Here's another image I've mirrored for the few people that will see this post. It's looking down the main isle, to the front.
I'm not karma-whoring, honest. If I am moderated up, well, I'm on a 33.6 and I could very well be Slashdotted worse than the original host of this web site. =)
This article seems sort of biased and heavy on the marketing hype. Here are my repsonses to a few notable things written in the article.
Windows 2000/XP/ME only. Win98 users need not apply. [page 1; image caption]
For what reasons is Windows 98 not supported? For the same reasons as Windows 98 won't run, will {Linux,{Free,Net,Open}BSD} be affected? They do not go into any detail on this whatsoever.
It is FINALLY time to get rid of that old Dexxa ball mouse that came with your first SVGA card. While you're at it, toss out that old Dot Matrix printer, and even the $13 keyboard with the ASDFJKL: keys completely rubbed off! Abit steps into a new era of computing with the MAX boards. [page 1]
What's wrong with my old ball mouse? It works. Why should PS/2 keyboards and mice be replaced by USB? It seems overkill, especially for keyboards, where things should always work, even if your OS is having problems seeing your USB controller. With USB, too many things can break, and leave your system in a hard-to-fix state. (Ever added 'usb-uhci' instead of 'usb-ohci' to/etc/modules, or equivalent? Many OSs don't have USB support in the installer; Debian doesn't at least, and I don't think any of the BSDs do. Does the BIOS emulate an AT interface for USB keyboards?) More importantly, new eras of computing have little to do with shedding "legacy" devices. New eras of computing are set in software ideology and design. Even as far as hardware goes, new eras of computing would be more akin to a next generation of processor, or a new archetecture in CPUs or busses enabling more than 16 IRQ lines. That last sentence seems very marketroid.
Despite having all the next-gen high performance capabilities, SCSI is still absent. This proves that while the AT7 is a very high end board, it is still targeted to the consumer market. Thanks, Abit! [page 1; bottom]
The "consumer" market doesn't use 12 IDE devices. The high-end hobbyist/server/gaming markets do. I think it's a shame they left out SCSI, which seems inconsistent with trying to shed "legacy devices". SCSI is great. Why are you thanking them?
Included are a set of nifty black IDE cables... This is a great way to have some nice looking custom IDE cables without worrying about using rounded cables (IDE cables are flat for a reason you know!). [page 2]
I use rounded cables (that I make myself) to improve airflow and increase the ease of routing/positioning cables inside my cases. I don't care about how they look. And no, I didn't know of a reason IDE cables are flat, besides conventional manufacturing techniques in use. Too bad you don't explain why, because the first thing I'd do with these cables is use a razor blade to slice them into segments, and bunch the segments together with zip-ties, as I would any other flat cable.
It is expected that most cases will start using MediaXP panels, rather than their own Mickey Mouse panels you see now. [page 2; bottom]
Uh huh. I take it these MediaXP panels are specific to Abit and their licensees. How much are you being paid by Abit again?
Gee, it sure looks like a big gain in performance, but this is like comparing dog poo to cat poo...
Ugh I can't bear to look at that... Let's get to the OpenGL numbers... [page 6; bottom]
Comparing "dog poo to cat poo"? Maybe you meant comparing apples and oranges. "Ugh"? The reviewer seems to have come straight from the AOL chat rooms... Okay, so I say "ugh" from time to time, but you don't use that word in a written review. That's fairly unprofessional.
One more gripe, but somewhat offtopic: at the bottom of every page, I'm told to use IE 5+ and a 1024x768x32-bit screen. There is no excuse for bad web design that depends on a specific browser configuration like this. </pissyrant>
Hackers obviously want anonymity when they're looking to trade personal information that they've obtained via identity theft, so Internet Relay Chat is a commonly used mechanism," says Harrington.
Anonymity? Let's see what happens when I/whois myself on my favorite IRC server:
Hey, that's my host name! That translates into a Real World IP address, 207.225.41.92. The website was specified in the "real name" value provided by one's IRC client, which is readily changable. Often, the user name ('piranha' in '~piranha@dialup...') is changable as well, without ident on the client's site. But the host name is not.
I don't see how IRC is any more anonymous than AOL Instant Messenger chat rooms or ezboard.com or other "mainstream" communications fora. (0-day credit card numbers on Slashdot?)
For those that don't know, any user can/whois any other user on IRC and get their IP address. Very, very few servers intentionally obfuscate the host name to prevent tracing it down to a real person. As far as I know, the major networks (EFnet, Dalnet, Undernet, etc) do not whatsoever. Some silly IRC servers don't check reverse lookups to see if they match IP addresses (for example, if I come from 11.22.33.44, and 44.33.22.11.in-addr.arpa resolves to www.whitehouse.gov, www.whitehouse.gov won't resolve to 11.22.33.44), but exploiting that requires that you have control of your own reverse DNS zone. (Did I get my terminology right?) And even if people do pull that off, an administrator at the IRC server's site can pull off one of probably several tricks to see what the real IP address is. (Like, start a tcpdump session, saving to a log file, and CTCP or/msg the person, and see where the packet goes to.)
By the way, it's sort of ironic I chose irc.2600.net for this example in a retort to this article; this server really isn't full of 3r33t h4x0rz, and I have observed no identity theft or other fraud on #2600. If you don't believe me, stop by and see for yourself, which is apparently a bit too much to ask for from the people at CNN.
All the socks that ever disappeared could simultaneously materialize in your dryer!
This all too strangely reminds me of a Ren and Stimpy cartoon...
Ren and Stimpy were in some sort of space vehicle, and they were pulled into a black hole. At the other end was a strange world, where all the real world's left socks had disappeared to...
Sort of. Mostly since I started encoding my CDs to Vorbis, I've also started to redundantly encode to Flac, a lossless format that gets about 2:1 compression. I'll eventually get around to burning these to data CD for archival. (On some, but not all audio CDs, I could fit two Flac'd CDs to a data disc, and data discs have higher error-correction than audio discs.)
I encode to Flac in case Vorbis hits the ground (maybe another format will supercede it?), or if I found myself in your situation, needing MP3s. With Flac I just have to decode them and reencode them to whichever format I want, with no loss in quality, and not spend time re-ripping and ensuring cdparanoia did its job properly.
This thing is nothing more than a PC with an LCD on the front...
I'll concur with that; this music appliance seems a bit overrated. I'd think they could at least make it significantly smaller than it presently.
Or for the geek that appreciates higher quality audio.
Vorbis audio is higher quality than MP3 at equivilant bitrates
I can hear artifacts in MP3s encoded at 128kbps (haven't tried Ogg Vorbis @ 128kbps)
Fraunhofer's (et al.) patent(s) on MP3 technology put a significant hurdle in front of free and commercial software writers, as well as developers of portable digital audio hardware, such as this device.
Xiphophoros's Vorbis libraries are licensed BSD-style, and the rest is licensed under the GPL.
If you don't like the licenses, the standard itself is 100% open, and you can create proprietary implementations that don't credit Xiph.
Ogg Vorbis is superior, save it's not as wide-spread as MP3. And it's free in both of the cliche senses. I'd recommend reading the FAQ.
Would you say Linux and BSD geeks just think that they are nonconformist when they don't use Windows?
It's not just ISP AUPs that could prevent me from legally reselling bandwidth. (Not that I would with my 33.6k!)
I work for a local commercial ISP, and among other things I signed away on my NDA, I can't sell internet service. I'd imagine that's the case with a lot of employers' NDAs and/or other employment agreements.
So someone starts sending fake summons via email with faked return addresses. How do you know what's real in E-mail.[?]
I'm personally a big fan of crypto, and the idea that lots of things (for instance, being served legal documents) can be conducted online securely with existing technology.
If you've ever played with PGP, you should know what a signature is. If not, try reading the GNU Privacy Handbook (some parts are GnuPG specific, but lots or most of it pertains to OpenPGP in general). The result is there could very well be a standard public key for every court jurisdiciton (and if you want to get detailed, these keys could have a chain of signatures, higher districts signing the keys of lower districts). These keys could be used to sign documents being served, and the availability of the public keys could be used to verify the document's authenticity.
AFAIK, crypto won't solve the issue of documents being served to dead addresses, or people denying that they were ever served.
If we're going to rely on PKE/PKI, we need to get this "security" (particularly Windows security) travesty in order. Public advocacy of the importance of general system security, and the responsibility of big software vendors (cough, Microsoft) to release patches quickly and make it easy for home end-users to get patched regularly (automatically?) could play a role in removing the number of worms and trojan horses spreading on Windows-based PCs. Lack of system security compromises the integrity of people's online identities and the potential role of PKE in the legal system.
IANARC (I Am Not A Real Cypherpunk); there's probably a host of issues I haven't covered that must be taken into consideration.
Ever hear of a hash? What about a signatures in a PKI environment? Have you ever considered their real-world usage?
Hypothetically, those papers could have had a unique 128-bit key on them; similar looking to an md5sum, but randomly generated. Mathematically unfeasible to guess, or even brute-force. (Increase bit-size appropriately if you disagree with me.) The supervisor reads the papers, then enters the number or reads a barcode with the number into his terminal. The terminal establishes a cryptographically secure connection to one of several computers, and downloads a signed copy of the exact same message that appears on paper, signed by an authorizing agent of the airline. The supervisor carefully compares the two documents to ensure the digitally signed copy and the hard-copy are the same.
Practical use of cryptography isn't so far fetched. With today's technology, forgery of paper documents like in this scenario could very well be prevented.
Some might ask, "why make all these special accommodations for the Cyborg Man?" I'd be seriously worried if the airlines weren't already using strong cryptography for digital communication. The infrastructure ought to be there already.
(Of course, I have no reason to believe the documents were ever digitally signed. I'm just pointing out a very realistic possibility.)
Again, as always, I blame Microsoft for dumbing-down the computer industry and removing functionality by making their crippled system the only standard people have to bother supporting.
Woah... what's that have to do with Microsoft and Windows or IE? The reason some misguided web site operators have programmed their scripts to assume one IP address is one individual end-user doesn't have anything to do with Microsoft. It has to do with ISPs handing out unique IP addresses to their customers (dialup, DSL, cable, or whatever). The ISP's customer could be using Windows or Linux or MacOS or AIX or whatever, and still get a unique IP address. This trend wasn't started by Microsoft, AFAIK.
There isn't anything in Microsoft software that I'm aware of to prevent multiple Windows users on one site from having the same outside, visible IP address. Microsoft's IP stack, as an internal client, works with various implementations of NAT just as well as any other IP stack, and MSIE can be configured to use an HTTP proxy server--another way two visitors from one site will appear to a web server to be one IP address. Lastly, a Windows computer, running Citrix software, can host multiple windowing sessions remotely accessable by a thin client--analogous to VNC clients or X servers on thin clients connecting to one server running the same software from that central server (Netscape, for example).
What I don't understand is the parallel you made with Microsoft and the practice of ISPs assigning one IP address to each individual end-user. Microsoft hasn't "removed functionality" that would otherwise allow multiple clients to use one IP address, and they haven't "crippled" their system in a way that pertains to this.
Jesus, people are silly. I have a feeling I'm being trolled here. But just in case...
Computers don't just break down. Sometimes the mechanical parts do, like hard disks and fans, and volatile components like CRTs. But, nobody told you to leave your computer on 24/7 for, eg, distributed.net. If you run distributed.net, you can leave off your computer in the same pattern that you do now, and still make a positive contribution towards such a project. Running distributed.net or some other distributed computing client will not shorten the lifespan of your computer.
Who said you can't control who's using your resources? The "Internet-Spanning OS" could very well be designed to allow end-users to weigh priorities to certain groups and individuals that will be consuming the resources. At present with systems like distributed.net and seti@home, you could choose one over the other, or configure one to run at a higher/lower CPU priority than the other.
Nobody said you had to spend more money on your computer to accommodate distributed applications. The whole idea is for the distributed clients to only use spare CPU cycles, spare disk space, etc. Ideally, you wouldn't notice any depletion of resources. And if you did, you could always switch to a more efficient distributed framework, or choose to not run one at all.
I'm not saying that everyone must allow their computer to participate in distributed computing, I'm just saying there's no harm in doing so, and very little would even notice that such computing is even taking place, after performing any initial installation of the client software. The car analogy doesn't work: you don't need to pay for or perform maintenance while running software like distributed.net, nor do you need to pay for additional resources; your CPU would only be running zeros otherwise, and you don't have to change your computer usage habits to run distributed clients.
Well, first off, there's some obvious heating implications with a thing like this. After all, you've wrapped a potentially (and probably quite) heat-emitting device in a warm, fuzzy, very insulating enclosure. It'd be interesting if they figured out a way to put in air ducts or something for cooling.
Besides the heat issue, I think this is a really cool idea, but I would have tried to be more decorative about the whole "Borg" thing, like looping lots of unused wires in and out of the "skin", especially around the head. Maybe mount a laser-pointer (or a CueCat?!) in front of one of the eyes.
"I ***know*** what it running on my system. I know this because I built the binaries myself. I know this because I can look at the source code and see what it does." (emphesis added)
A couple points:
You can look at the source yourself? Woohoo. But do you actually audit the code yourself? Are you competent enough to not only know what every line of code is doing, but know enough to look out for obvious backdoors, and not-so-obvious backdoors disguised as unintentional security holes? Compiling from source is little better than using the binaries unless you've answered 'yes' to these questions.
Let's say you finished looking at the source code you use and finished auditing it thoroughly, on a box called 'Sparcy' some 2 months ago. You even wrote down the version numbers. Now let's say you download the source again because Sparcy's hard drive crashed, and you need to recompile everything. You hit the usual sites and download the same versions. How do you know the distributors or mirror operators didn't add a few unmentioned features? What about your ISP, and their uplink? Can you trust everyone from point A to point B? The same would apply to one of your e-mail buddies or even a reputable friend telling you GnuPG 1.06 is "clean".
For the latter situation, you should write down md5sums for anything important that you audit like that. But mostly the point was this really isn't practical.
Suggestions? Maybe rely on a group of trusted people to audit code for you and the general public, PGP-signing the md5sums as their official seal-of-approval. How about using a system like Freenet to collaborate and organize such an effort? Freenet is uncensorable, and provides for anonymous publication and retrieval of content. A method of code peer-review that the feds/aliens/Elvis/Jesus/Dr. Evil can't stop, bribe, or censor, or otherwise put to a halt. Unfortunately, Freenet is far from mainstream. However, it's stuff like this Freenet is designed for, and does the best job for.
By the way, if such a group/organization is formed, there's no reason they couldn't compile binaries in the exact environment popular distribution packagers build packages, that is, by using the exact same libraries, etc. Theoretically, if a source package can be marked clean, and a distributor's diffs can be marked clean, the exact binary could be reproduced, and the package management utilities could recreate what should be the official package binary. If it really is exactly the same as the official package, that package (and the md5sum!) could be declared as "clean" as well. I don't know about other distributions/OSs, but Debian and NetBSD provide for detailed build logs, and probably give enough information to recreate the official packages.
Dude. My main PC is a 333MHz Pentium II with 192 megs of RAM. I run 2.4.17, and my system boots probably in under a minute. That's more than acceptable for me. I run XFree86 4.1 and GNOME 1.4 (with the default GTK theme; no pixmaps), and Sawfish 1.0. No issues. Takes about a minute or two for GNOME to fire up completely, but I rarely need to restart my X server.
The X clients I use most commonly are xchat, xmms, Eterm, Everybuddy, Galeon, and Opera. None of those programs get slow enough to the point of "pain," or even exceedingly testing my patience.
The only times I have issues with speed are when I work on large, multilayered images in Gimp, or boot up Windows 98 under VMware (with 32 megs of RAM allocated to the VM) to test a site in MSIE.
I consider my computer to be somewhat older, but not ancient, and certainly reasonable as a box I spend 90% of my non-working, non-sleeping time in front of. People really make out the slightly older hardware to be a huge pain in the ass. Give me a break.
After reading a few posts, I was inspired and came up with some ideas for a neat 3D user environment:
More or less, this environment would be a virtual-reality sort of thing. The user customizes the layout of the 3D scene that they inhabit on their computer, with both physical elements (like structures; walls, ceilings, trees, whatever), and programmable elements
Have an integrated, virtual X server, to run X normal apps and clients. A virtual X server would render the appearance of the clients' windows into dynamic bitmaps. The bitmaps could be placed in the 3D environment as if it were a tile, in ways not possible without the use of a virtual X server (that I can think of), like tilting the window in different directions, or perhaps shaping it to a 3D object, like a convex surface.
Programmable elements would allow a user to use this VR environment as their primary interface to the computer (for better or worse). The shape or behavior of objects and structures could change, or new physical elements could be created in the environment, based on internal conditions (user walked to a certain corner of the room), or conditions external to the environment (user has new mail).
Have the option of running your own little VR server, so that other people may use their VR client to explore and enjoy your space, or designated areas of your space. The transition to another person's "space" could be very obvious, like explicitly starting your VR client from a shell prompt, and stating the remote person's VR server as an arguement. Or, you could program a portal into your own space; the portal could be animated and mysterious-looking, or could just be represented as the threshold of a door in your space. Perhaps a user and their friends could set up a small network of portals between their spaces.
Design the whole thing so people can trade room/environment designs and elements, as well as scripting code, so that people that don't have a lot of time on their hands or don't know the scripting language can still enjoy the system.
Other possibilities could include some sort of a gaming environment integrated into the system. (It could be a distributed, open-source EverQuest. =) The more open it is and the more effort put into it to ensure expandability, the more possibilities down the road. (As if I or anyone else is going to make it happen, anyway.)
> Get a journaled FS. The reason is that as long as your system is up and running, having a fs like ext2 is no problem. But if you ever have crashes, long fsck (that something fail) means downtime.
Well, my company has a very production Mandrake 7.1 box running ReiserFS. This box went down unexpectedly due to a shot CPU fan, and I had to try getting the system up as fast as possible. It booted, and reiserfsck worked its five-second magic as expected, and the system worked for all of about 20 seconds, until a fatal filesystem error popped up on the console (forgot what it was exactly) and I had to reboot into single-user mode to try to fix the problem. On a couple of the filesystems, reiserfsck told me that I needed to use the --rebuild-tree option, which it said was a somewhat risky thing to do, and that I should back up the filesystem I was going to use it on. So I did. That took about 10 minutes. When I actually ran reiserfsck --rebuild-tree, it took at least 15 or 20 minutes to complete.
In the end, reiserfs costed at least half an hour of downtime on a production box, but the backup I made wasn't needed, as there appeared to be no lost files. To be fair, the kernel was 2.2.15-4mdk, so the implementation is probably a bit dated. This post is just meant to serve as a reminder that not all FS implementations are years mature, and in some cases you may run into little glitches. That aside, reiserfs is great.
Slashdot Mirror
I'm not karma-whoring, honest. If I am moderated up, well, I'm on a 33.6 and I could very well be Slashdotted worse than the original host of this web site. =)
For what reasons is Windows 98 not supported? For the same reasons as Windows 98 won't run, will {Linux,{Free,Net,Open}BSD} be affected? They do not go into any detail on this whatsoever.
What's wrong with my old ball mouse? It works. Why should PS/2 keyboards and mice be replaced by USB? It seems overkill, especially for keyboards, where things should always work, even if your OS is having problems seeing your USB controller. With USB, too many things can break, and leave your system in a hard-to-fix state. (Ever added 'usb-uhci' instead of 'usb-ohci' to /etc/modules, or equivalent? Many OSs don't have USB support in the installer; Debian doesn't at least, and I don't think any of the BSDs do. Does the BIOS emulate an AT interface for USB keyboards?) More importantly, new eras of computing have little to do with shedding "legacy" devices. New eras of computing are set in software ideology and design. Even as far as hardware goes, new eras of computing would be more akin to a next generation of processor, or a new archetecture in CPUs or busses enabling more than 16 IRQ lines. That last sentence seems very marketroid.
The "consumer" market doesn't use 12 IDE devices. The high-end hobbyist/server/gaming markets do. I think it's a shame they left out SCSI, which seems inconsistent with trying to shed "legacy devices". SCSI is great. Why are you thanking them?
I use rounded cables (that I make myself) to improve airflow and increase the ease of routing/positioning cables inside my cases. I don't care about how they look. And no, I didn't know of a reason IDE cables are flat, besides conventional manufacturing techniques in use. Too bad you don't explain why, because the first thing I'd do with these cables is use a razor blade to slice them into segments, and bunch the segments together with zip-ties, as I would any other flat cable.
Uh huh. I take it these MediaXP panels are specific to Abit and their licensees. How much are you being paid by Abit again?
Comparing "dog poo to cat poo"? Maybe you meant comparing apples and oranges. "Ugh"? The reviewer seems to have come straight from the AOL chat rooms... Okay, so I say "ugh" from time to time, but you don't use that word in a written review. That's fairly unprofessional.
One more gripe, but somewhat offtopic: at the bottom of every page, I'm told to use IE 5+ and a 1024x768x32-bit screen. There is no excuse for bad web design that depends on a specific browser configuration like this. </pissyrant>
Anonymity? Let's see what happens when I /whois myself on my favorite IRC server:
Hey, that's my host name! That translates into a Real World IP address, 207.225.41.92. The website was specified in the "real name" value provided by one's IRC client, which is readily changable. Often, the user name ('piranha' in '~piranha@dialup...') is changable as well, without ident on the client's site. But the host name is not.
I don't see how IRC is any more anonymous than AOL Instant Messenger chat rooms or ezboard.com or other "mainstream" communications fora. (0-day credit card numbers on Slashdot?)
For those that don't know, any user can /whois any other user on IRC and get their IP address. Very, very few servers intentionally obfuscate the host name to prevent tracing it down to a real person. As far as I know, the major networks (EFnet, Dalnet, Undernet, etc) do not whatsoever. Some silly IRC servers don't check reverse lookups to see if they match IP addresses (for example, if I come from 11.22.33.44, and 44.33.22.11.in-addr.arpa resolves to www.whitehouse.gov, www.whitehouse.gov won't resolve to 11.22.33.44), but exploiting that requires that you have control of your own reverse DNS zone. (Did I get my terminology right?) And even if people do pull that off, an administrator at the IRC server's site can pull off one of probably several tricks to see what the real IP address is. (Like, start a tcpdump session, saving to a log file, and CTCP or /msg the person, and see where the packet goes to.)
By the way, it's sort of ironic I chose irc.2600.net for this example in a retort to this article; this server really isn't full of 3r33t h4x0rz, and I have observed no identity theft or other fraud on #2600. If you don't believe me, stop by and see for yourself, which is apparently a bit too much to ask for from the people at CNN.
This all too strangely reminds me of a Ren and Stimpy cartoon...
Ren and Stimpy were in some sort of space vehicle, and they were pulled into a black hole. At the other end was a strange world, where all the real world's left socks had disappeared to...
Strange, indeed.
Sort of. Mostly since I started encoding my CDs to Vorbis, I've also started to redundantly encode to Flac, a lossless format that gets about 2:1 compression. I'll eventually get around to burning these to data CD for archival. (On some, but not all audio CDs, I could fit two Flac'd CDs to a data disc, and data discs have higher error-correction than audio discs.)
I encode to Flac in case Vorbis hits the ground (maybe another format will supercede it?), or if I found myself in your situation, needing MP3s. With Flac I just have to decode them and reencode them to whichever format I want, with no loss in quality, and not spend time re-ripping and ensuring cdparanoia did its job properly.
I'll concur with that; this music appliance seems a bit overrated. I'd think they could at least make it significantly smaller than it presently.
Ogg Vorbis is superior, save it's not as wide-spread as MP3. And it's free in both of the cliche senses. I'd recommend reading the FAQ.
Would you say Linux and BSD geeks just think that they are nonconformist when they don't use Windows?
I work for a local commercial ISP, and among other things I signed away on my NDA, I can't sell internet service. I'd imagine that's the case with a lot of employers' NDAs and/or other employment agreements.
I'm personally a big fan of crypto, and the idea that lots of things (for instance, being served legal documents) can be conducted online securely with existing technology.
If you've ever played with PGP, you should know what a signature is. If not, try reading the GNU Privacy Handbook (some parts are GnuPG specific, but lots or most of it pertains to OpenPGP in general). The result is there could very well be a standard public key for every court jurisdiciton (and if you want to get detailed, these keys could have a chain of signatures, higher districts signing the keys of lower districts). These keys could be used to sign documents being served, and the availability of the public keys could be used to verify the document's authenticity.
AFAIK, crypto won't solve the issue of documents being served to dead addresses, or people denying that they were ever served.
If we're going to rely on PKE/PKI, we need to get this "security" (particularly Windows security) travesty in order. Public advocacy of the importance of general system security, and the responsibility of big software vendors (cough, Microsoft) to release patches quickly and make it easy for home end-users to get patched regularly (automatically?) could play a role in removing the number of worms and trojan horses spreading on Windows-based PCs. Lack of system security compromises the integrity of people's online identities and the potential role of PKE in the legal system.
IANARC (I Am Not A Real Cypherpunk); there's probably a host of issues I haven't covered that must be taken into consideration.
Yeah, and CueCats changed how I use the Internet forever.
(OTOH, CueCats were marginally neat. I'll be interested in knowing if these are any fun, as well.)
Hypothetically, those papers could have had a unique 128-bit key on them; similar looking to an md5sum, but randomly generated. Mathematically unfeasible to guess, or even brute-force. (Increase bit-size appropriately if you disagree with me.) The supervisor reads the papers, then enters the number or reads a barcode with the number into his terminal. The terminal establishes a cryptographically secure connection to one of several computers, and downloads a signed copy of the exact same message that appears on paper, signed by an authorizing agent of the airline. The supervisor carefully compares the two documents to ensure the digitally signed copy and the hard-copy are the same.
Practical use of cryptography isn't so far fetched. With today's technology, forgery of paper documents like in this scenario could very well be prevented.
Some might ask, "why make all these special accommodations for the Cyborg Man?" I'd be seriously worried if the airlines weren't already using strong cryptography for digital communication. The infrastructure ought to be there already.
(Of course, I have no reason to believe the documents were ever digitally signed. I'm just pointing out a very realistic possibility.)
Woah ... what's that have to do with Microsoft and Windows or IE? The reason some misguided web site operators have programmed their scripts to assume one IP address is one individual end-user doesn't have anything to do with Microsoft. It has to do with ISPs handing out unique IP addresses to their customers (dialup, DSL, cable, or whatever). The ISP's customer could be using Windows or Linux or MacOS or AIX or whatever, and still get a unique IP address. This trend wasn't started by Microsoft, AFAIK.
There isn't anything in Microsoft software that I'm aware of to prevent multiple Windows users on one site from having the same outside, visible IP address. Microsoft's IP stack, as an internal client, works with various implementations of NAT just as well as any other IP stack, and MSIE can be configured to use an HTTP proxy server--another way two visitors from one site will appear to a web server to be one IP address. Lastly, a Windows computer, running Citrix software, can host multiple windowing sessions remotely accessable by a thin client--analogous to VNC clients or X servers on thin clients connecting to one server running the same software from that central server (Netscape, for example).
What I don't understand is the parallel you made with Microsoft and the practice of ISPs assigning one IP address to each individual end-user. Microsoft hasn't "removed functionality" that would otherwise allow multiple clients to use one IP address, and they haven't "crippled" their system in a way that pertains to this.
If I'm mistaken, please tell me how.
Computers don't just break down. Sometimes the mechanical parts do, like hard disks and fans, and volatile components like CRTs. But, nobody told you to leave your computer on 24/7 for, eg, distributed.net. If you run distributed.net, you can leave off your computer in the same pattern that you do now, and still make a positive contribution towards such a project. Running distributed.net or some other distributed computing client will not shorten the lifespan of your computer.
Who said you can't control who's using your resources? The "Internet-Spanning OS" could very well be designed to allow end-users to weigh priorities to certain groups and individuals that will be consuming the resources. At present with systems like distributed.net and seti@home, you could choose one over the other, or configure one to run at a higher/lower CPU priority than the other.
Nobody said you had to spend more money on your computer to accommodate distributed applications. The whole idea is for the distributed clients to only use spare CPU cycles, spare disk space, etc. Ideally, you wouldn't notice any depletion of resources. And if you did, you could always switch to a more efficient distributed framework, or choose to not run one at all.
I'm not saying that everyone must allow their computer to participate in distributed computing, I'm just saying there's no harm in doing so, and very little would even notice that such computing is even taking place, after performing any initial installation of the client software. The car analogy doesn't work: you don't need to pay for or perform maintenance while running software like distributed.net, nor do you need to pay for additional resources; your CPU would only be running zeros otherwise, and you don't have to change your computer usage habits to run distributed clients.
Besides the heat issue, I think this is a really cool idea, but I would have tried to be more decorative about the whole "Borg" thing, like looping lots of unused wires in and out of the "skin", especially around the head. Maybe mount a laser-pointer (or a CueCat?!) in front of one of the eyes.
Or, maybe I'm getting carried away.
A couple points:
For the latter situation, you should write down md5sums for anything important that you audit like that. But mostly the point was this really isn't practical.
Suggestions? Maybe rely on a group of trusted people to audit code for you and the general public, PGP-signing the md5sums as their official seal-of-approval. How about using a system like Freenet to collaborate and organize such an effort? Freenet is uncensorable, and provides for anonymous publication and retrieval of content. A method of code peer-review that the feds/aliens/Elvis/Jesus/Dr. Evil can't stop, bribe, or censor, or otherwise put to a halt. Unfortunately, Freenet is far from mainstream. However, it's stuff like this Freenet is designed for, and does the best job for.
By the way, if such a group/organization is formed, there's no reason they couldn't compile binaries in the exact environment popular distribution packagers build packages, that is, by using the exact same libraries, etc. Theoretically, if a source package can be marked clean, and a distributor's diffs can be marked clean, the exact binary could be reproduced, and the package management utilities could recreate what should be the official package binary. If it really is exactly the same as the official package, that package (and the md5sum!) could be declared as "clean" as well. I don't know about other distributions/OSs, but Debian and NetBSD provide for detailed build logs, and probably give enough information to recreate the official packages.
Dude. My main PC is a 333MHz Pentium II with 192 megs of RAM. I run 2.4.17, and my system boots probably in under a minute. That's more than acceptable for me. I run XFree86 4.1 and GNOME 1.4 (with the default GTK theme; no pixmaps), and Sawfish 1.0. No issues. Takes about a minute or two for GNOME to fire up completely, but I rarely need to restart my X server.
The X clients I use most commonly are xchat, xmms, Eterm, Everybuddy, Galeon, and Opera. None of those programs get slow enough to the point of "pain," or even exceedingly testing my patience.
The only times I have issues with speed are when I work on large, multilayered images in Gimp, or boot up Windows 98 under VMware (with 32 megs of RAM allocated to the VM) to test a site in MSIE.
I consider my computer to be somewhat older, but not ancient, and certainly reasonable as a box I spend 90% of my non-working, non-sleeping time in front of. People really make out the slightly older hardware to be a huge pain in the ass. Give me a break.
But you could remove the plexiglass cover while it's running and draw a spiral on it with a marker. What fun!
- More or less, this environment would be a virtual-reality sort of thing. The user customizes the layout of the 3D scene that they inhabit on their computer, with both physical elements (like structures; walls, ceilings, trees, whatever), and programmable elements
- Have an integrated, virtual X server, to run X normal apps and clients. A virtual X server would render the appearance of the clients' windows into dynamic bitmaps. The bitmaps could be placed in the 3D environment as if it were a tile, in ways not possible without the use of a virtual X server (that I can think of), like tilting the window in different directions, or perhaps shaping it to a 3D object, like a convex surface.
- Programmable elements would allow a user to use this VR environment as their primary interface to the computer (for better or worse). The shape or behavior of objects and structures could change, or new physical elements could be created in the environment, based on internal conditions (user walked to a certain corner of the room), or conditions external to the environment (user has new mail).
- Have the option of running your own little VR server, so that other people may use their VR client to explore and enjoy your space, or designated areas of your space. The transition to another person's "space" could be very obvious, like explicitly starting your VR client from a shell prompt, and stating the remote person's VR server as an arguement. Or, you could program a portal into your own space; the portal could be animated and mysterious-looking, or could just be represented as the threshold of a door in your space. Perhaps a user and their friends could set up a small network of portals between their spaces.
- Design the whole thing so people can trade room/environment designs and elements, as well as scripting code, so that people that don't have a lot of time on their hands or don't know the scripting language can still enjoy the system.
Other possibilities could include some sort of a gaming environment integrated into the system. (It could be a distributed, open-source EverQuest. =) The more open it is and the more effort put into it to ensure expandability, the more possibilities down the road. (As if I or anyone else is going to make it happen, anyway.)Well, my company has a very production Mandrake 7.1 box running ReiserFS. This box went down unexpectedly due to a shot CPU fan, and I had to try getting the system up as fast as possible. It booted, and reiserfsck worked its five-second magic as expected, and the system worked for all of about 20 seconds, until a fatal filesystem error popped up on the console (forgot what it was exactly) and I had to reboot into single-user mode to try to fix the problem. On a couple of the filesystems, reiserfsck told me that I needed to use the --rebuild-tree option, which it said was a somewhat risky thing to do, and that I should back up the filesystem I was going to use it on. So I did. That took about 10 minutes. When I actually ran reiserfsck --rebuild-tree, it took at least 15 or 20 minutes to complete.
In the end, reiserfs costed at least half an hour of downtime on a production box, but the backup I made wasn't needed, as there appeared to be no lost files. To be fair, the kernel was 2.2.15-4mdk, so the implementation is probably a bit dated. This post is just meant to serve as a reminder that not all FS implementations are years mature, and in some cases you may run into little glitches. That aside, reiserfs is great.