s2e18: Sequencing; Spectrum
0.0 Sitrep
21:31 Central Time, August 13, 2015
16-Asilomar osDarwin 0z4:xnu/RELEASE_ARM dm: 3-Azimuth
... Instantiation successful.
> look
You are sitting at a dining table in the farmhouse. There are two computers in front of you.
> look at computers
To the left, there is an black Gateway laptop. To the right, there is a silver Apple MacBook Pro.
> examine black Gateway laptop
The black Gateway laptop looks like it is running Windows 7. It is covered in stickers that tell you that it has an Intel processor inside. There appears to be a lot of malware on it, and it looks like someone has installed a remote login program typically used by scammers pretending to be Microsoft support.
> sigh
I know, right?
1.0 Sequencing
Apparently we've only just got around to sequencing the octopus genome and there's a whole bunch of super interesting stuff in there[0, 1]. This shouldn't really be a surprise because observation and experiments have shown that octopuses are a) very smart, b) very flexible, c) really alien looking because have you seen what they do with those suckers and those chromatophores. And that's before we even get started with the squid! So anyway, we haven't even *finished* sequencing *a* octopus genome yet and we've already discovered what we think are thousands of new genes[2]. If you're like me, the last time you thought really hard about genes was probably when you were around 14-18 years old and learning science or biology at school, so it's refreshing to know that *we still aren't really sure what genes are* because although it's true that they're regions of DNA that code for, well, *things happening*, they don't even have to be regions of DNA that are adjacent to each other because although one bit describes, well, the gene-y bit, another bit of DNA controls how it's regulated[2]!
Anyway, that's a digression. Octopuses. We don't even know what we don't know. We haven't even sequenced a whole octopus genome yet - they come in at around 2.7 million bases long, which is near enough to the roughly 3 million bases in the human genome, and we're only done sequencing about 83% of those bases.
So if you *want* to introduce some germ-line therapy to introduce chromatophores into your skin or to go even a bit more crazy and go with a radial symmetry body plan, then you're going to need to chuck a lot more money at the marine genetic biologists who're busy figuring out how octopuses work.
[0] The octopus genome and the evolution of cephalopod neural and morphological novelties : Nature
[1] Octopus’ sophistication driven by hundreds of previously unknown genes | Ars Technica
[2] Gene - Wikipedia, the free encyclopedia
2.0 Spectrum
Every so often I go spelunking in Spectrum, the magazine of the institute of electrical and electronics engineers, to see what's up in the, uh, world of electrical and electronic engineering. Here's the stuff that caught my attention and will probably bubble up somewhere else through my head in a few weeks time:
- Here's a good writeup of the 2015 Blackhat presentation that explains how Charlie Miller and Chris Valasek remotely controlled that jeep through the internet: Jeep Hacking 101
- Our children (well, to be perfectly honest, the study uses Japanese children so we need to replicate, right?) are being mean to robots, so robots are learning how to run away from children: Children Beating Up Robot Inspires New Escape Maneuver System
- I think this means that if you buy two off-the-shelf FPGA transceivers - or anything else that can function as a software-defined radio, right? - you can get either a) Superman x-ray vision wherever there's good wifi coverage or b) vision like Robocop used in the first movie to find the bad guys through the wall, but either way, we now know the maths to use the passive radiation from ambient wifi (2.4 and 5GHz bands) and GSM and LTE (no idea which frequency bands) to pick out moving subjects through 25 centimeters worth of masonry. So expect to see this as an app on a phone within about 18 months: See Through Walls by the Glow of Your Wi-Fi
- DARPA, the Defense Advanced Research Projects Agency that amongst other things was responsible for the system of tubes through which you're receiving this newsletter, is like a bratty kid who writes unreasonable descriptions of things in its Christmas list. It now wants a spaceplane that can fly 10 times in 10 days, fly faster than Mach 10, carry about 2,200 kilograms and costs less than $5 million per flight. For reference, current launches cost about $50 million per flight. Billionaire genius playboy philanthropist Tony Stark^W^WElon Musk *isn't* involved in this particular competition, but Boeing (with Jeff Bezos' Blue Origin), Northrop Grumman (with Scaled Composites and Virgin Galactic) and Masten Space Science Systems (with XCOR) are. Oh, right, and they want to fly the prototype in about 3 years time. That's about when the iPhone 8 will come out. DARPA Funds Stage 2 of XS-1 Spaceplane Design Competition
- Terahertz rays are cool and we might have figured out how to make them, also we get to call them T-rays. Simple Device Could Convert DC Electric Field To Terahertz Radiation
- Supercomputers have stopped getting fast as fast as they used to be getting fast, so now instead of getting an exascale computer around 2020, we might have to wait until 2023. Some people think this is because of Moore's law, some other people think this is because no-one wants to pay for it. China still has the fastest supercomputer, also, they're mining a lot of bitcoin but those two things are totally unrelated. Why Aren't Supercomputers Getting Faster Like They Used To?
- This one has nothing to do with IEEE Spectrum but the nVidia DIGITS dev box is in my mind what a Silicon Graphics Indy looked to me as a 14 year old: kick-ass workstations with unreasonable amounts of power are now not used just to render graphics, they're used to simulate or run giant neural networks to increase the number of slugs we can find in photos uploaded to the internet. The specs on the box look completely crazy and I just wish it had similarly awesome industrial design and didn't just look like a high-end gaming PC or a first-generation glowing green Xbox. Because if you're going to have that much computing power, you need to make sure that it looks like a Gibson pizza box or whatever. NVIDIA® DIGITS™ DevBox
--
OK, that's it for today. I'm still glaring at that Windows 7 laptop. Odds on having imposed a new Mac on the father-in-law by the end of the week are unsurprisingly high.
Best,
Dan