> Maybe we should organize a field trip to go > through their "that will never work" idea room. > I'm sure we could find a use for some of the > stuff in there.
One shot, and they've got my whole house mapped out, right down to the ant crawling on the floor. When they come to bust be for hacking my grades on the school's system, I'll have no where to hide.
Screw petty computer criminals; think of the potential this has for suppressing those nasty political dissidents!
They were shining light thru skin and measuring some parameter X that was correlated with sugar level...
Could the kind of interferometry described here give better results? ie better correlation?
No, just more accurate measurements. Which might improve results, but not the correlation you're measuring. The actuall degree of correlation doesn't change. There's an analogy in applied statistics: more careful measurements don't help if your samples aren't any good.
If the technology ever hits the commercial market, it'll likely be a bag of several hundred little marble-like sensors that you scatter around the area you want to film. They'd communicate using RF or IR or something wireless, and you'd have a receiver hooked up to your box that would collect the data from all of them.
Then, you'd run the happy little proprietary number-crunching software and it'd build the 3d models for you. Probably not in realtime, for a desktop machine.
One difficult thing, though, is that you need to know the relative locations of all the sensors at recording time to some decent level of precision (i.e. they need figure out where they all are relative to each other, or you need to indicate their positions somehow, in 3d yet), or the data is pretty worthless. You could I suppose lay them out on a gridded mat or similar, but that's not convenient for filming in a lot of settings.
By the way, if USB were still involved, while USB is kind of complex to hack, the groundwork for Linux generic USB support is pretty much laid now. It might not be as bad as you think.
Not like it matters; most of the magic here is in the user-land signal-processing software, not the device drivers. (don't think the algorithms aren't patented; fat chance seeing libre software doing this anytime in your lifetime) ---
I suspect the way it works, since they brought interferometry into it, is that they have a large number of small sensors (mini-cameras) scattered around and throughout the area being "filmed".
You'd have to apply algorithms similar to those used for CT to reconstruct the a 3d model from the data from the sensors, and interferometry comes in to play to compensate for the low spatial resolution of the individual sensors (as well as removing the need for them to be "focused" -- the more sensors, the sharper the detail).
Of course you'd still have problems with opacity, but with enough sensors scattered around that wouldn't be such an intractable problem. You just couldn't see the insides of stuff that you didn't have a window of some kind into and couldn't put a sensor inside of.
Stuff wouldn't need to be in full view of all of the sensors, either -- just, the fewer sensors that can see a feature, the more fuzzy ("out of focus") it'd be. ---
Brady's research was funded in part by the Department of Defense Advanced Research Projects Agency (DARPA), which he said would like to use it for military applications.
A camera that worked without having to focus would be ''smarter,'' he said. ``They have cameras spread throughout the world -- a lot more cameras than people,'' he said. These include camera viewing from satellites.
> Heh look, there are people out there BREATHING > for free. Dammit, I want to develop an AIR > product and sell it, but all of this free air > stuff is narrowing consumer choice and stifling > innovation in the air market.
That argument reminds me of Gates' infamous "Open Letter to Hobbyists", which ushered in the current age of proprietary software. ---
> If we didn't have two competing browsers, maybe > we could just stick to the open standards.
Doubtful. Any time you only have one primary implementation of something, that implementation (with all its flaws) will supercede any defined standards whether or not they are open.
It's sad but true: no matter who the coders are, if they don't need to worry about interoperability with other implementations, they aren't going to give a shit about adhering to standards, open or not.
This is as true for Free Software authors as it is for Microsoft.
[ think about it: if Mozilla was the _only_ web browser to exist, do you seriously think they _wouldn't_ start making extensions/modifications to the standards that would be adopted as the de-facto new standards? (although likely they'd document them properly and keep them open, as well as putting them through the proper standards processes eventually) ] ---
Because 3d is much, much harder to design. Right now, 2d is relatively easy for a human designer to keep track of, but 3d is very very hard to visualise without severe loss of information.
Additionally, routing software and other tools related to design right now just aren't equipped to deal with especially 3-dimensional designs. Throw a third dimension in and you complicate routing dramatically.
Then there's the problem of heat dissipation. It'll get real hot in the middle of that silicon cube.
i.e. 3d chip designs are doable, yes, but they're so much trouble that most designers/producers don't feel that it's worth it right now. I'm sure we'll get to it eventually when we run out of other options. ---
> Who are the unwashed hordes of barbarians at the > gate who eventually win... the goths, visigoths, > vandals and franks?
...us hairy Unix folk, of course. Don't tell me that the sysadmin next door with the two-foot beard, sandals and braided hair doesn't look JUST LIKE SAID BARBARIANS...
...although personally I look like a strung-out accountant with the goatee, rockports and slicked-back hair... ---
Check out the history of the clone() system call. I might be wrong about the specific patchlevel, but I know it was added sometime in the 1.3 series (the number 1.3.19 keeps coming to mind now).
I have absolutely no idea about SMP support. Was primitive SMP availible in 2.0.0?
We could use some better kernel support for some thread-related things like shared POSIX semaphores (SYSV semaphores don't count) and things, but we do already have kernel threads themselves.
We've had soft realtime since I don't know when, but certainly in 2.0... (I still run 2.0.36, and I'm the maintainer of some shell utilities for dealing with the realtime scheduler (check out rt-utils on Freshmeat); I should know that it's there). For hard realtime, we just need the rtLinux merge.
> Why anyone would want to use an OS that can't do > something as simple as bind 4-8 eathernet cards > together to create one fat pipe is beond me.
Why anyone would want to use an OS that doesn't quite seem to quite "get" large portions of the TCP/IP spec is beyond me.
But yeah, binding the ethernet cards to the same IP like that is a nice feature. Although I find myself wondering how that works without severely mangling ARP/RARP...
[ nitpick: there's no 'a' in "ethernet" ]
> In NT you can do that in under 10 mouse clicks. > Ahahahaha No waite 7 mouse clicks.
If this is such a large site, you've got some kind of automated setup procedure for these servers and workstations, right? I can't imagine having to walk around to each and every machine and execute 10 or 7 or even 5 mouse clicks...
[ nitpick: wait is spelled W-A-I-T, and you're missing some punctuation ]
> I work for a Fortune 100 company and we did our > owne in house benchmarks on VT vs. Linux and > that's why we now are 100% NT+NTworkstation. We > discovered that being able to put 2 100Mbit > eathernet cards in each Workstation and 4 on the > server gave us several times the performace that > the Linux server (by the wasy was twice as > powerfull in CPU Mhz than the NT server ) gave > us.
I won't ask what you people are doing that requires this kind of bandwith, since you're obviously only concerned with peak throughput...
...but let me get this straight: in a situation where the main bottleneck is network I/O, you're deriding Linux networking because of performance comparisons between a Linux server with one NIC, and an NT server with four?
I trust you are aware that you can take advantage of multiple network interfaces even without binding them to the same IP? (and in fact you can get some additional advantages there, if you're clever about switched ethernet and routing)
Also, you _did_ use the same client configuarations for both tests, didn't you?
[ nitpick: no 'e' in "own". and there's no 'a' in "ethernet" ]
> And since our network is firewalled ( FW-1 ) > then we don't have any security problems to > worry about. Not even Viruses cuz they are > stoped at the Firewall.
I assume you mean proxy+firewall, not just firewall. I sincerely hope that you're not in any way responsible for security at your site; this is an extremely naive attitude.
Firewalls don't magically make your network secure by themselves (yes, they _help_, but...). Crackers like networks with a hard candy shell and a soft, gooey center. You want to be giving the bastards jawbreakers to suck on instead. Preferably jalapeno flavor.
[ nitpick: the plural of Virus is Virii. I'll overlook the informal "cuz". "stopped" has two p's. ]
[ self-nitpick: I'm missing the accents in naive and jalapeno ]
It's a good thing you didn't reveal what corporation you work for; your post does not exactly scream professionalism. ---
At minimum, I do a thorough skimming and grepping before I kick off./configure. Something insanely huge like gcc, I obviously couldn't do the whole thing, but I try to get a good look, and grep for exec() and system() at least.
There are a lot of folks who are even more paranoid than I, certainly about particular packages -- i.e. ones that they have some interest in developing.
So, while I suspect the majority of people do not in fact look at the source of the OSS app they install, it doesn't really matter; compared to CSS, there are AT LEAST several hundred times more people scrutinizing the source code, a good number of which are not affiliated with the original author. That increases my comfort level a lot.
So what if OSS doesn't reach 100% eyeball utilization? The fact is that compared to any other approach, OSS consistently gets the maximum number of eyeballs, period.
That being said, I do strongly encourage those that aren't reading source already but can to take more of an interest in the security of your system. Take the time to peruse the source you download. It benefits us all, and it's often a good opportunity to learn all kinds of things, too! ---
So let's assume that the government has a hypercluster of computers that are a billion billion times faster, en masse, than the ENTIRE distributed.net.
It would still take them 1e20/1e18=100 years to break _ONE_ 128 bit key.
--- snip ---
That's only true if they don't know about any analysis techniques or weaknesses in the algorithm that we don't. I don't think that's a safe assumption. ---
> Won't argue with you there. It isn't any wonder, > and it does kinda-sorta invalidate some of the > utopian "we can all work together" propaganda.
No argument there. Anyone who thinks that people won't be real jerks at least some of the time (even where there is considerable counterincentive) need to get in better touch with reality.
> The people promoting stuff like this are > shineing a really bright light on the whole OSS > concept.
Are they? Last time I checked, SETI@Home certainly was not OSS, and their decsision to avoid OSS doesn't appear to have helped them significantly either.
You can maybe say something about shining a really bright light on the inherent good of mankind concept, but this is neiter a vindication nor an indictment of OSS specifically. OSS depends on the same kinds of you-scratch-my-back-i'll-scratch-yours economic exchanges we're used to, just as much as it does on altrusim.
[ n.b. economic != monetary ]
Very few people would do OSS if they couldn't realisitically expect to get something back, and that's usually more than just a clean conscience. However, neither is it World Domination...
> One world, one nation, one GNU. er... let me > rephrase that....
Slashdot Mirror
*falls off chair in spasmodic fits of laughter as co-workers look on bewilderment*
---
> Maybe we should organize a field trip to go
> through their "that will never work" idea room.
> I'm sure we could find a use for some of the
> stuff in there.
Hey, it worked for Steve Jobs.
*grins*
---
What about a non-reversible hash of the IP, though?
---
One shot, and they've got my whole house mapped out, right down to the ant crawling on the floor. When they come to bust be for hacking my grades on the school's system, I'll have no where to hide.
Screw petty computer criminals; think of the potential this has for suppressing those nasty political dissidents!
---
They were shining light thru skin and measuring some parameter X that was correlated with sugar level...
Could the kind of interferometry described here give better results? ie better correlation?
No, just more accurate measurements. Which might improve results, but not the correlation you're measuring. The actuall degree of correlation doesn't change. There's an analogy in applied statistics: more careful measurements don't help if your samples aren't any good.
---
Read my speculations on how it works:
Hrm; I just discovered these people's site on a post below, too... it looks like my conjectures were pretty dead on:
http://www.phs.uiuc.edu/4Is/
---
If the technology ever hits the commercial market, it'll likely be a bag of several hundred little marble-like sensors that you scatter around the area you want to film. They'd communicate using RF or IR or something wireless, and you'd have a receiver hooked up to your box that would collect the data from all of them.
Then, you'd run the happy little proprietary number-crunching software and it'd build the 3d models for you. Probably not in realtime, for a desktop machine.
One difficult thing, though, is that you need to know the relative locations of all the sensors at recording time to some decent level of precision (i.e. they need figure out where they all are relative to each other, or you need to indicate their positions somehow, in 3d yet), or the data is pretty worthless. You could I suppose lay them out on a gridded mat or similar, but that's not convenient for filming in a lot of settings.
By the way, if USB were still involved, while USB is kind of complex to hack, the groundwork for Linux generic USB support is pretty much laid now. It might not be as bad as you think.
Not like it matters; most of the magic here is in the user-land signal-processing software, not the device drivers. (don't think the algorithms aren't patented; fat chance seeing libre software doing this anytime in your lifetime)
---
I suspect the way it works, since they brought interferometry into it, is that they have a large number of small sensors (mini-cameras) scattered around and throughout the area being "filmed".
You'd have to apply algorithms similar to those used for CT to reconstruct the a 3d model from the data from the sensors, and interferometry comes in to play to compensate for the low spatial resolution of the individual sensors (as well as removing the need for them to be "focused" -- the more sensors, the sharper the detail).
Of course you'd still have problems with opacity, but with enough sensors scattered around that wouldn't be such an intractable problem. You just couldn't see the insides of stuff that you didn't have a window of some kind into and couldn't put a sensor inside of.
Stuff wouldn't need to be in full view of all of the sensors, either -- just, the fewer sensors that can see a feature, the more fuzzy ("out of focus") it'd be.
---
Brady's research was funded in part by the Department of Defense Advanced Research Projects Agency (DARPA), which he said would like to use it for military applications.
A camera that worked without having to focus would be ''smarter,'' he said. ``They have cameras spread throughout the world -- a lot more cameras than people,'' he said. These include camera viewing from satellites.
Got that warm and fuzzy feeling, uh-huh.
---
> Heh look, there are people out there BREATHING
> for free. Dammit, I want to develop an AIR
> product and sell it, but all of this free air
> stuff is narrowing consumer choice and stifling
> innovation in the air market.
That argument reminds me of Gates' infamous "Open Letter to Hobbyists", which ushered in the current age of proprietary software.
---
> If we didn't have two competing browsers, maybe
> we could just stick to the open standards.
Doubtful. Any time you only have one primary implementation of something, that implementation (with all its flaws) will supercede any defined standards whether or not they are open.
It's sad but true: no matter who the coders are, if they don't need to worry about interoperability with other implementations, they aren't going to give a shit about adhering to standards, open or not.
This is as true for Free Software authors as it is for Microsoft.
[ think about it: if Mozilla was the _only_ web browser to exist, do you seriously think they _wouldn't_ start making extensions/modifications to the standards that would be adopted as the de-facto new standards? (although likely they'd document them properly and keep them open, as well as putting them through the proper standards processes eventually) ]
---
> why don't designers use 3D designs?
Because 3d is much, much harder to design. Right now, 2d is relatively easy for a human designer to keep track of, but 3d is very very hard to visualise without severe loss of information.
Additionally, routing software and other tools related to design right now just aren't equipped to deal with especially 3-dimensional designs. Throw a third dimension in and you complicate routing dramatically.
Then there's the problem of heat dissipation. It'll get real hot in the middle of that silicon cube.
i.e. 3d chip designs are doable, yes, but they're so much trouble that most designers/producers don't feel that it's worth it right now. I'm sure we'll get to it eventually when we run out of other options.
---
> if you weren't white you might understand.
I happen to be white, and even I understand enough to be at least mildly offended by this dork
---
There was a /. article on this yesterday or the day before, I believe.
---
> but the same logic says that they should also
> try to protect the pseudo-humans called fetuses
Although I somewhat argree with your general point, just how are they pseudo-humans?
---
> Who are the unwashed hordes of barbarians at the
> gate who eventually win... the goths, visigoths,
> vandals and franks?
...us hairy Unix folk, of course. Don't tell me that the sysadmin next door with the two-foot beard, sandals and braided hair doesn't look JUST LIKE SAID BARBARIANS...
...although personally I look like a strung-out accountant with the goatee, rockports and slicked-back hair...
---
Check out the history of the clone() system call. I might be wrong about the specific patchlevel, but I know it was added sometime in the 1.3 series (the number 1.3.19 keeps coming to mind now).
I have absolutely no idea about SMP support. Was primitive SMP availible in 2.0.0?
---
We could use some better kernel support for some thread-related things like shared POSIX semaphores (SYSV semaphores don't count) and things, but we do already have kernel threads themselves.
We've had soft realtime since I don't know when, but certainly in 2.0... (I still run 2.0.36, and I'm the maintainer of some shell utilities for dealing with the realtime scheduler (check out rt-utils on Freshmeat); I should know that it's there). For hard realtime, we just need the rtLinux merge.
---
sendmail.
---
...representation of web data, which is patented if I recall. There was a /. article on it a while back.
---
> Why anyone would want to use an OS that can't do
> something as simple as bind 4-8 eathernet cards
> together to create one fat pipe is beond me.
Why anyone would want to use an OS that doesn't quite seem to quite "get" large portions of the TCP/IP spec is beyond me.
But yeah, binding the ethernet cards to the same IP like that is a nice feature. Although I find myself wondering how that works without severely mangling ARP/RARP...
[ nitpick: there's no 'a' in "ethernet" ]
> In NT you can do that in under 10 mouse clicks.
> Ahahahaha No waite 7 mouse clicks.
If this is such a large site, you've got some kind of automated setup procedure for these servers and workstations, right? I can't imagine having to walk around to each and every machine and execute 10 or 7 or even 5 mouse clicks...
[ nitpick: wait is spelled W-A-I-T, and you're missing some punctuation ]
> I work for a Fortune 100 company and we did our
> owne in house benchmarks on VT vs. Linux and
> that's why we now are 100% NT+NTworkstation. We
> discovered that being able to put 2 100Mbit
> eathernet cards in each Workstation and 4 on the
> server gave us several times the performace that
> the Linux server (by the wasy was twice as
> powerfull in CPU Mhz than the NT server ) gave
> us.
I won't ask what you people are doing that requires this kind of bandwith, since you're obviously only concerned with peak throughput...
...but let me get this straight: in a situation where the main bottleneck is network I/O, you're deriding Linux networking because of performance comparisons between a Linux server with one NIC, and an NT server with four?
I trust you are aware that you can take advantage of multiple network interfaces even without binding them to the same IP? (and in fact you can get some additional advantages there, if you're clever about switched ethernet and routing)
Also, you _did_ use the same client configuarations for both tests, didn't you?
[ nitpick: no 'e' in "own". and there's no 'a' in "ethernet" ]
> And since our network is firewalled ( FW-1 )
> then we don't have any security problems to
> worry about. Not even Viruses cuz they are
> stoped at the Firewall.
I assume you mean proxy+firewall, not just firewall. I sincerely hope that you're not in any way responsible for security at your site; this is an extremely naive attitude.
Firewalls don't magically make your network secure by themselves (yes, they _help_, but...). Crackers like networks with a hard candy shell and a soft, gooey center. You want to be giving the bastards jawbreakers to suck on instead. Preferably jalapeno flavor.
[ nitpick: the plural of Virus is Virii. I'll overlook the informal "cuz". "stopped" has two p's. ]
[ self-nitpick: I'm missing the accents in naive and jalapeno ]
It's a good thing you didn't reveal what corporation you work for; your post does not exactly scream professionalism.
---
Run a nested X server as the sandbox user.
---
At minimum, I do a thorough skimming and grepping before I kick off ./configure. Something insanely huge like gcc, I obviously couldn't do the whole thing, but I try to get a good look, and grep for exec() and system() at least.
There are a lot of folks who are even more paranoid than I, certainly about particular packages -- i.e. ones that they have some interest in developing.
So, while I suspect the majority of people do not in fact look at the source of the OSS app they install, it doesn't really matter; compared to CSS, there are AT LEAST several hundred times more people scrutinizing the source code, a good number of which are not affiliated with the original author. That increases my comfort level a lot.
So what if OSS doesn't reach 100% eyeball utilization? The fact is that compared to any other approach, OSS consistently gets the maximum number of eyeballs, period.
That being said, I do strongly encourage those that aren't reading source already but can to take more of an interest in the security of your system. Take the time to peruse the source you download. It benefits us all, and it's often a good opportunity to learn all kinds of things, too!
---
--- snip ---
So let's assume that the government has a hypercluster of computers that are a billion billion times faster, en masse, than the ENTIRE
distributed.net.
It would still take them 1e20/1e18=100 years to break _ONE_ 128 bit key.
--- snip ---
That's only true if they don't know about any analysis techniques or weaknesses in the algorithm that we don't. I don't think that's a safe assumption.
---
> Won't argue with you there. It isn't any wonder,
> and it does kinda-sorta invalidate some of the
> utopian "we can all work together" propaganda.
No argument there. Anyone who thinks that people won't be real jerks at least some of the time (even where there is considerable counterincentive) need to get in better touch with reality.
> The people promoting stuff like this are
> shineing a really bright light on the whole OSS
> concept.
Are they? Last time I checked, SETI@Home certainly was not OSS, and their decsision to avoid OSS doesn't appear to have helped them significantly either.
You can maybe say something about shining a really bright light on the inherent good of mankind concept, but this is neiter a vindication nor an indictment of OSS specifically. OSS depends on the same kinds of you-scratch-my-back-i'll-scratch-yours economic exchanges we're used to, just as much as it does on altrusim.
[ n.b. economic != monetary ]
Very few people would do OSS if they couldn't realisitically expect to get something back, and that's usually more than just a clean conscience. However, neither is it World Domination...
> One world, one nation, one GNU. er... let me
> rephrase that....
Godwin's law, you lose.
---