Slashdot Mirror
The Year In Ideas
No_Weak_Heart writes "The New York Times Magazine (registration required) presents its annual compendium of ideas. The list ranges from acoustic keyboard eavesdropping to land-mine-detecting plants to water that isn't wet. What catches your fancy? And what do you think is missing?"
I didn't RTFA, but I noticed my ski bike isn't on there. Neither is my shopping cart grocery trailer! Whats is this, a popularity contest?
http://craig.backfire.ca/imgbrowse/ski-bike/
It is a hydraulic engine with which you can build motors of any size. Want to rotate the Pentagon? It is possible with the Hercules motor:
http://www.indrives.com/frameset.html
I think that land-mine plant could have an extra benifit - if countries who refuse to sign land-mine bans continue to use them (COUGH USA COUGH) someone needs to fill a plane with these seeds and drop them everywhere they think land mines are being used - but not after the war, during the war! render them totally useless as a weapon by revealing their locations days after they have been set! Although the plant still seems a little creepy...
This comment does not represent the views or opinions of the user.
>You can even use the time between strokes as a crude measure of distance between (unknown) keys, or as a hint as to what kind of stuff is being typed (c code will sound different from a memo, even if the keys are all the same) to improve your frequency analysis
g .pdf
My advisor (Dawn Song) has a paper (with other people, of course) about timing analysis of interactive ssh sessions. Basically, the upshot is that you can watch how long it is between packets that come out, and you get one packet per keystroke (iirc), so you can use this to learn about what they're typing. It's reasonably difficult, of course, but the microphone attack does gain extra information which the ssh attack does not.
If you're interested, a pdf is at http://www.ece.cmu.edu/~dawnsong/papers/ssh-timin
Lea
Saw a Danish documentary on those landmine detecting plants. Funny enough they used the dehydrated water to "water" their plants. It was due to the seeds being so small that they could be carried away by the wind. On a different note, it seems as if the guys that developed the plants are having a hard time in getting the right clearances, some english chap that was in the documentary, working for the team as an observer, and link to the African country's government, ended up trying to wreck the whole projekt because he was afraid of genetically engineered plants.........(Note to guy, if your country is chock full of landmines, a few extra plants is not going to ruin your day)
Tax intellectual property the way we tax real property. If you don't value it enough to pay the tax, sell it or put it in the public domain.
This would bring unused ideas into the mainstream.
Thanks, that's a nice paper; I'll have to read the Viterbi algorithms in more detail later though it's nice to see pure info. theory put to nefarious ends. =)
The upshot of an empirical 50x reduction in workload for password-cracking from timing information alone is surprising and disturbing. However, even the rudimentary position info from a multi-microphone analysis should at least double that (in the case of a high-latency digraph it can give order information, e.g. tell OZ from ZO; otherwise, it will tell you where the flurry of keystrokes is occurring and possibly let you break up the HMM into one for each hand if you have a model for how humans type).
In the "A Fire Upon The Deep" universe, the Powers of the higher computational zones are hypothesized to be able to perform powerful computations on minimal data.
The keyboard thing is a great example of that; with scanty data you can reverse engineer what keys are being tapped.
I'd bet with a bit more work you wouldn't even need to calibrate the device, just collect a lot of keypresses, classify them blind, and apply known probability distributions to the data. With that you could get a high probability analysis of the keypresses. (After all, if the two most probable passwords are "thebeatles" or "theb]atles", which do you think it is?)
A single picture or a short sound doesn't have a lot of data in it, but a long sound sample or video file has a lot of data in it. Expect this to be just the beginning.
The NYT has my real e-mail address and in return I find real NYT news content in my in-box each morning, something I want and need. I suspect that is true of most of those who register.
The tinfoil hat market being what it is this days, I doubt the Times worries much about the Slashdot demographic.