Nah, there's no exemption for roman alphabet spelling, they get that wrong too. Not long ago there was a US Secretary of State with a French name "Boucher" didn't get pronounced "booshay", but "bowtshur". This guy must have had the whole of the corps diplomatique giggling inside.

So when Johann Cmythe is wanted, the various combinations of Jan, Jon, John, Joanne, Jane, Joan, Smythe, Schmidt, Shmitt, and Smith will all be queued up at the gate, waiting their turn for a cavity search. That'll work!

Yeah, but the formula used to compress pi down to one bit sucks at compressing e.

So we're looking for the C. elegans of software?

Well, Linus uses a broader definition of kernel than is customary, referring to a monolithic (macro) kernel. If you really seek elegance in OS code, you start by looking at microkernels, stripped of all the device-dependent clutter. Harmony, for instance was a mere 20 kbyte kernel, even on 68K architectures. http://books.google.com/books?id=xvOpC0_r14wC&pg=PA100 http://www.researchgate.net/publication/234826460_Harmony_as_an_object-oriented_operating_system This led to the even more succinct MQX, which is embedded in quajjillions of devices.http://en.wikipedia.org/wiki/MQX

"I don't know, Yogi. Ranger Smith isn't gonna like this..."

Placebos are pretty well understood. The BS enters when you pretend that one sugar pill is vastly more valuable than another and charge accordingly. Of course big pharma does the same with real drugs (look into "evergreening" sometime), but for them, it's more of a sideline. The basic idea is still to find treatments that actually do work better (and safer) than currently available options.

There's a reason there's an entire field called evidence-based medicine [wikipedia.org], which from its very name makes it distinct from just plain-old normal "medicine."

If you prefer, we could just distinguish "real medicine" from "pretend medicine"... Of course individual practioners use a mix of the two, particularly when there's no real treatment for a diagnosed condition that is either untreatable or harmless. That in no way makes pretending a full-time substitute for evidence.

So, less than five cents per ticket on a 200 pax aircraft could pay for sending basic telemetry: lat, lon, airspeed, elevation every four minutes. That doesn't sound like much, until you think that there are 3 billion tickets a year. Still, $150 million/year to avoid having to run searches for missing aircraft might be a good deal, especially in cases where there are survivors to be rescued.

Looking at http://pubs.acs.org/doi/abs/10... and at http://nanoresearchul.org/ it seems that Ryan et al. have a new way of making stable anodes, losing 10% of the new charge rate per 1000 discharge cycles. A car that still has 80% power after 2000 commutes starts to look credible.

You've forgotten that the Hz represents cycles/s, not just 1/s. Hence their so-called "spectral efficiency" is 5.7 bits/cycle. The problem of course is that the article does not address the SNR nor the BER that it took them to get that 5.7 result. If you need cryo-tech photodetectors and massive FEC to get that result, it's less impressive than doing it with the existing kit and minor data redundancy.

An there we were thinking that Target had all those real-time customer-linked analytics watching our eyeballs as we strolled the shelves and trotting out the offers we couldn't refuse. Now it turns out they're just as incompetent as the next bunch of monkeys. We can breathe again, at least for a little while...