it as a database key. This is cheaper than doing their own unique keys, and it means that organizations like credit reporting companies can use that one number to track all your bank accounts. That wasn't a big problem in the 1930s-1950s, when records were kept on paper, with occasional help from Business Machines that sorted punch cards that disliked being Bent, Folded, Spindled, or Multilated. Around the 60s, computers started becoming affordable to medium-large businesses that could correlate information with them, though it was still pretty tedious using magnetic tape to handle large quantities of records. By the early 90s, anybody could afford computers faster than the government used in the 60s, and by the late 90s, anybody could afford high-speed storage on their desktops bigger than the off-line storage government could afford in the 60s, and could carry more CPU in their shirt pockets than the government could afford in the 60s, and database queries are no longer an arcane process requiring extensive development budgets - they're just something you type into websearch engines or your PC.
An alternative - Suppose you had a stack of Taxpayer ID numbers, and could give a different number to everybody who needs to know. The Tax Agencies would still be able to coordinate them, since computers and databases are affordable and cheap, but nobody else could.
Medical Records keyed on Taxpayer IDs - In the US, the government provides medical insurance for old people who've paid into the Social Security tax system, and they use the Social Security Number as their database identifiers, and force any medical providers to use that number to be reimbursed for the costs of medical services to old people. Therefore, almost all medical insurance companies use that as their database keys, and the insurance companies and government force doctors to use them as identifiers as well. Furthermore, tax policy strongly encourages businesses to provide their employees with medical insurance, and therefore employers need to use SSNs as their interface to medical insurance companies. And increasing social pressure about making employer-funded medical insurance pay for drugs makes pharmacies use SSNs as well. Do you really want your employer to know what medicines you're taking? It's much harder to solve this one than the tax issues, because the insurance process is very deeply tangled, and because even aside from the money there are medical benefits to sharing of information between anybody who a given patient interacts with.
Driver's Licenses tied to SSNs, Citizenship Papers. The Feds first permitted the states to use SSNs as a database key, and since then they've essentially made it mandatory. This does reduce the extent to which bad drivers maintain multiple licenses so they can still drive after being convicted of bad driving, but it's increasingly being used for enforcement of social policy. For instance, the State of California believes that Driving While Speaking Spanish is unsafe, so they've been requiring citizenship documentation to discourage Un-Americans from getting drivers' licenses. To some extent, this decreases the number of immigrants who get CA drivers' licenses or car insurance, which is directly counter-productive, but it also increases the financial advantage to Motor Vehicle Department employees to accept bribes in return for otherwise-unavailable paperwork services.
Permission To Work Tied To Centralized Databases - First there was the reprehensible policy of requiring anybody who wants to work as an employee in the US to provide Citizenship Papers and fill out forms with the Immigration Cops, but that was basically a one-way information flow. Since then, the Deadbeat Dads law has created a requirement that you not only tell the government you're hiring somebody, but ask permission first, in case they might be a father who's criminally failed to pay child support - even if they're not a Dad, or a Dad Required By A Court To Pay Child Support, or a Deadbeat, you're still required to treat your potential employees as if they might be, and get the government's permission first. This means there's a centralized database of US Citizens Not Permitted To Work, which is relatively simple to query, and it's probably possible for non-government employees to access the database of people who are known to be working - it's certainly simple for government employees to query the database, whether they're from the tax agencies or other parts of the government. This radically increases the consequences of inaccuracies in government databases, and also creates a strong incentive for identity theft by Personna Not Grata, while increasing the bribery potential of people who have access to the data.
Professional Licenses Tied To Central Databases and Social Policy - The Deadbeat Dad stuff has also spawned a requirement that cities and states which grant professional licenses withhold them from anybody who might be Listed, which has similar effects to the other privacy-reduction regulations. The intent is to force any Deadbeat Dads who are actually making money to pay up, which is fine, but the consequences for non-deadbeats can be significant.
Draft Registration - yes, it's been decades since the US government got into a war that required more cannon fodder than the politicians have been able to get volunteers for. But after the Vietnam Police Action ended, and the authorization for a draft ended, the Pentagon was able to talk Congress into re-creating the regulation for universal registration of young men, and they do extensive work with external databases that may provided pointers to non-registrants. Bill Clinton had the opportunity to end the draft, given one bill (military budget, I think) that got through Congress, but because of his Personal Background Problem he wasn't politically well-positioned to get rid of the draft that he'd had enough sense to dodge for himself. Phat chance that Bush will drop it.
Telecom and postal regulations allowing collection of user information without wiretap authorizations - The Feds and local police generally don't need specific wiretap authorization to collect telephone billing records or record who you receive Post Office mail from. There've been cases where they've subpoenaed phone bills from hundreds of thousands of non-involved people to find out if any of them might have called the person they _were_ authorized to surveil. And y'all know a lot about Carnivore and its friends The fun of being a leftover monopoly is that the government can do all sorts of things they couldn't get away with in an independent industry.
Add your own favorite example here!
There are lots of things the government can do to help privacy, but the first step has to be reducing the number of ways that the government is harming privacy. It's a slow process, and there are some regulatory steps they can take that may help while they're getting their act together on the real issues. I personally expect most of those regulations to cause some harm along with any good they cause, and the good parts of the regulations can be repealed while leaving the harmful parts as legal precedent, but hey, that's cynicism for you.
* Harassing the Europeans is a job for a separate posting. They've got similar problems with common identification numbers and the economics of computers, and while European Data Privacy Laws may be slightly larger bandages, they also provide government visibility into privately held databases (your pocket organizer or mobile phone's number list are databases, and they can go fishing in your machines for other data you might be suspected of having), they've decently demonstrated a continuing Willingness to Throw Them Over When Their Police Ask Nicely.
Horses do just fine, and the article commented on the authors expectation that they were related to the movements of the mummy's people. Euro-American invaders to Central and Western North America (:-) used horses and wagons to cross the place fairly rapidly, though later followers brought railroads. Herding cultures in the area drove cattle hundreds of miles, perhaps up to a thousand, just for market reasons, hauling them from grasslands to urban areas where they could be put on railroads. Railroads did simplify transportation of larger numbers of people, and simplified extermination of the buffalo that supported competing hunter-gatherer popoulations, which simplified expansion of the railroads and invasion of less-mobile crop-based populations, who resolved their differences with racially similar herder populations in much lower-intensity conflicts.
Peer-to-Peer may or may not have centralized indexing, and this can affect the ability to sell content. ICQ and Napster both used centralized systems to detect who was online, and used the peer-to-peer mechanism to deliver the bulk traffic. This can scale extremely well, if you implement it right - the servers handle the lightweight presence server tasks and optionally somewhat heavier indexing tasks, but the bulk distribution job not only doesn't use the servers, but the popular objects become available from large numbers of peers while the less popular objects don't go where they're not needed. (In the case of Napster, this scaling is theoretically 1:1, though it doesn't need to be, and in practice you'd prefer to use peers who have faster network connections more often than peers on analog dialup or slow radio links.)
To distribute Linux releases, you'd probably want to modify the mechanism somewhat - Nx600MB is a large chunk of data to need to transmit reliably, compared to the reliability of modems or servers. (Linux servers *are* much more reliable than Windows, but the reason you're distributing a new Linux release is so people can shut their systems down while they install new software on them.) So you'd need to distribute the thing as some collection of smaller packages, plus the software's critical enough that you'd need a centrally distributed digital signature mechanism (which is primarily just another file to ship along with the rest of the distro, except that everybody would probably want to verify the file centrally.)
I'm sorry, but this one trips by bogometer. Just about all the network connectivity into Alaska comes into Anchorage, where most of the people live - anything going up to the North Slope either gets there by way of one or two fiber routes, or by satellite, which are both expensive chunks of bandwidth. Yes, real estate's cheap, but realestate outside of Spokane Washington or Kansas City or for that matter Phoenix is also pretty cheap, and you can get multiple fiber routes connecting you to the Real World. Network Delay is another problem - round-trip time in fiber is 20ms per 1000 miles (or 1500km if you prefer), and the North Slope is Way Up There - companies like Akamai and AT&T and Digital Island get a lot of money for locating distributed web caching centers at network locations all around the US just to shave a few milliseconds of response time, and they're suggesting building far away? Trained workers are another problem - anybody in the Internet business knows that getting and keeping skilled operators who are trained on your particular system is a constant challenge, even in the current bust cycle; getting people who want a short gig up to the North Slope may be easy (I'd sure do it, if I could take a sabatical from my day job), but keeping people up there for a long time is something else entirely.
Is this really a media hack to tweak people about the Arctic National Wildlife Refuge and the Buch League's plans to drill it for oil and cut down all the trees to make government paperwork?
You'd have much better luck putting a hosting center on one of the First Nations reservations outside of Victoria BC. Some of them don't have treaties with the Canadian government, so there're interesting possibilities for using their sovereignty rights and tax status, and they're English-speaking and near the networks.
Nokia misfeatures curable by Open Source Music
on
Ring-Tone Royalties
·
· Score: 2
I never wanted my cell phone to play 20 bars for F"ur Elise or Flight of the Bumblebee or various race-horse music, though it would have been nice if the basic Big Ben clock-tower chimes were in the list of 40 or so tunes in my phone. I'd much rather just type in a few notes so I could tell my phone from the other nearby Nokia users' phones. 4-5 tones is generally enough - it really annoys me to listen to the shorter tunes go on for a dozen or so notes when they've already made their point. Yes, some people like that sort of stuff, and they did gain a lot of popularity by people playing with their phone music out in public where other people could notice it.
But the problem is that the phone doesn't have Open Source Music. Tunes are objects that you can download by SMS, if you happen to live in GSM-land where SMS works, but at least the earlier and dumber phone versions didn't let you enter your own. It would also have saved them a bunch of memory by providing half as many tunes and letting you enter your own.
And this whole intellectual property nonsense could be pretty much avoided. Beethoven's dead, and from a copyright standpoint he doesn't care whether you enter just 4 notes of his 5th Symphony (fair use, if trite) or the whole first movement. Walt Disney on the other hand, is merely spending the century dead for intellectual property law reasons so you may not be able to use much Mickey Mouse music. Brittany Spears is still alive, so if you want your phone to play Oops I called you again you might ask your lawyer whether the Digital Millenium Copyright Atrocity cancels out fair use for programming 6 notes into your phone but I wouldn't bother.
Does the RIAA/ASCAP/MPAA mafia extract money from the people who make those car horns that play 20 different annoying tunes?
Alcohol ~100K/year, not including car accidents, maybe another 10-30K
Prescription Drugs ~20K - includes suicides, accidental deaths from recreational use, deaths from non-recreational use.
Caffeine ~5-10K/year - contributes to heart attacks, strokes, blood pressure. My personal favorite.
Cocaine ~2500/year (~1990 figures) - a major contributor is quality differences in street drugs, which can be blamed on the Drug War, and also bad synthetic cocaine relatives accounted for 1/3 in ~1990, again blamed on the War.
Heroin ~2000/year -Again, quality differences due to the War On Drugs are a major contributor to death rates, but additionally there are serious nutritional problems related to spending your money being a junkie. Since medical opiates cost Meth, PCP, etc. - ~700 in ~1990 - meth is probably higher today - partly a popularity thing, and partly it's a substitute for cocaine, which is overpriced due to the war on drugs.
LSD - Zero. The major health risk to LSD is well-intentioned bad treatment by medical personnel, primarily with Thorazine or other dangerously strong tranquilizers. LSD can occasionally lead to Confused Behavior, which has its dangers.
Marijuana - Zero. The stuff isn't toxic. You can't kill yourself by consuming mass quantities - it's been tried. Marijuana can lead to Stupid Behavior, which *can* be fatal, and can lead to Pizza Overdoses and similar bad health effects - but the appetite increase can be a major health benefit for people with cancer, AIDS, and some other diseases. Some research suggests that smoking burning marijuana leaves can cause lung damage, particularly cancer, just as tobacco does, but high-quality modern marijuana (the stuff that D.A.R.E. tells your parents is ten times as strong as the stuff Cheech & Chong smoked which your parents could never score any of) uses much smaller quantities of burning leaves, and since marijuana doesn't have the addictive properties of nicotine, it's *really* hard to smoke two packs a day. Also, lung damage can be reduced by using bongs instead of those little easily-hidden pipes (but the War On Drugs bans this harm-reduction technique) or by consuming your dope in brownies or tea.
Summary - 99% from legal drugs or misuse (possibly legal) of prescription drugs.
Deaths from the War On Drugs:
Gunshots - Some high fraction of the 30000 US gunshot deaths are from the War On Drugs and Drug-War-funded gang activity.
Heterosexual AIDS - almost entirely transmitted through needle sharing or sex with partners who share needles. ~100-200K/year, entirely because the War On Drugs prevents most junkies from buying clean needles at the drugstore. (It's been shown that junkies with diabetes don't get AIDS - because they can use clean needles.) Some fraction of the homosexual AIDS epidemic is also drug-related, and enough heterosexuals have AIDS at this point that the drug can spread epidemically even without the needle-sharing that catalyzed it.
Bad/Variable Quality Street Drugs - as noted, this is a high fraction of the drug deaths.
Latin America murders and civil war deaths, from drug cartels, drug-funded rebel armies, and US-drug-war-funded government armies.
Lack of medical marijuana for cancer and other diseases where appetite is a problem.
Inadequate quantities of painkillers and substitution of inadequate or more dangerous painkillers for opiates, primarily morphine but sometimes also heroin. This is primarily a quality-of-life issue rather than a death-causing issue, but it's a serious problem for many elderly people, cancer patients, and people with other serious injuries or diseases. It's partially driven by medical concerns about avoiding addiction, but primarily by the political correctness enforced by government and medical standards bodies against the use of opiates.
Lives wasted in prison. Being in prison isn't as bad as being dead, but it's still slavery - and if you lose 10 years of your 70-80 behind bars, that's a major loss, affecting your family as well as you. Some US states do have the death penalty for selling large enough quantities of drugs (even marijuana), and some other countries like Singapore are much more enthusiastic about killing you for it. (The US Prison Growth Industry would rather have you as one of their customers than dead.)
Deaths from legalization
Then there's all the scary "but if drugs were legal, more people would use them and die" crap. If you look at government figures on death rates and numbers of users, it's 4-5 times safer to be a heroin user than a tobacco user, and about equally dangerous to be a heroin addict as a tobacco addict (the difference between the two figures is that most heroin and cocaine users are not addicts, just occasional users, while 95% of tobacco users are addicts.) Some people who aren't junkies or potheads would waste their lives that way if it were convenient and legal, just as some waste their lives as drunkards. On the other hand, most people would treat marijuana as an occasional drug to use, just as most people consume small quantities of alcohol but aren't frequent drunks. There would be some increased deaths from car accidents - the right way to deal with that is increased enforcement of driving-under-the-influence laws, and it's balanced by the reduction in shootings, drive-by and otherwise.
Dynamic web pages let you give each viewer different information. Caching lets you give different people the *same* information. Most ISPs and most corporate firewalls do a lot of caching, so if you can manage your application to maximize the amount of cacheable material, that helps. Fruhead.com has done a good job of minimizing unnecessary graphics and frills, and seems to do a lot of for discussion-board applications, which do need to be pretty dynamic, though there are sometimes ways to hack around that. So this advice isn't very helpful to you, but for lots of sites it makes a big difference.
As blair1q commented, I also did a doubletake on the name. The i860 was a really kinky chip that did some things very fast, though it appeared to be too weird for most compilers to do a good job of letting C language tell you which part of the processor to run your stuff on.
"Redundancy" has two substantially different problems: How do you initiate connections to outside internet sites, and how do they initiate connections to you. It's pretty easy to handle the outbound problem - most users have some kind of proxy firewall that handles their web and email traffic, and depending on what routing protocols your ISPs use, it's easy enough to find one route that works, especially if the main failure mode you're worried about is the access between you and your ISP's router. You don't need BGP for that, though it can be fun, you just need to know what locations you can reach by what paths, and nobody's bothered by the fact that sometimes your address space is from ISP1 and sometimes from ISP2.
The harder part is giving other people multiple paths to reach you. One way is to get yourself a routable address block (your local policies will indicate whether this is/19 or longer), and use BGP to advertise yourself to multiple ISPs, who forward those advertisements to the world. You need to be tolerably large to do this. Another way is to use a fancy DNS version that advertises different routes to you (www.you.co.nz gets advertised as a.a.a.x or b.b.b.x, using some load-balancing that also detects failure.) This isn't perfect, because DNS caches will prevent some outsiders from getting your current address quickly, but it's a good start. Another is to have a server in a hosting center that has multiple highly-reliable internet connections, so not only can you provide your web servers there where the response time and price of bandwidth are better than hosting them in your home office space, without risking backhoe fade, but you can use that to forward email and other services to your real IP addresses, whichever ones are working best this minute.
I can't speak for New Zealand - between physical isolation and occasional entertaining telecom and business regulation laws, there's lots of specialty detail involved. In particular, there may be fewer providers who can get you real paths off the islands, and you have to care a lot more about their service quality, but you still have a lot of flexibility for accessing local sites.
Slashdot Mirror
It's true that the FCC are a bunch of spectrum thieves, who nationalized spectrum in the Roosevelt days to protect the big-money communications companies from competition (even if they made lots of hype about protecting the public's interest in the public's airwaves), and US and European spectrum regulators figured out that the hype about Next Generation Wireless Services could be used to put a big hidden tax on wireless telephony and low-speed data services, which is showing up as huge debts by cellphone companies, just in case any of them weren't getting themselves into debt trouble investing in the fiber bandwidth glut
Fortunately, there's enough unlicensed spectrum to build some reasonable collections of services, but it'll take a lot of coordination. Metricom / Ricochet tried for a while; unfortunately they couldn't make enough money at it, but maybe another generation of providers will succeed, using faster commodity equipment. It's possible to do freenets if you can find a way to coordinate them (tough). But Starbucks is starting to offer commmercial wireless services for $X/month, and so are a few other companies, targeting either the coffeeshop market or the airport market, where there are enough business users with laptops and possibly wireless networks. Not much help if you live in the burbs, but here in San Francisco there's a law requiring a Starbucks on every other block, so if you live in the dot-com live-work loft district, that may be an attractive way to get service. For urban residential areas, where there's enough density for wireless nets to work, it's hard to say whether freenets or businesses will be more successful.
Software is also an issue - 802.11b tries very hard to look like an Ethernet card with a few extra hooks, but the 802.11a stuff wants more of the complex processing done by the CPU, in a WinModem-ish fashion, which says there'll be some major issues getting Linux drivers done for a while unless one of the two maufacturers decides it's worth helping out a lot
Basically, if the LA tax goons push too hard, they'll just lose, and make themselves look stupid as opposed to just greedy.
HTML and its relatives retain the structure of the document, they're editable and cut&paste-able, and even though MSWord generates ugly clueless HTML code, you can still reuse stuff.
The social engineering process you use is worthwhile, though.
Wait a minute here! Zinc Oxide is the stuff you use to keep damaging UltraViolet sunlight off your face. Now they're using it to turn it into UV Lasers? I can feel my nose burning already. Ouch!
Disc-lamer: I *am* making much of this up. I haven't tried to turn this into working code. Take it with several grains of salt. But the principles are pretty much correct.
I usually give crypto-cracking speeds (for adequately strong algorithms) in terms of planet-sized computers and billions of years, because that's obviously infeasible to crack, and if it's not, you should have made the keys a few bits longer. For RC4, that doesn't even cost you anything :-) Since you know how to calculate using exponentials, keep in mind that given good algorithms, it's trivial to make things that take that long to crack, and are so far out of reach of intelligence agencies that you should be worrying about other threats, like keyboard sniffers planted in your phone or passwords on yellow sticky notes. Single-DES can be brute-forced - John Gilmore proved that with the EFF Deep Crack machine, and the distributed crackers also showed they can do it. But Triple-DES isn't just 3 times as hard - it's 2**56 times as hard (total strength is only 112 bits, not 168, because there's a meet-in-the-middle attack that uses 2**56 pieces of memory, which is currently impractical.) RC4 is adjustable from near-0 to 255 bits of key length, with much less work per key brute-forced, but 128 bits is enough. The new US NIST Advanced Encryption Standard (contest won by Rijndael from Belgium) has modes for 128, 192, and 256 bits, if I remember correctly - even the weakest mode is strong enough for Earth-bound attackers.
The hard part of the crypto isn't the symmetric algorithm - it's the public-key part. I suppose they *could* have used 128-bit algorithms for that, but Elliptic Curve isn't strong enough at that length, and they'd be expected to know it. If you're not worried about traffic analysis, you could build a Kerberos-like system using 3DES or AES that fits in 128-bit keys.
The other obvious approach is to add a cellular modem to the cellphone, as long as it can get at least ~6.5kbps of throughput (one of the tighter compressions used in US digital cellphones) and set up a modem call. This needs a bit more hardware, but modems can be pretty compact, and again you've already done the compression in an ASIC. If you can't get fast enough modem speeds, you either need a tighter but nastier-sounding codec, e.g. 4800 baud or (gak!) 2400 baud or 1200 baud LPC (Speak-And-Spell is a trademark of somebody or other.) Or you can cheat and make a double-sized cellphone that's doing two simultaneous calls - klugey, but if you can afford DM6000 for a phone, you should be able to live with a much cheaper phone that burns minutes twice as fast.
Another approach is to wait for those 3G phones that the EU governments scammed their phone companies into paying giga-Euros of debt money for in the license auctions. Shouldn't cost any DM6K for one of those.
But the main reason you'd do crypto in hardware in a cellphone is that callphones tend to do the heavy lifting in ASICs and not have a lot of general-purpose computing horsepower or memory - it's easier to put the crypto into the ASIC than find somewhere else to wedge it.
Check out the Cypherpunks archives on the net.
GSM doesn't use ECC - it uses a couple of algorithms called A5, A8, etc. which look something like a fast fourier transform. Ian Goldberg, a Berkeley grad student, cracked them over lunch one day (he's not Israeli, just Canadian.) The authentication is a bit stronger than the message encryption. One of the entertaining results of the crack was the discovery that, while the keys are too short to start with, most of them have 10 bits set to 0, so they're even easier to crack, which is a strong argument that there was government pressure on the development process.
They're just pod bay doors, Hal, sure you can close them...
...
...
And wiretapping your cable company to steal internet access is likely to be difficult, depending on which protocols your cable provider is using - even if the transmissions aren't encrypted, the address assignment is usually tied into the billing mechanism somehow. You're better off paying your neighbor to put in an 802.11b wireless LAN instead.
I'll probably go back to a chiropractor, since some of that insight is how my spine and hips move together, and that's some thing chiropractors mess with a bit more directly that may help me benefit from some of the other techniques. Also, now that I'm getting my lazy butt into the gym more often, I'm finding that alignment issues affect my running, and the hip bone really *is* connected to the leg bone, etc.
Some of the newer foam technology is really nice, and helped a lot for some of my problems. I don't know if waterbeds are a good idea long-term - some people swear by them, some people view them as leftover hippie stuff (some of us view them nostalgicly as old hippie stuff :-) The newer versions give you ways to cut down on sloshing around and make the things easier to handle. I've never been that impressed with futons, but they work for some people. Some people do well with extra padding, though too much usually means you're not getting enough support, which is bad for your lower back, though it may be just great for your shoulders.
So experiment. Hack your bed.
Sounds suspicious to me.... Depending on whether the "centralized" box is really a centralized box run by his company or only a centralized-per-customer firewall-like-thing, it could be a golden opportunity for wiretapping the paranoid, or it could be just watered-down explanations given to the non-technical press by the Corporate Speaker-To-Publicists.
There are lots of things the government can do to help privacy, but the first step has to be reducing the number of ways that the government is harming privacy. It's a slow process, and there are some regulatory steps they can take that may help while they're getting their act together on the real issues. I personally expect most of those regulations to cause some harm along with any good they cause, and the good parts of the regulations can be repealed while leaving the harmful parts as legal precedent, but hey, that's cynicism for you.
* Harassing the Europeans is a job for a separate posting. They've got similar problems with common identification numbers and the economics of computers, and while European Data Privacy Laws may be slightly larger bandages, they also provide government visibility into privately held databases (your pocket organizer or mobile phone's number list are databases, and they can go fishing in your machines for other data you might be suspected of having), they've decently demonstrated a continuing
Willingness to Throw Them Over When Their Police Ask Nicely.
Horses do just fine, and the article commented on the authors expectation that they were related to the movements of the mummy's people. Euro-American invaders to Central and Western North America (:-) used horses and wagons to cross the place fairly rapidly, though later followers brought railroads. Herding cultures in the area drove cattle hundreds of miles, perhaps up to a thousand, just for market reasons, hauling them from grasslands to urban areas where they could be put on railroads. Railroads did simplify transportation of larger numbers of people, and simplified extermination of the buffalo that supported competing hunter-gatherer popoulations, which simplified expansion of the railroads and invasion of less-mobile crop-based populations, who resolved their differences with racially similar herder populations in much lower-intensity conflicts.
To distribute Linux releases, you'd probably want to modify the mechanism somewhat - Nx600MB is a large chunk of data to need to transmit reliably, compared to the reliability of modems or servers. (Linux servers *are* much more reliable than Windows, but the reason you're distributing a new Linux release is so people can shut their systems down while they install new software on them.) So you'd need to distribute the thing as some collection of smaller packages, plus the software's critical enough that you'd need a centrally distributed digital signature mechanism (which is primarily just another file to ship along with the rest of the distro, except that everybody would probably want to verify the file centrally.)
Is this really a media hack to tweak people about the Arctic National Wildlife Refuge and the Buch League's plans to drill it for oil and cut down all the trees to make government paperwork?
You'd have much better luck putting a hosting center on one of the First Nations reservations outside of Victoria BC. Some of them don't have treaties with the Canadian government, so there're interesting possibilities for using their sovereignty rights and tax status, and they're English-speaking and near the networks.
For that matter, you'd have much better luck putting a hosting center on some or a slightly-used nuclear missile bunker in the UK.
But the problem is that the phone doesn't have Open Source Music. Tunes are objects that you can download by SMS, if you happen to live in GSM-land where SMS works, but at least the earlier and dumber phone versions didn't let you enter your own. It would also have saved them a bunch of memory by providing half as many tunes and letting you enter your own.
And this whole intellectual property nonsense could be pretty much avoided. Beethoven's dead, and from a copyright standpoint he doesn't care whether you enter just 4 notes of his 5th Symphony (fair use, if trite) or the whole first movement. Walt Disney on the other hand, is merely spending the century dead for intellectual property law reasons so you may not be able to use much Mickey Mouse music. Brittany Spears is still alive, so if you want your phone to play Oops I called you again you might ask your lawyer whether the Digital Millenium Copyright Atrocity cancels out fair use for programming 6 notes into your phone but I wouldn't bother.
Does the RIAA/ASCAP/MPAA mafia extract money from the people who make those car horns that play 20 different annoying tunes?
Deaths from the War On Drugs:
- Gunshots - Some high fraction of the 30000 US gunshot deaths are from the War On Drugs and Drug-War-funded gang activity.
- Heterosexual AIDS - almost entirely transmitted through needle sharing or sex with partners who share needles. ~100-200K/year, entirely because the War On Drugs prevents most junkies from buying clean needles at the drugstore. (It's been shown that junkies with diabetes don't get AIDS - because they can use clean needles.) Some fraction of the homosexual AIDS epidemic is also drug-related, and enough heterosexuals have AIDS at this point that the drug can spread epidemically even without the needle-sharing that catalyzed it.
- Bad/Variable Quality Street Drugs - as noted, this is a high fraction of the drug deaths.
- Latin America murders and civil war deaths, from drug cartels, drug-funded rebel armies, and US-drug-war-funded government armies.
- Lack of medical marijuana for cancer and other diseases where appetite is a problem.
- Inadequate quantities of painkillers and substitution of inadequate or more dangerous painkillers for opiates, primarily morphine but sometimes also heroin. This is primarily a quality-of-life issue rather than a death-causing issue, but it's a serious problem for many elderly people, cancer patients, and people with other serious injuries or diseases. It's partially driven by medical concerns about avoiding addiction, but primarily by the political correctness enforced by government and medical standards bodies against the use of opiates.
- Lives wasted in prison. Being in prison isn't as bad as being dead, but it's still slavery - and if you lose 10 years of your 70-80 behind bars, that's a major loss, affecting your family as well as you. Some US states do have the death penalty for selling large enough quantities of drugs (even marijuana), and some other countries like Singapore are much more enthusiastic about killing you for it. (The US Prison Growth Industry would rather have you as one of their customers than dead.)
Deaths from legalizationThen there's all the scary "but if drugs were legal, more people would use them and die" crap. If you look at government figures on death rates and numbers of users, it's 4-5 times safer to be a heroin user than a tobacco user, and about equally dangerous to be a heroin addict as a tobacco addict (the difference between the two figures is that most heroin and cocaine users are not addicts, just occasional users, while 95% of tobacco users are addicts.) Some people who aren't junkies or potheads would waste their lives that way if it were convenient and legal, just as some waste their lives as drunkards. On the other hand, most people would treat marijuana as an occasional drug to use, just as most people consume small quantities of alcohol but aren't frequent drunks. There would be some increased deaths from car accidents - the right way to deal with that is increased enforcement of driving-under-the-influence laws, and it's balanced by the reduction in shootings, drive-by and otherwise.
Dynamic web pages let you give each viewer different information. Caching lets you give different people the *same* information. Most ISPs and most corporate firewalls do a lot of caching, so if you can manage your application to maximize the amount of cacheable material, that helps. Fruhead.com has done a good job of minimizing unnecessary graphics and frills, and seems to do a lot of for discussion-board applications, which do need to be pretty dynamic, though there are sometimes ways to hack around that. So this advice isn't very helpful to you, but for lots of sites it makes a big difference.
As blair1q commented, I also did a doubletake on the name. The i860 was a really kinky chip that did some things very fast, though it appeared to be too weird for most compilers to do a good job of letting C language tell you which part of the processor to run your stuff on.
The harder part is giving other people multiple paths to reach you. One way is to get yourself a routable address block (your local policies will indicate whether this is
I can't speak for New Zealand - between physical isolation and occasional entertaining telecom and business regulation laws, there's lots of specialty detail involved. In particular, there may be fewer providers who can get you real paths off the islands, and you have to care a lot more about their service quality, but you still have a lot of flexibility for accessing local sites.