Spammers' favorite big registrar! Spammer location obfuscation! Boo! Evil Protectors of Miscreants!
It's the same thing, of course - some of the biggest users of privacy are annoying resource thieves. Oh, well.
Also, GoDaddy is a US company, so anybody in the US who wants to drop legal papers on them can pretty much do so, so your activities may not be outlawed yet, but that could change, and you've got no protection for any useful information you've provided them if you do. For a while, Gandi.net was the politically correct choice for registration - they weren't Verisign, did good customer service, weren't Verisign, were based outside the US so you have to trigger two governments' stupidities before you get hassled, weren't Verisign, charged a reasonable fee (12 euros was pretty low back then, though others are lower), and they weren't Verisign. OpenSRS seems to be the current politically correct approach more recently.
Seems like he might be getting back on the Peace Train, even though he got kicked off the plane. His websites are getting a bit Slashdotted since the press announcements, but GoogleCache He says he never endorsed the Fatwa, and that he was misquoted - he was explaining that Islamic law is quite clear about killing blasphemers, but also about going through local legal process, and since that's not the law in Britain, Muslims shouldn't go lynching people, they should just get involved in politics. I consider that a pretty wimpy defense, but he apparently wasn't in favor of the Fatwa itself and was really upset when the press reported that he was.
He has made a bunch of statements opposing terrorism over the years, including the recent school invasion in Russia and the 9/11 attacks and the US invasions of Iraq.
If you're engaged in an activity which the State has gotten away with demanding that you have a license for, e.g. driving, police may be able to demand that you show them license paperwork. Otherwise, you don't have to tell them anything. They can use information you give them voluntarily, and the whole Miranda process is designed to make sure that police cannot force you to disclose information when you're in a non-voluntary situation, or at least can't use that information against you in court. As an American resident, you're presumed to know this, so anything the police tell you in a non-custody situation is presumed to be voluntary. Cases like Brown vs. Texas hold that you don't even have to identify yourself to police if you're arrested - they can book you as "John Doe" if you don't give them a better name.
The one time I've had my Miranda rights read to me, the cops violated them and the Federal Privacy Act after we got back to the cop shop, or as the sergeant said "This isn't a threat, it's just a choice you can make". (I'd been photographing misbehaving small-town cops, and they didn't appreciate it, but this was pre- Rodney King, so it was easier for them to get away with things, and I was supposed to go on vacation the next day and didn't want to spend the weekend in jail instead of getting on my plane, which the cops were quite correct that they could do.) Charges were later dropped, but it was annoying.
Until the recent Hiibel case , the courts were really clear about this; it's an ugly mess, and the Supremes upheld an Nevada law permitting cops to ask people to identify themselves, in spite of the fact that that's not what the cop did in the events under consideration.
It's better to be able to simply not provide a phone number, but that's available if the registrar has an automated form that insists on something. (Kind of annoying if your main phone isn't part of the North American Numbering Plan and the web page is over-enthusiastic about validation.)
I'm glad it's gotten watered down - the people who asked the legislators to push this bill have been twisting what Whois records are, and many members of the public, apparently including you, haven't understood them well enough not to be misled.
The WHOIS records aren't a legal declaration of your True Name, True Legal Domicile, Phone number you agree to be reached at 24 hours a day by anyone who wants, ICBM address, Subpoena Acceptance Address, Mother's Maiden Name, Fingerprints, and RIAA pre-approved guilty plea that you give The Authorities in exchange for permission to speak on the Internet. They're simply administrative contact information people can use to try to reach you if your system is having trouble. There's a billing address there so that the Registrars can reach you if they want more money. There's a technical address to reach you if things are broken. There's an administrative address for general administrative requests. If somebody can't reach you because your information is out of date or incorrect, that doesn't mean you're an evil miscreant, it just means that you won't get proactive billing notices, and if your DNS isn't working right, people can't reach you to let you know.
It is possible to give the registrars fraudulent information - if you're impersonating someone else who really exists, but that's adequately covered by existing fraud laws. But if you give your name as "Johnny Smith" and put your address as "111 Main Street, Bogustown, USA" or "1600 Pennsylvania Ave, Washington DC 90210", that's nobody's business, that's just not a very useful contact handle you're giving somebody. If your payment to the registrar works, it works. And here in California, it was common-law right to use any name you wanted to except for purposes of fraud, though apparently the DMV got that changed a decade or so ago and insists that you need papers from some government or other to have a name.
ICANN seems to have been one of the early prime movers in True Name Whois Information, in spite of the damages to privacy that it causes (e.g. spammers hitting your published admin address.) The "IP" that they're interested in has always been "Intellectual Property", not "Internet Protocol", and they're really grouchy about the concept that anybody could ever use a domain name without agreeing to provide an always-updated True Name and Legal Process Server Address so that trademark owners can find you and sue you if they think they've got a claim on a domain name you're using. More recently, though, the RIAA/MPAA have taken up the cause, because they want to be sure that if you ever even think about sharing copyrighted music on line, they want to be able to drop handcuffs on you. If there's a dispute about domain name ownership, and your Registrar is unsuccessful in contacting you using the contact information you provide, for some reasonable period of time, it's reasonable for them to bounce your domain name.
On the other side of the argument, while I strongly value privacy, most of the time when I try to track down spammers using whois records, the information is bogus, which is annoying, and it's almost always either obviously bogus or else some foreign address that looks hard to track down. The main exceptions are in-your-face spammers like Spamford or Scotty Richter, and spammers with corporate shells to hide behind (e.g. one spammer had a mailbox at the street address of The Company Corporation, which is in the business of setting up cheap Delaware corporations), so they're effectively untraceable.
Meanwhile, if you're a "hardcore libertarian", you need to think about what rights mean. Saying somebody doesn't have the right to do something isn't just a statement about ethics - its equivalent to saying that you have the right to beat them up if they do it. Falsifying your personal resume is attempting to deceive somebody about your skills so they'll give you something that they wouldn't if you'd been honest, and of course that's wrong.
As they say, if car technology evolved the way computer technology did, compared to the cars of the 1980s, you should expect that today you can get a Rolls Royce that goes 800 mph, gets 400 miles per gallon, holds 82 passengers, and explodes twice a week killing everybody inside.
Individual pieces of equipment seldom get above 4-5 nines of reliability (4 nines is about 1 hour of downtime a year; 5 is about 6 minutes.) That's fine - so you use duplicate equipment, with a third piece of equipment watching the two working pieces to be sure they're both running correctly, or N+2 pieces of critical equipment if failures are obvious to the operators, and you make it hot swappable, so that if one piece fails, you can replace it while the other one's running, and you do a lot of work to prevent common-mode failures and undetected failures.
Of course, the most important thing is to spend a lot of time carefully defining what events are or are not failures, because that can make a couple of nines difference in what you call the reliability numbers...
For instance, what about preventive maintenance? If Thing1 and Thing2 each have 99.99% reliability, and you take Thing1 down for an hour for maintenance, have you blown your 8 9's reliability for the year because there was a 0.0001 chance of Thing2 failing during that hour? If so, then you need triplicate equipment, not just duplicate. And remember that the 1970s backup equipment is running the whole system for 4 hours a night to keep it working and keep the operators trained - can you fire it up during your 1-hour preventive maintenance and not get dinged for failure risk?
What about partial failures? If you've got a box that's supposed to manage 100 radar lines, obviously an event that takes down all 100 is a "failure", so you have to duplicate enough to cut the probability of that event down below your spec. But what if just one radar line fails - do you need to make the line cards supporting _each_ line 10 nines reliable, so that the chance that not all lines are working is 8 nines, or is it good enough to make sure that each radar line is independently 8 nines reliable (including per-line and per-half-system and per-system failures)? Hint: If your management is too conservative about the decision they make, you need triplicate line card support, which is much much harder than duplicate. And this gets to be _really_ annoying if each radar line your processor supports is on a telco circuit that's only 3-4 nines reliable in itself, but you're running it on a 10 nines triplicate set of line cards....
And yes, the FAA has always been on drugs. One of the drugs they're on is knowing that if there's an airplane crash and hundreds of dead bodies due to problems with air traffic control, they get infinite amounts of political heat, whereas if major hub airports don't have enough capacity because the ATC system is antiquated, well, that's only money, and usually somebody else's money at that, and if there are appalling delays and cost overruns, maybe it takes a bit longer to get promoted, but often you can _get_ more budget, because if two 747s full of school children crash over LAX a month before Election Day due to ATC glitches, nobody wants to be the Congresscritter who voted against fixing the ATC system. So the system's rigged against them, forcing them to be overconservative, and to _look_ extremely conservative, except that every once in a while the fragility and brokenness of the system catches up with them and forces them to do something in a hurry, especially if there's going to be an election where the top people get replaced for partisan political reasons, which gives them an opportunity to let the outgoing guy take any blame after he's gone. So just because they're on drugs doesn't mean that it doesn't suck to be them...
On the other hand, you really can get equipment that reliable if you're willing to pay for it, and component reliability has improved wonderfully since the 1980s, e.g. disk drive MTBFs of 500000-1M hours instead of 10,000 hours, so you really can wait until midnight slowdown to rebuild the RAID partition after you hot-swap the drive, and computers are a couple orders of magnitude faster so you need fewer of them to get a given job done fast enough, making it much easier to make subsystems reliable and monitor their status.
Memory can leak because of applications. Memory can leak because of operating systems. Memory can leak because of obscure timing bugs nobody can find between the OS and the application's garbage collector. Memory can leak because the hardware clock that drives the timer for the garbage collector sometimes skips a beat because of bus loading. Memory can leak because the moon is full, affecting the frequency of cosmic rays hitting the memory chips. Memory can even leak because somebody didn't RTFM.
Fragmentation is another problem besides leaking, but it can also lead to systems getting progressively slower until they drop below some critical performance threshold.
And disk drives _do_ fill up with log files unless you do something about it.
Back when my department used Vaxen, we'd reboot them every Friday night, fsck the disks, and do backups. Around the time we were running SVR2, the file system really was stable enough and the removable disk packs high enough quality that fsck seldom found anything and didn't need manual intervention, and the rebooting process was reliable enough that we could let a cron job run it, and while we could have cut back to monthly, people had gotten in the habit of knowing the machine would be down, so they could get a life, and it was a good schedule for the times we actually did want to to upgrades or hardware maintenance.
Backup plans, and people who know how to use them, are really much more critical, because sometimes you will need them.
Failover clusters aren't trivial - I worked on a non-winning design for one of the predecessors to this system back in the late 80s (fortunately for us, we lost, and unfortunately for IBM, they won.) Yes, you can have two, three, or N of everything, but then you need a lot of code watching the redundant components to see if any of them appear to be failing, and code deciding which redundant subsystem is correct if two or more of them disagree, and code watching the watchers to make sure they're still watching well, and data communication protocols that work ok when all messages are transmitted redundantly to the redundant processors, possibly getting different results at microsecondly different times. One of my coworkers had worked with an early "fault-tolerant computer" system which had triply or quadruply redundant hardware, but had an operating system that crashed at least weekly because it was too complex.
You also have to be extremely careful and flexible in your design for the granularity of the redundant subsystems - if you make the separately processed chunks too big or too small, you can have an order of magnitude change in performance and sometimes several orders of magnitude change in reliability, and then there's the problem that the definition of "reliability" includes "probability that the calculation finishes in N milliseconds", so it's inextricably linked with performance.
Moore's Law is really your friend here. Improved performance means you can use a lot fewer parts, which reduces complexity and failures. Disk drives are more than an order of magnitude more reliable, and the increase in size means that a cluster of disks containing N gigabytes is several orders of magnitude more reliable because it's a lot fewer disks, and CPUs that are 2+ orders of magnitude faster mean that it's easier to guarantee that something happens in a given time, and cuts down on communications steps between different modules, so you cut down on all the failure modes for those communications, and on the monitoring software watching for failures, and on the failures of the monitoring software. On the other hand, Moore's Law lets operating system vendors and application vendors bloat their software with features - X Windows 11R3 ran just fine on my 386/25 machine with 8MB RAM, but of course I was using twm, not Enlightenment or Gnome.
Backup plans do introduce the danger of complexity - the FAA doctrines of the 1980s were that any new system had to be able to interoperate with everything its predecessor interoperated with, because you weren't going to flash-cut upgrade everything at once. That meant that everything you designed had to be bug-for-bug backwards compatible with the predecessor's interfaces, and when they redesigned the thing _your_ system interoperates with, it has to be bug-for-bug compatible with everything your system does, which means being compatible with its predecessor, which was compatible with your system's predecessor, etc. It's a vicious circle similar to the messes Windows and Intel CPUs had to put up with, except that while the 8088 and MS-DOS 1.0 were *ugly*, they were at least small and well-documented late-70s technology, as opposed to poorly-documented 1960s JOVIAL and 1950s analog.
What was the old line about laws and sausages, that if you're fool enough to like either one of them you ought to be forced to see them being made? Yes, the old system was "tested", probably, but the loads it was tested for don't resemble the modern airspace environment.
The 40-year-old system was pretty much the Mos Eisley of software design - you'll never see a more wretched hive of scum, villainy, and undocumented unmaintainable Jovial code running on IBM 360/50 and 360/90 hardware. The backup system was much cleaner (and much dumber); I think the main thing they did in the 1970s enhancement was retread the design to use transistors instead of vacuum tubes, though I never worked directly with that side.
Yes, Sun and IBM machines fail - that's why all of the critical parts in our designs had to be at least doubly redundant, and often triply redundant, because the design spec of "Eight 9s of reliability" meant that doing an hour a year of preventive maintenance might expose you to too much risk from the backup system failing. I haven't seen IBM's design; I was on the lucky team that didn't win the bidding to build the final system, unlike the poor suckers at IBM who had to implement theirs, but the requirements were not only insanely non-implementable, they were excessively focussed on No Possible Downtime Ever, because if anything goes wrong resulting in an airline crash, the FAA gets insane amounts of political heat. Doesn't matter if the system is N years late, because you can try to blame the contractor for that, or if you can't fly supersonic planes across the Continental US because they're too fast for the new ARTCCs, because tough luck for the French and for bi-coastal business travellers.
Of course, that doesn't mean that Im inclined to trust a system running on Windows, either...
Back when I was working on ARTCC replacement in the late 80s, during the daytime they were running the "modern" 1960s IBM System 360/90 system, which was an ugly undocumented unmaintainable hack job written mostly in JOVIAL. For about four hours a night, they'd run the backup system EDARC, which was an 1970s "Enhanced" version of the 1950s "DARC" radar controller. There were all sorts of parts you couldn't get back in the 1980s - IBM had stopped making the "Serpentine" cable connector, for instance.
I was on the lucky team that *lost* the bidding for the replacement system; IBM's team were the poor bastards who won, and were stuck investing seven years into building an unbuildable replacement, pouring billions of dollars down the drain while being micromanaged by the FAA, who didn't know much about software design or reliability in spite of having a methodology that required producing 175 design documents over the optimistically 3-year design period.
The wireless telcos are totally clueless about the potential size of the market for these devices, because they keep assuming they can charge rip-off prices per bit/packet/message just like they do with text messaging services, and they're unwilling to risk reducing their revenue from texting by changing the economics. The right price is something competitive with DSL / Cable Modems, e.g. $30-60 flat rate per month for all the bits you can eat, trading off convenient access for a bit more money.
Supposing they could make a profit at that price, they'd be able to sell to a lot of people. The price of the raw internet bandwidth feeding their towers has come way down; the real question is how many users their frequency bands can handle if they're actually using the system.
Back when Metricom was around, and DSL mostly wasn't, most of my geeky friends had it, for prices like $40/month, compared to $20 for dialup, and at least one non-geeky coworker had it supporting the four users in his household who'd otherwise all have phone lines and modems. (Of course Metricom _died_ at those prices, but they were running really customer equipment back when raw Internet upstream was expensive, and they never got the customer volume to get economies of scale. By contrast, wireless phone companies already have tons of volume - they're just short on clues.)
My Hitachi Visionbook Traveller weighs 2.7 pounds (about 1.2kg) and has a ~10" screen, and it was what passed for a cheap subnotebook back in its day, and it has three PCMCIA card slots, two of which are aligned so you can use a fat-form-factor PCMCIA disk if you want.
Alas, like many lightweight machines, part of the lightness was because of flimsy construction, so it's gradually fallen apart, and you can no longer buy critical parts like the little plastic doors that hold in the batteries. (Too bad - it's sensibly designed to use standard video-camera batteries, so they'd be replaceable at a quasi-rational price if the silly doors weren't missing.) That all became less relevant when the screen cracked, so the machine now looks like a nice little keyboard that has VGA and Ethernet interfaces instead of PS2.
Most hax0ring doesn't require you to be 31337 any more. Once in a while somebody does an interesting hack, but most of it these days is just assembly-line work, run automated scriptz you got from the otha k1dd13z, and if you've seen one you've seen them all.
It's kind of like hacking cars - taking off the muffler might have been fun for 15 minutes when you were 16, but everybody's heard it before and it just sounds like you did it because your muffler had rusted out anyway, so no sense annoying the neighborhood, and that 42-tune electronic horn widget had gotten old by 1980 and makes you look dorky, not retro. There's the occasional quasi-new hack, like the Bubb Rubb Car Whistle, but even that just invites people to break your car windows or spackle your muffler shut.
So all that 1337 h4x0rin's become just another day job, like stealing hubcaps for profit, or graffitiing telephone poles with signs about "Make Money Fast - call 1-800-SCAM-MER"; it's mostly taken over by automated systems or underpaid losers selling to spammers. If you want to have _fun_ hacking cars, you can just as well do legitimate things like make art cars or improve your gas mileage, and if you want to do 31337 on your computer, might as well do something interesting like write new software, or hunt down spammers, or at least find ways to hack MMORPGs so you can frag your friends.
But you _can_ play a game called Linux, and it lets you play games like Adventure and Nethack and TuxRacer (actually, I don't know if TuxRacer runs on the PS2, not having one to try it with), and play as many dimensions of MUD as you want to...
Can you tell if it's USB1/1.1 or USB2? That'd let you use a USB drive, though it's obviously much more interesting if it's USB2, and (unlikely) extra credit if it's bootable from USB (but still good enough for/home even if you boot from CD/DVD.)
Also, extra credit if you can run this as a Beowulf cluster, though of course the really interesting Playstation cluster system uses the Emotion Engine graphics chip to do number crunching rather than the relatively tame CPU so it's not a straight Beowulf system. This new Playstation version is 1/3 the size and 1/2 the price of the ones NCSA used, so it's an even more attractive idea...
So run Knoppix from the CDROM or DVD with your/home directory mounted across the network. (Actually, if the new one only has 32MB of RAM, you probably want some smaller CDROM-bootable competitor.) You've presumably got _some_ kind of Unix machine available as a server, or else you can run a Windows file sharing protocol and SAMBA.
If you're a kid, and your parents buy you the gamer box for Christmas, they're probably not buying every game out there along with it; they're buying you the box and a couple of popular games. You can use that money from Granny to buy a couple more games once you're bored, e.g. the Decmeber 27th sales.
Sure, as an individual you want to be able to have a phone that works in emergencies - but most of the discussion about this topic is run by policy wonks from one group or another. I suppose I did leave out the FCC types who are annoyed that VOIP doesn't look like their current regulatory structure, and it's hard to justify radically increasing the cost of an essentially free service with the excuse that you're doing it to make service affordable.
Just used up my mod points on another discussion, so I can't mod you up, but you're quite right about "good riddance". Of course, if you were using Mozilla, you wouldn't see popups anyway.
The reason bots are potentially bad for business is that they could drive away other players, though of course the house doesn't care if a table full of bots are playing each other as long as there are enough of them.
First of all, your cell phone will work just fine even if your power is out. If the VOIP companies were clueful, they'd have somebody build them a VOIP router with a cell-phone circuit built in so you can make emergency calls.
Of course VOIP and 911 don't get along - 911 was designed to work in a landline environment, with communications architectures tightly tied to Class 5 telco switches and database architectures designed for phones that stay in one place, and the 911 folks haven't been willing to adapt their systems to accept VOIP connections even though it wouldn't be that hard. VOIP, like wireless, presents some new technical challenges because the equipment is portable, and if you bring your VOIP box on a business trip with you and have to call the fire department, you want firetrucks showing up where you are, not back at your house. But there are ways to design around it, whether you do something with GPS or adapt your DHCP servers to pass you geographical info or whether you have the VOIP box/software/etc. let you tell it your address.
Complaints about VOIP and 911 are usually a cover for real complaints about VOIP and wiretapping. The folks who like wiretapping are annoyed that changing technology makes their tools obsolete, and want to force the technology to adapt to them, rather than the other way around, and they tend to use 911 as a lever to do that. After all, you want an ambulance to be able to find you if you're hurt, but you probably don't want the police to be able to locate you within 10 meters and follow you all day, so that's not the motivation they advertise for mandating that new cell systems provide user location. Similarly, the wiretappers _really_ don't like peer-to-peer flexible technology, and they're used to having hooks into traditional telcos to control them.
Re:This isn't Everest, and astronomers are indoors
on
Antarctic Telescope?
·
· Score: 1
My family gets together in the Colorado mountains in the summer, at about 8300 feet. It does take me a day or so to acclimatize, if I haven't already been in Denver on business. My grandmother had to stop going in her early 90s because the air was a bit too thin, but she was in pretty good shape for her age. (Ok, she mostly sat around in her rocking chair, but by those years that was what she usually did back in Kansas City as well, and she liked the mountains better.) After a day at 8300, going up to 11000 feet is still hard to do much physical exertion without dizziness, but it's not bad. 14000 was a lot more trouble, but back when I was hiking at that altitude I was one of those "scientists not well known for their physical abilities"; now I'm in my late 40s instead of late 30s, but I was already fat by then:-) On the other hand, I've known a lot of scientists who are in much better physical condition, especially younger ones who haven't gotten fat. Some of the people I met while hiking out there lived at 10000 feet, so for them it was a piece of cake, and the ultramarathoners who I met halfway up the mountain (me going up, them coming down) had been around there training for a week. And the cold is partly from dryness - tweaking humidity helps a lot.
This isn't Everest, and astronomers are indoors
on
Antarctic Telescope?
·
· Score: 1
Sure, above 7000 meters, it's nice to have oxygen tanks, and above 8000, it's really nasty not to have it, though a few people don't use it. But astronomers aren't likely to be that high up, and even if they are, they do most of their work indoors, and the buildings can have pressurized air or oxygen concentrators (plus the buildings can have heat, which is a real problem for mountain climbers.) Meanwhile, the telescope gets to look through all the lack of atmosphere above it.
So build another in the Arctic
on
Antarctic Telescope?
·
· Score: 2, Interesting
This one's in Antarctica because Australians have the silly idea that it's closer to them. Building in the Arctic gets you the other half of the year.
And the cost of building two of these things is much less than twice the cost of building a single one, because a large fraction of the cost is developing all the tools and technology to build it, and they can crank out two or three more for not much extra cost. (Obviously building the base and staffing it are duplicated costs.) By contrast, building all the launching systems for the Hubble is so expensive that you're not going to build a couple of clones and launch them, you're going to wait another decade and develop most of the system from scratch using the technology of the time again.
Slashdot Mirror
Spammers' favorite big registrar! Spammer location obfuscation! Boo! Evil Protectors of Miscreants!
It's the same thing, of course - some of the biggest users of privacy are annoying resource thieves. Oh, well.
Also, GoDaddy is a US company, so anybody in the US who wants to drop legal papers on them can pretty much do so, so your activities may not be outlawed yet, but that could change, and you've got no protection for any useful information you've provided them if you do. For a while, Gandi.net was the politically correct choice for registration - they weren't Verisign, did good customer service, weren't Verisign, were based outside the US so you have to trigger two governments' stupidities before you get hassled, weren't Verisign, charged a reasonable fee (12 euros was pretty low back then, though others are lower), and they weren't Verisign. OpenSRS seems to be the current politically correct approach more recently.
He has made a bunch of statements opposing terrorism over the years, including the recent school invasion in Russia and the 9/11 attacks and the US invasions of Iraq.
The one time I've had my Miranda rights read to me, the cops violated them and the Federal Privacy Act after we got back to the cop shop, or as the sergeant said "This isn't a threat, it's just a choice you can make". (I'd been photographing misbehaving small-town cops, and they didn't appreciate it, but this was pre- Rodney King, so it was easier for them to get away with things, and I was supposed to go on vacation the next day and didn't want to spend the weekend in jail instead of getting on my plane, which the cops were quite correct that they could do.) Charges were later dropped, but it was annoying.
Until the recent Hiibel case , the courts were really clear about this; it's an ugly mess, and the Supremes upheld an Nevada law permitting cops to ask people to identify themselves, in spite of the fact that that's not what the cop did in the events under consideration.
It's better to be able to simply not provide a phone number, but that's available if the registrar has an automated form that insists on something. (Kind of annoying if your main phone isn't part of the North American Numbering Plan and the web page is over-enthusiastic about validation.)
The WHOIS records aren't a legal declaration of your True Name, True Legal Domicile, Phone number you agree to be reached at 24 hours a day by anyone who wants, ICBM address, Subpoena Acceptance Address, Mother's Maiden Name, Fingerprints, and RIAA pre-approved guilty plea that you give The Authorities in exchange for permission to speak on the Internet. They're simply administrative contact information people can use to try to reach you if your system is having trouble. There's a billing address there so that the Registrars can reach you if they want more money. There's a technical address to reach you if things are broken. There's an administrative address for general administrative requests. If somebody can't reach you because your information is out of date or incorrect, that doesn't mean you're an evil miscreant, it just means that you won't get proactive billing notices, and if your DNS isn't working right, people can't reach you to let you know.
It is possible to give the registrars fraudulent information - if you're impersonating someone else who really exists, but that's adequately covered by existing fraud laws. But if you give your name as "Johnny Smith" and put your address as "111 Main Street, Bogustown, USA" or "1600 Pennsylvania Ave, Washington DC 90210", that's nobody's business, that's just not a very useful contact handle you're giving somebody. If your payment to the registrar works, it works. And here in California, it was common-law right to use any name you wanted to except for purposes of fraud, though apparently the DMV got that changed a decade or so ago and insists that you need papers from some government or other to have a name.
ICANN seems to have been one of the early prime movers in True Name Whois Information, in spite of the damages to privacy that it causes (e.g. spammers hitting your published admin address.) The "IP" that they're interested in has always been "Intellectual Property", not "Internet Protocol", and they're really grouchy about the concept that anybody could ever use a domain name without agreeing to provide an always-updated True Name and Legal Process Server Address so that trademark owners can find you and sue you if they think they've got a claim on a domain name you're using. More recently, though, the RIAA/MPAA have taken up the cause, because they want to be sure that if you ever even think about sharing copyrighted music on line, they want to be able to drop handcuffs on you. If there's a dispute about domain name ownership, and your Registrar is unsuccessful in contacting you using the contact information you provide, for some reasonable period of time, it's reasonable for them to bounce your domain name.
On the other side of the argument, while I strongly value privacy, most of the time when I try to track down spammers using whois records, the information is bogus, which is annoying, and it's almost always either obviously bogus or else some foreign address that looks hard to track down. The main exceptions are in-your-face spammers like Spamford or Scotty Richter, and spammers with corporate shells to hide behind (e.g. one spammer had a mailbox at the street address of The Company Corporation, which is in the business of setting up cheap Delaware corporations), so they're effectively untraceable.
Meanwhile, if you're a "hardcore libertarian", you need to think about what rights mean. Saying somebody doesn't have the right to do something isn't just a statement about ethics - its equivalent to saying that you have the right to beat them up if they do it. Falsifying your personal resume is attempting to deceive somebody about your skills so they'll give you something that they wouldn't if you'd been honest, and of course that's wrong.
But "Trav
So you thought you like Fly-By-Wire airplanes?
Of course, the most important thing is to spend a lot of time carefully defining what events are or are not failures, because that can make a couple of nines difference in what you call the reliability numbers...
And yes, the FAA has always been on drugs. One of the drugs they're on is knowing that if there's an airplane crash and hundreds of dead bodies due to problems with air traffic control, they get infinite amounts of political heat, whereas if major hub airports don't have enough capacity because the ATC system is antiquated, well, that's only money, and usually somebody else's money at that, and if there are appalling delays and cost overruns, maybe it takes a bit longer to get promoted, but often you can _get_ more budget, because if two 747s full of school children crash over LAX a month before Election Day due to ATC glitches, nobody wants to be the Congresscritter who voted against fixing the ATC system. So the system's rigged against them, forcing them to be overconservative, and to _look_ extremely conservative, except that every once in a while the fragility and brokenness of the system catches up with them and forces them to do something in a hurry, especially if there's going to be an election where the top people get replaced for partisan political reasons, which gives them an opportunity to let the outgoing guy take any blame after he's gone. So just because they're on drugs doesn't mean that it doesn't suck to be them...
On the other hand, you really can get equipment that reliable if you're willing to pay for it, and component reliability has improved wonderfully since the 1980s, e.g. disk drive MTBFs of 500000-1M hours instead of 10,000 hours, so you really can wait until midnight slowdown to rebuild the RAID partition after you hot-swap the drive, and computers are a couple orders of magnitude faster so you need fewer of them to get a given job done fast enough, making it much easier to make subsystems reliable and monitor their status.
Fragmentation is another problem besides leaking, but it can also lead to systems getting progressively slower until they drop below some critical performance threshold.
And disk drives _do_ fill up with log files unless you do something about it.
Back when my department used Vaxen, we'd reboot them every Friday night, fsck the disks, and do backups. Around the time we were running SVR2, the file system really was stable enough and the removable disk packs high enough quality that fsck seldom found anything and didn't need manual intervention, and the rebooting process was reliable enough that we could let a cron job run it, and while we could have cut back to monthly, people had gotten in the habit of knowing the machine would be down, so they could get a life, and it was a good schedule for the times we actually did want to to upgrades or hardware maintenance.
Failover clusters aren't trivial - I worked on a non-winning design for one of the predecessors to this system back in the late 80s (fortunately for us, we lost, and unfortunately for IBM, they won.) Yes, you can have two, three, or N of everything, but then you need a lot of code watching the redundant components to see if any of them appear to be failing, and code deciding which redundant subsystem is correct if two or more of them disagree, and code watching the watchers to make sure they're still watching well, and data communication protocols that work ok when all messages are transmitted redundantly to the redundant processors, possibly getting different results at microsecondly different times. One of my coworkers had worked with an early "fault-tolerant computer" system which had triply or quadruply redundant hardware, but had an operating system that crashed at least weekly because it was too complex.
You also have to be extremely careful and flexible in your design for the granularity of the redundant subsystems - if you make the separately processed chunks too big or too small, you can have an order of magnitude change in performance and sometimes several orders of magnitude change in reliability, and then there's the problem that the definition of "reliability" includes "probability that the calculation finishes in N milliseconds", so it's inextricably linked with performance.
Moore's Law is really your friend here. Improved performance means you can use a lot fewer parts, which reduces complexity and failures. Disk drives are more than an order of magnitude more reliable, and the increase in size means that a cluster of disks containing N gigabytes is several orders of magnitude more reliable because it's a lot fewer disks, and CPUs that are 2+ orders of magnitude faster mean that it's easier to guarantee that something happens in a given time, and cuts down on communications steps between different modules, so you cut down on all the failure modes for those communications, and on the monitoring software watching for failures, and on the failures of the monitoring software. On the other hand, Moore's Law lets operating system vendors and application vendors bloat their software with features - X Windows 11R3 ran just fine on my 386/25 machine with 8MB RAM, but of course I was using twm, not Enlightenment or Gnome.
Backup plans do introduce the danger of complexity - the FAA doctrines of the 1980s were that any new system had to be able to interoperate with everything its predecessor interoperated with, because you weren't going to flash-cut upgrade everything at once. That meant that everything you designed had to be bug-for-bug backwards compatible with the predecessor's interfaces, and when they redesigned the thing _your_ system interoperates with, it has to be bug-for-bug compatible with everything your system does, which means being compatible with its predecessor, which was compatible with your system's predecessor, etc. It's a vicious circle similar to the messes Windows and Intel CPUs had to put up with, except that while the 8088 and MS-DOS 1.0 were *ugly*, they were at least small and well-documented late-70s technology, as opposed to poorly-documented 1960s JOVIAL and 1950s analog.
The 40-year-old system was pretty much the Mos Eisley of software design - you'll never see a more wretched hive of scum, villainy, and undocumented unmaintainable Jovial code running on IBM 360/50 and 360/90 hardware. The backup system was much cleaner (and much dumber); I think the main thing they did in the 1970s enhancement was retread the design to use transistors instead of vacuum tubes, though I never worked directly with that side.
Yes, Sun and IBM machines fail - that's why all of the critical parts in our designs had to be at least doubly redundant, and often triply redundant, because the design spec of "Eight 9s of reliability" meant that doing an hour a year of preventive maintenance might expose you to too much risk from the backup system failing. I haven't seen IBM's design; I was on the lucky team that didn't win the bidding to build the final system, unlike the poor suckers at IBM who had to implement theirs, but the requirements were not only insanely non-implementable, they were excessively focussed on No Possible Downtime Ever, because if anything goes wrong resulting in an airline crash, the FAA gets insane amounts of political heat. Doesn't matter if the system is N years late, because you can try to blame the contractor for that, or if you can't fly supersonic planes across the Continental US because they're too fast for the new ARTCCs, because tough luck for the French and for bi-coastal business travellers.
Of course, that doesn't mean that Im inclined to trust a system running on Windows, either...
I was on the lucky team that *lost* the bidding for the replacement system; IBM's team were the poor bastards who won, and were stuck investing seven years into building an unbuildable replacement, pouring billions of dollars down the drain while being micromanaged by the FAA, who didn't know much about software design or reliability in spite of having a methodology that required producing 175 design documents over the optimistically 3-year design period.
Supposing they could make a profit at that price, they'd be able to sell to a lot of people. The price of the raw internet bandwidth feeding their towers has come way down; the real question is how many users their frequency bands can handle if they're actually using the system.
Back when Metricom was around, and DSL mostly wasn't, most of my geeky friends had it, for prices like $40/month, compared to $20 for dialup, and at least one non-geeky coworker had it supporting the four users in his household who'd otherwise all have phone lines and modems. (Of course Metricom _died_ at those prices, but they were running really customer equipment back when raw Internet upstream was expensive, and they never got the customer volume to get economies of scale. By contrast, wireless phone companies already have tons of volume - they're just short on clues.)
Alas, like many lightweight machines, part of the lightness was because of flimsy construction, so it's gradually fallen apart, and you can no longer buy critical parts like the little plastic doors that hold in the batteries. (Too bad - it's sensibly designed to use standard video-camera batteries, so they'd be replaceable at a quasi-rational price if the silly doors weren't missing.) That all became less relevant when the screen cracked, so the machine now looks like a nice little keyboard that has VGA and Ethernet interfaces instead of PS2.
It's kind of like hacking cars - taking off the muffler might have been fun for 15 minutes when you were 16, but everybody's heard it before and it just sounds like you did it because your muffler had rusted out anyway, so no sense annoying the neighborhood, and that 42-tune electronic horn widget had gotten old by 1980 and makes you look dorky, not retro. There's the occasional quasi-new hack, like the Bubb Rubb Car Whistle , but even that just invites people to break your car windows or spackle your muffler shut.
So all that 1337 h4x0rin's become just another day job, like stealing hubcaps for profit, or graffitiing telephone poles with signs about "Make Money Fast - call 1-800-SCAM-MER"; it's mostly taken over by automated systems or underpaid losers selling to spammers. If you want to have _fun_ hacking cars, you can just as well do legitimate things like make art cars or improve your gas mileage, and if you want to do 31337 on your computer, might as well do something interesting like write new software, or hunt down spammers, or at least find ways to hack MMORPGs so you can frag your friends.
But you _can_ play a game called Linux, and it lets you play games like Adventure and Nethack and TuxRacer (actually, I don't know if TuxRacer runs on the PS2, not having one to try it with), and play as many dimensions of MUD as you want to...
Also, extra credit if you can run this as a Beowulf cluster, though of course the really interesting Playstation cluster system uses the Emotion Engine graphics chip to do number crunching rather than the relatively tame CPU so it's not a straight Beowulf system. This new Playstation version is 1/3 the size and 1/2 the price of the ones NCSA used, so it's an even more attractive idea...
So run Knoppix from the CDROM or DVD with your /home directory mounted across the network. (Actually, if the new one only has 32MB of RAM, you probably want some smaller CDROM-bootable competitor.) You've presumably got _some_ kind of Unix machine available as a server, or else you can run a Windows file sharing protocol and SAMBA.
If you're a kid, and your parents buy you the gamer box for Christmas, they're probably not buying every game out there along with it; they're buying you the box and a couple of popular games. You can use that money from Granny to buy a couple more games once you're bored, e.g. the Decmeber 27th sales.
Sure, as an individual you want to be able to have a phone that works in emergencies - but most of the discussion about this topic is run by policy wonks from one group or another. I suppose I did leave out the FCC types who are annoyed that VOIP doesn't look like their current regulatory structure, and it's hard to justify radically increasing the cost of an essentially free service with the excuse that you're doing it to make service affordable.
The reason bots are potentially bad for business is that they could drive away other players, though of course the house doesn't care if a table full of bots are playing each other as long as there are enough of them.
... is like "military intelligence" or "jumbo shrimp".
Of course VOIP and 911 don't get along - 911 was designed to work in a landline environment, with communications architectures tightly tied to Class 5 telco switches and database architectures designed for phones that stay in one place, and the 911 folks haven't been willing to adapt their systems to accept VOIP connections even though it wouldn't be that hard. VOIP, like wireless, presents some new technical challenges because the equipment is portable, and if you bring your VOIP box on a business trip with you and have to call the fire department, you want firetrucks showing up where you are, not back at your house. But there are ways to design around it, whether you do something with GPS or adapt your DHCP servers to pass you geographical info or whether you have the VOIP box/software/etc. let you tell it your address.
Complaints about VOIP and 911 are usually a cover for real complaints about VOIP and wiretapping. The folks who like wiretapping are annoyed that changing technology makes their tools obsolete, and want to force the technology to adapt to them, rather than the other way around, and they tend to use 911 as a lever to do that. After all, you want an ambulance to be able to find you if you're hurt, but you probably don't want the police to be able to locate you within 10 meters and follow you all day, so that's not the motivation they advertise for mandating that new cell systems provide user location. Similarly, the wiretappers _really_ don't like peer-to-peer flexible technology, and they're used to having hooks into traditional telcos to control them.
My family gets together in the Colorado mountains in the summer, at about 8300 feet. It does take me a day or so to acclimatize, if I haven't already been in Denver on business. My grandmother had to stop going in her early 90s because the air was a bit too thin, but she was in pretty good shape for her age. (Ok, she mostly sat around in her rocking chair, but by those years that was what she usually did back in Kansas City as well, and she liked the mountains better.) After a day at 8300, going up to 11000 feet is still hard to do much physical exertion without dizziness, but it's not bad. 14000 was a lot more trouble, but back when I was hiking at that altitude I was one of those "scientists not well known for their physical abilities"; now I'm in my late 40s instead of late 30s, but I was already fat by then :-) On the other hand, I've known a lot of scientists who are in much better physical condition, especially younger ones who haven't gotten fat. Some of the people I met while hiking out there lived at 10000 feet, so for them it was a piece of cake, and the ultramarathoners who I met halfway up the mountain (me going up, them coming down) had been around there training for a week. And the cold is partly from dryness - tweaking humidity helps a lot.
Sure, above 7000 meters, it's nice to have oxygen tanks, and above 8000, it's really nasty not to have it, though a few people don't use it. But astronomers aren't likely to be that high up, and even if they are, they do most of their work indoors, and the buildings can have pressurized air or oxygen concentrators (plus the buildings can have heat, which is a real problem for mountain climbers.) Meanwhile, the telescope gets to look through all the lack of atmosphere above it.
And the cost of building two of these things is much less than twice the cost of building a single one, because a large fraction of the cost is developing all the tools and technology to build it, and they can crank out two or three more for not much extra cost. (Obviously building the base and staffing it are duplicated costs.) By contrast, building all the launching systems for the Hubble is so expensive that you're not going to build a couple of clones and launch them, you're going to wait another decade and develop most of the system from scratch using the technology of the time again.