Slashdot Mirror
Spray-On Computers
Jack William Bell writes "Edinburgh University has funding for a program to create spray on computers. The basic idea is to make thousands of tiny 'silicon specks' or 'smart sand' (a step larger than smart dust) which work together via wireless networking to provide 'ubiquitous computing.' No, the idea itself isn't new. But it is interesting to see someone actually working on it. The initial application is a spray you apply to the chest of heart patients, creating a sensor array to report their health back to the hospital."
that geeks start coding in graffiti?
[Insert Spray-On Hair-Club-For-Men Joke Here]
.
_
Then the patient could really be in trouble if some one accidentally sent "dd if=/dev/zero of=/dev/hda"
boy I sure would not give out my root password.
OH THE SHAME I fell off the wagon and use sigs again!
Most of the geeks I work with can't even figure out how to spray on deodorant.
The idea of spraying silicon chips directly onto patients should be approached with caution.
Great! So I guess in the future, I can look forward to have my daily /. page
generated by a squadron of albino spray-on paint.
Sometimes I wonder if this effect i planned by the IT industry. With quantum and DNA computing on the way, we will see in a few decades computers which are extremely powerful but also also extremely cheap. Obiviously the profit margins of the industry will drop below vaccuum energy levels. Therefore they have to find a way to make people more and more computers besides the incredible power of a single machine. The easiest way to do this is the make computers more powerful but less efficient.
I would even conjecture that this idea is behind the introduction of XML, web services and grid computing. Normal computer operations are overlayed with bloated protocols and documents to decrease to efficiency of modern servers and workstations forcing people into new upgrade cycles.
Ever wondered by why XML is not binary based ? Computers don't care if humans can read their data. Or why bloated XML is used for web services where simple binary based RPC would do the same job ? Or ever the perversion of putting it on top of HTTP, introducing new security holes by making trditional firewalls useless ?
Owner of a Mensa membership card.
I want some high-tech painball loads of this stuff, so I can get a tactical advantage feeding the enemy's location into a little heads-up-display...
Okay, I'll just stick with Unreal Tournament for now. Proper exercise can be a unimplemented goal, can't it...?
~ Whence do you come, slayer of men, or where are you going, conqueror of space?
I don't know about you, but I don't really see the point of using an aerosol can to make computers. Yeah, I get the medical thing (and I think it's a good idea) but why do you have to spray it on? Why not simply apply it with some tweezers and glue? Is it funner to shake up a can, say "Close your eyes!" and press down on the little button?
With this you could take a pen and spray it and it becomes an interface in its own right.
Umm... okay... So everyone has a can of "Kwik-n-Ez Spray-on computers" sitting on their desk and then they say, "I want to interface with my computer!" shake up the can and spray their pen. "There! I typed a command, now I'd better throw the pen out and when I need to interface again I'll shake it up and spray!"
1.3 million pound grant eh?
/usr/bin/complain >
do these devices have enough compute horsepower to handle encryption? do you really want your heart status to be "broadcast out"? can they autonegotiate proper encryption for correct data exchange? all these smart dust, smart pebbles, etc. plans (especially those intended for military purposes) - it looks like major security needs to be built in, but at least so far the track record has been pretty bad (WEP).
I am afraid this is just another grant hunt. There is no battery that would be suitable for this project.
Ever wondered by why XML is not binary based ? Computers don't care if humans can read their data.
Oh, but humans do. If you're developing such systems it's many times easier to debug human-readable text instead of binary stuff. You can also see what's going on more easily. See below.
Or why bloated XML is used for web services where simple binary based RPC would do the same job ?
Because being plain text allows for easier access, modification and searching by humans AND computers. No more locking in to proprietary binary formats. Would you like each manufacturer to have their own binary RPC protocol, all of them being incompatible with all others?
Or ever the perversion of putting it on top of HTTP, introducing new security holes by making trditional firewalls useless ?
It needs to have something as bearer, so why not use something tried and tested. What would you have suggested? A completely new protocol? What a waste of time and resources. Besides, a firewall alone does not quarantee your safety. It's not like if you allow web traffic, suddenly everyone and their dog is able to r00t your box. You allow port 80 but you disable RPC stuff in the server and that's it.
Have I been trolled?
First of all, DNA computers will probably never be practical. Not because of the equipment involved, but because their is a hard limit on the amount of dna that can be involved in a single computation.
Also, you are correct that it is not feasable for a network to achieve linear time speed-up in the number of processors. Theta(n/log(n)) would be great, though, and sufficient for most applications.
http://yetanotherpoliticalrant.blogspot.com
I'm sure a lot of the grant money is for working out how to encapsulate the sillicon chips with teflon or titanium (or other stuff like exotic carbon) in strategic arrangements so that their outsides are biocompatible and conductive in just the right places, so that one day Bones can wave some strange glowey thing over you, say "Well, i'm just a simple country doctor," and bring you back from the dead.
On the other hand poisonous smart sand would make a fantastic weapon. Make it be able to hop or crawl by giving it little piezoelectric actuators and you could make an ordinary looking patch of sand suddenly come to life when activated by an enemy soldier's touch, swarming over his body and forcing itself under his gas mask seals and into his eyes, nose, ears and mouth. And then it could hide out in the body, waiting for the next victim.
lastly, a listening device the size of a grain of sand could be put into your bedroom and you'd never find it. (hands up who regularly even vaccuumes their room? I always forget. I don't even know how to spell the verb properly.) They say that the goal is to computerise objects by simply spraying smart sand onto it: What if somebody decided to (without your knowledge) computerise your pillow stuffing? Do you know how many Americans pray out their deepest darkest secrets on their knees by their bedsides each night?
If these walls could talk...
Would you get a spontaneous Beowulf cluter?
I'm not sure if you're joking or not, so I'll answer seriously: The whole idea of nanotech is to be small and cheap. A lot of useful research can be done with $2 million (equivalent) if you're not wasting money on a marketing dept., executive salaries, and all the other corporate bullshit. The academic research environment has shown time and time again that it gets more done dollar for dollar (pound for pound, euro for euro, whatever) than the corporate research environment, which is why most genuine science (as opposed to engineering) comes out of academic labs. In fact, it's a nice refutation of the "free market is always more efficient" mantra beloved of both "privatize everything" politicians and university administrators -- who are basically whores for the oxymoronic "business schools" and sports teams.
;)
[steps back, looks at post]
Wow, that was a bit of a rant, wasn't it? I didn't realize I had that much venom stored up on this subject.
The correlation between ignorance of statistics and using "correlation is not causation" as an argument is close to 1.
XML's textual nature keeps XML documents "open". I did a Google search a while ago when trying to determine whether there was a standard scheme for putting binary data into XML, and came across somebody discussing parse times for XML. He'd written a program which parsed XML and saved the parsed version in a binary format, and found that it was actually faster to just parse the XML again than to reconstruct the information from his binary format.
On one hand, sure, perhaps his coding wasn't up to the level of those who'd written the XML parser, which might account for the slower loading.
On the other hand, there's a good chance that someone writing an XML parser intended for general inclusion within other programs is paying a lot more attention to doing things right (in terms of speed, security, etc.) than the person who writes a quick data-parsing routine.
Computers don't care if humans can read their data, sure. But, humans do. Parsing a simple, well-defined text format isn't computationally expensive, and makes the job easier for those who might have cause to view the data.
Finally, XML tags can and do carry important semantic information. It's much easier to write a program to parse an arbitrary unknown XML schema (say, GnuCash's file format) than to have to reverse-engineer an arbitrary unknown binary format (yes, I've done both). This is important because it helps to ensure that the data isn't quite as bound to the program -- parsing and conversion between schemas is generally much easier than, say, translating a WordPerfect document to MS Word format.
Hans Reiser has even decided to use text in his transaction-control syscalls:
Text parsing isn't as bad as people like to make it, as long as you aren't parsing a horribly ugly specification (like, say, C++ code).
Besides, how is carrying something on top of HTTP going to introduce new security holes? I haven't been able to decide where you are implying the issues might arise.
Somebody get that guy an ambulance!