After All Those Silver Linings: A Cloud

The electronics retailier Best Buy, as you may have read, has begun phasing out its CD and DVD sales. It was inevitable even in the conventional sense of technology progression. The compact disc is a nearly 30-year-old format, which worked fine for music playlists of the usual album length (back when measuring things that way mattered more) and you could back up a few dozen photos once they made 'em writable, but the RW was on the wall from the beginning. They just didn't have enough capacity for what people were already using magnetic media for. Not that magnetic tape wasn't shit. The sleek, discreet (and discrete!), silver-lined disks* only have to worry about getting damaged from the rigors of storage and abuse, not so much in the course of using them. But it took a while for video signal to make the awkward transition off tape. Anyone remember those early video disks? The size of a platter, and you had to flip them over halfway through. Got a friend from college who's probably still geeking out to The Empire Strikes Back on that thing.

If you look back on bubble-vintage tech literature (e.g.), they were very excited about the upcoming 4.7 GB capacity in optical media (which would be your standard DVD), and the promises to extend that to 70 GB this decade (which is approximately what you get out of your standard one-sided blu-ray disk). Big fucking deal. I fell into an interesting conversation yesterday about media formats and obsolescene. (These were old photographic film industry people, with a perspective on their former company's innovations and terminal flaws.) All this investment in optical disks—and the read/write technology is pretty impressive from a materials science geek's perspective—and pretty much no one was thinking about how boring old magnetic media would come along and totally kick their ass.

In terms of data density—how many ones or zeroes you can fit per unit area—a blu-ray gets about 12 GB per square inch. From your hard disk, you're getting 200. Today, a hundred-dollar external hard disk can store all of your music, and a dozen or two high definition movies in a device the size of a deck of cards. A DVR can store the week's worth of television programming that most people care about. It is expected that magnetic data storage will top out at a TB per square inch(!).

The promise of optical computing is speed and maybe bandwidth, not really the inherent size. Optical resolution is usually diffraction limited, meaning, in storage media, that the smallest spot is depends on wavelength of the reading laser (and aperture of the focusing optics). This limits how tiny a bit you can read. I had mistakenly thought that the innovation of DVD was merely doubling up the layers to increase the density, but they also adopted a new laser diode. The old CDs use a red laser with 705 nm wavelength, and the move to a DVD was to a 605 nm laser which shrunk the smallest bits that could be addressed to about 4 um. (Obligatory points of reference: there are a thousand nm in a micrometer; a typical atom is about 1/10 nm; a typical cell is about 10 um, or 10000 nm; record grooves are about 50 um. I am not going to compare to human hairs.) Commercial blu-ray technologies has been enabled in part by the invention of reliable 405 nm laser diodes. The bumps on the blu-ray are down to about 150 nm, and you get more storage off of one layer than you do on the doubled-up DVDs. (The blu-ray disks also managed to get content closer to the surface to increase the aperture.) You can sort of see where it goes, as photolithography had the same issues in making objects of sufficiently small size, but it probably needs better diodes again to happen: it's hard to imagine adopting excimer lasers, mercury lamps, or immersion techniques for my home video. (For optical storage the obvious-ish way to go subwavelength is to go near-field, and who knows, given the tolerances of magnetic disks, it might be made to work. Holographic storage has always been up and coming, but isn't enough of a breakthrough to even propose better than magnetic, even though it's extremely cool. Holography is diffraction, and the data limit, Wikipedia says anyway, is about 1 bit per cubic wavelength.)

The fact that you can write, in controlled circumstances, 150 nm features in your cheap home drive boggles my mind. Just like records, they used to make and distribute music cds from master templates. With a stash of proprietary plates, injection mold the polycarbonate in custom machines, microscopic grooves and all, stamping them out like license plates, and then metallizing them, before packing and shipping out pallets of Better than Ezra's testament to the musical suckiness of the late 90s. (This is essentially what my dad has spent a lifetime doing in a plant that makes decorative plastic containers. Not at quite the same level of precision.) I don't really know if its cheaper to stamp out higher density optical disks than it is to thermally write them, and I don't know how they mass produce the ones you purchase from Best Buy, but you definitely can do it that way. People have imagined this kind of pattern transfer as the cheap way out of photolithography (which is also subject to diffraction limits) for electronics production for many years now (although of course you still need to make a master somehow), and claim resolutions below 10 nm. It is, in fact, on the official roadmap. Even now, a couple hundred nanometers in a production setting is pretty feasible.

A magnetic domain in a ferromagnetic solid can be as small as 20 nm, which, keen observers will note is somewhat smaller than the wavelength of your normal laser-emitted photon. There's a lot of research that has gone into into manipulating magnetic domain size, and I fail you by having kept up with none of it over the years, and I compound that failure by having over the course of the day decided that it was more than I cared to study for a blog post. So how it ends up making sense in polycrystalline materials, thin films, etc., I don't really know, but I suspect is impressive. (I will mention that I have written proposals saying that making magnetic crystals several nanometers across might necessarily be single domain and make for some interesting physical comparisons, but those were justly unrewarded.) That you can move a read head with sufficient precision with something as primitive as a voice coil is fucking insane; the engineering of these devices is really fabulous when you get down to it. On the other hand, it's a little disappointing that the basic tech is as old as my parents, and we're merely spinning ever-smaller plates to get the information out. (Digitization was a big conceptual advance of course, but in some ways, the fact that analog devices work predictably impresses me more. Making record grooves and encoding data without knowing what the code is, that's pretty radical. I mean, I know in my mind that it's not any weirder than a guitar pickup, but still...)

They still make magnetic tape, and the market is data less-than-urgent data storage, such as nightly backups or the kind of stuff you back up or hold in a box for a half-century. The company I spoke to yesterday make optical tape, sort of a reel-to-reel DVD, which offers amazing storage ona bytes-per-dollar basis, and is intended for archival information. It doesn't sit completely well with me. It's probably better than magnetic tape, which is subject to activated bit-flipping with warmth and time, but on the other hand, I think of classic celluloid reels rotting in their cans, or of how Stevie B got deep and garbled at either end of the casette back in 1988 after a couple months of play.** Maybe lets etch it all in stone.

The musical revolutions of the living generations were pioneered by storage technology (and other technology miniaturization). We can trace our experience of music from records to tape to optical media to high denisty magnetic and dynamic memory. Records brought the concert to our living room, and Walkmen to our gym. God knows, enough low-value digital copy has been expended speculating on the low value of digital copying. What's it like when it's pure information, etceterblah. Making information storage and transfer cheap made content even cheaper, and for consumers, it's been pretty great. (Although I have read that the digitization rate of CDs is the bare minimum to fool the ear, just over 44 kHz. That was pretty goddamn lazy, and if I can't hear the difference, diligent audiophiles claim they can. One reason to cling to analog.)

And it's a funny thing. Consumers love storage capacity, so manufacturers like to sell it to them (free markets as actually working). On the other hand, content providers fucking hate the capacity to store and share fungible bits (free markets as they prefer to fail to work). Sometimes the conflict falls within the same entity. Apple has recently made the future clear, let's make the content user-independent. I say, fuck the cloud.

The whole move is to subscription-based media. That's what net neutrality is all about, and while I've never overestimated the democratizing ability of the internet, fair access to greater information has never been a bad thing. It's not so terrible in all the aspects--it would be nice to listen to your music library anywhere, and yeah, I've grown content accessing my email this way, and maybe I'd rather pay a little for the handful of shows I watch a week than let the networks rent my eyeballs to advertisers. But music hits me harder. Maybe it was all the arguments I remember from the Napster days. It's good to pay artists (although bundling album sales must have improve the sales of their deeper tracks), but the RIAA and the broadcasters are the essence of evil. They would like the control over content that used to bring predictable record sales. There's precious few movies I'd rather keep copies of, and normally renting a viewing isn't a biggie. Music though? It's good to borrow before owning, but it's too integral and too much a part of my life to be constantly on lease. I romance a body of music, I fall in love with it, marry it, and then slowly grow apart. Music is a commitment, for which the law only needs to be involved at critical stages. (Books are somewhere in between. I'd prefer to own them as a permanent record, for all sorts of reasons ranging from physical pleasure to the throwback notions of knowledge guarded.) Music needs to stay at home.

I don't know why DVR has to be a subscription service either,and I don't like it. It's a local disk, and I used to be able to program my VCR just fine. It pisses me off that an unnecessary transmission service gets disguised as content. It reminds me of when AT&T under their monopoly used to rent us telephones . When it gets down to it, these fuckers have a fine history of treating both ends of the transaction like shit, making a fortune off of limiting the flow. I have no burning desire to get yet another middleman skimming money from the daily interaction of me and my life. Or, for that matter, to obviate my stored media in a controlled fashion so that I can pay for a lifetime of upgrades and format changes. They've been trying to make music subscription based for years now, and I don't want to restore the greedy bastards to power.


* One year, a friend of mine gave me the gold CD edition of Rush's 2112 for Christmas, the idea being that the oxidative stability of gold will keep the integrity of the bits indefinitely. That's not quite one of the 15 personally influential albums that I would prefer not be admitted, but it's pretty close.

** There's the second-most embarrassing, and probably the first-most dorky, one of those, were I willing to be honest. For my 15th birthday, Mom and Dad got me a home weightlifting set that I set up in the basement, and that's what I worked out to for a few months. "Spring Love" spoke to me, man. Shut up.

[Also, the loudmouth thing again. Edits no doubt to follow.]