Stanford engineer aims to connect the world with ant-sized radios

Actually, most of the uses of the word "emergent" I encounter are exactly as the LW article describes. If you'd replace the word "emergent" with "magic", you'd learn nothing more and nothing less from a sentence. [0]

As for feeling foolish after playing with evolutionary algorithms - I'm not the OP but I can relate somewhat given how I saw people learning evolutionary algorithms and neural networks at my university (and I'm pretty sure it's not a local phenomenon). Evolutionary algorithms are usually explained as inspired by biological evolution, with implicit (and sometimes explicit) note that "evolution made us, therefore evolution is superpowerful, therefore evolutionary algorithms - which are just evolution in code - will be superpowerful too!". Except they're not, and the whole concept is bullshit. It's a belief in Random Number God. Throw enough shit at the wall and something will stick. Evolution is terribly, terribly inefficient, and so are the evolutionary algorithms.

Sure, this inefficiency gives them some interesting properties that may help them avoid particular types of local optimas, etc. But those are mathematical features of an algorithm type, and have nothing to do and share no power with evolution, or magic.

The whole problem stems from people trying to transfer virtues of biology to computing by using a surface metaphor. There's a post on LW that covers it nicely:

http://lesswrong.com/lw/vx/failure_by_analogy/

"So... why didn't the flapping-wing designs work? Birds flap wings and they fly. The flying machine flaps its wings. Why, oh why, doesn't it fly?"

Or about neural networks,

"A backprop network with sigmoid units... actually doesn't much resemble biology at all. Around as much as a voodoo doll resembles its victim. The surface shape may look vaguely similar in extremely superficial aspects at a first glance. But the interiors and behaviors, and basically the whole thing apart from the surface, are nothing at all alike. All that biological neurons have in common with gradient-optimization ANNs is... the spiderwebby look."

I encounter a lot of similar "medieval thinking" in CS departments. I don't know why. It probably goes in common with the concept of not caring about how the world works.

Oh, this one is also good: http://lesswrong.com/lw/rj/surface_analogies_and_deep_causes....

[0] - it's also a good trick I picked up while hanging on LW; if you don't know why something happens, label it as unknown explicitly. Say "this process is driven by magic", or "caused by Divine Intervention" instead of trying to invent equivalently-informative but sciency-sounding labels like "emergent behaviour" or "spontanous self-organization". This way you'll never forget that your theory still has holes that need to be filled in, and you won't accidentally confuse yourself (or others).

You'd think it might work better if they had a bit of wire attached as an arial. Then they could use lower frequencies, get more range and so on, perhaps? Having mucked about connecting bits of wire to oscilloscopes it seems most of the signal that you could use to power something is 50/60Hz picked up from the mains. It would still cost cents given a bit of wire is not terribly expensive.

Its far-field, so it couples completely differently than NFC tags. Also, NFC communicates by reflecting signals, this device communications by active transmission (according to the article, I haven't read their paper yet).

What's new? Its tiny, and I'm very curious how they got that an oscillator to work at extremely low powers. But, it is really just an extension of RFID/NFC/IoT miniaturization work. But then again, just about everything starts out that way.

Maybe you can describe this a bit more. The Great Seal Bug was used for wireless audio recording in the 40's, but this operated by using a microphone/cavity to modulate the load of an antenna; thereby embedding audio in the reflected fields.

Bell did a similar demonstration in the 1890s using a mirror to embed reflections in ambient light and demonstrated wireless audio transmission over 200 meters.

However, I'm not sure why audio would cause a frequency shift in a radio circuit. Perhaps you are meaning the antenna will be perturbed and that could possibly be recovered? I'd be interested to know.

In the times that I was still building radio transmitters (for a very illicit living, selling them to pirate radio stations in Amsterdam) I had to hot-melt each and every long wire and coil in place so that it wouldn't vibrate.

The oscillator circuitry of a transmitter is (even when crystal controlled) sensitive to mechanical perturbation, which typically leads to spurious AM and FM modulation of the outgoing signal. To demonstrate the effect I once held a half our session on air with a guy on the other side of the city by just talking to the circuit board.

In a PLL or crystal controlled transmitter modulating the carrier in such a coarse way is much harder. Typically the modulation is done using a capacitive diode (a varicap) which is a diode whose capacitance changes with the reverse voltage. Because this voltage has to be applied to the diode somehow (in the days before SMD) this meant that that wire was again susceptible to microphony because air pressure on the wire changed it's location relative to the ground plane and that caused a measurable frequency shift. Not nearly as big a shift as in the older stuff but it was definitely a factor.

Wifi radios as much more robust than the stuff that I built. But I suspect that given a sensitive enough detector a residual audio component might be extracted from an otherwise non-audio signal by direct interaction between the sound waves and the transmitter hardware.

In a nutshell, it is very much harder to make something that does not exhibit microphony than to make something that does. You'd have to take that into account from the beginning of the design.