Mad Dog 21/21: Bone Appetite

Computing seems to have become, coincidentally, an aspect of cannibalism, just as cannibalism has once again become a prominent aspect of computing. In Germany, a computer technician is on trial for eating a part of an engineer he met via the Internet, killing the engineer, and subsequently eating a lot more engineer steak. In Northern California, Japan, Taiwan, and Korea, among other places, cannibals in the storage, electronics, and computing industries armed with flash memory are killing off and consuming the markets for a lengthening list of gadgets.

The trial of Armin Meiwes, the computer tech cannibal, is unprecedented. Germany has no law against cannibalism. The prosecution is basing its case on various theories. And it does have what may be a fallback basis for keeping Meiwes behind bars: the crime of assisting another in a suicide. This last legal approach is necessary in the Meiwes affair because Bernd-Jurgen Brandes, who came to and was dinner, participated voluntarily.

The court knows this because Meiwes videotaped the events on which the trial is based. The tapes show, among other things, Meiwes and Brandes sitting down to dine on the part of Brandes that had been severed. It was some ten hours later that Meiwes allegedly killed Brandes. We say allegedly because, notwithstanding press reports that seem to show Meiwes stabbing Brandes to death, Brandes might have died primarily as a result of chopping for the first dinner, in which case it is hard to say how the court will rule. However the trial proceeds, it does not seem too likely that Meiwes will be back on the streets and the Internet soon, or ever, trolling for another freezer load of flesh. Whether or not Meiwes had eaten his fill by the time he finished off the forty-odd pounds of Brandes he butchered, froze and subsequently ate, we somehow doubt the cannibal will get another bit of the cheery engineer.

This situation stands in stark contrast to the situation in computing, where cannibalism is a way of life and death. Standardized hardware and software are busy eating the markets established by proprietary alternatives. Flat displays are gobbling up the desktop space that is still held by CRTs. X86 processors are eating proprietary and RISC architectures. But the most dramatic cannibalism of the moment is taking place in storage, where flash memory is directly and indirectly enabling a stunning variety of new products to rapidly chew to bits older alternatives.

Armin Miewes: Volunteer victim is food for thought

Many of the older products being slaughtered and eaten are aptly named consumer devices, such as cell phones and cameras. But the stunning impact of progress in flash memory, which has proven to be as vast in its effects as the transistor and integrated circuit, has yet to run into boundaries. It’s a flash flood as solid-state nonvolatile memory washes away established market turf, soaks up product development efforts, and seeps into every corner of the electronic universe.

The industry that has been most visibly affected by flash memory is photography, where, as digital memory has replaced film, camera makers are only the first companies directly affected. Film processing firms are forced to become image-processing firms, moving from chemical to digital printers; they also must compete with inexpensive photo printers attached to PCs.

The connection between computers and photography has become a significant factor in the sale of single-user systems. Hewlett-Packard, the market leader in small and midrange printers, now not only offers photo printers but has a line of digital cameras and provides PCs configured with digital shutterbugs in mind. Gateway also offers its own brand of digital cameras. Dell sells its own printers alongside those from other vendors and peddles cameras from a number of makers on its web site. Windows has a media edition, aimed at hobbyists who want to fiddle around with pictures (as well as videos and sound tracks).

The sales volume of digital cameras has swamped that of film cameras, forcing the film segment of the business to retreat to high and low ground. Professional photographers continue to buy film cameras at the high end, while single-use cameras at the low end have yet to go digital. While established camera companies still have the lead in brand recognition, optics and well-established distributions channels, new players (in addition to the PC vendors) have emerged to challenge them. The situation looks a lot like the PC business during the 1980s–ripe for a shakeout.

From SLRs (this one from Canon) down to miniatures (from Minolta), digital cameras are eating the film camera market

The stronger camera companies are already eating the weaker ones much the way members of the Donner party consumed their fallen fellow pioneers. On their way from Illinois to California during the winter of 1846 and 1847, the Donner party was trapped by snow in the Sierra Nevadas, so while more than 80 people started the trip, only seven survived. Some of those lost were eaten by others members of the expedition. The details of this misadventure are in dispute, but there’s no question those who became just meat didn’t sign up for extreme KP on the Internet or its predecessor, the telegraph, which at the time was just beginning to catch on. Incidentally, none of the Donner survivors were convicted of cannibalism.

By contrast, 28 years later Alfred Packer, who similarly ran out of provisions while in the Rockies, was convicted of murder for killing and eating fellow travelers, although at times he claimed he killed only in self-defense. After some amazing judicial acrobatics, Packer went to prison, but was later paroled. He died in Deer Creek, Colorado, by then, reputedly, a vegetarian. There is an Alfred E. Packer Memorial Grill at the University of Colorado in Boulder. Whether this is recognition of Packer’s innocence or just grisly college humor is an open question.

The picture in the camera business is and equally stark and, unlike that of Packer, quite clear. The market has reached an inflection point. Quality digital cameras with the apparent resolution of 35 mm film cameras, are still costly, although they have fallen to available for well less than $1,500, although the very finest equipment can cost five or six times as much. The key breakthrough seems to be the Foveon sensor, which stacks photocells three deep, gathering three times as much visual data per unit of sensor area as alternative technologies. (There is academic data showing it takes a lot more than 10 or 12 megapixels to match the resolution of 35 mm film, particularly the finer grades like Kodachrome, but as a practical matter there are almost no situations in which the experience of photographers matches the theoretical analysis.)

For ordinary consumers, high end means a camera that captures at least 5 megapixels per shot, which these days means a retail cost of less than $400. This price is a bit of an illusion, however, because these cameras typically come with starter flash memory cards holding 32 MB, while cameras in the 3-megapixel, $300, class may come with flash that holds only 8 MB or 16 MB. Consumers discover have to buy much more capacious memory cards right away.

Alleged cannibal Alfred Packer: Maybe he did, maybe he didn’t

The need for more picture storage capacity does not present a technical problem. Depending on the type of storage technology the camera uses, flash cards can store 512 MB, 1 GB, 2 GB or 4 GB. (There are some older flash technologies that max out at 128 MB, but they are supported only in backwards-compatible mode on current cameras.) A physically small flash card (some the size of a postage stamp) can hold a lot of photographs. Typically, the fine mode JPG recording compression reduces a typical 3-megapixel image to no more than 1.5 MB and still yields a tolerable 8×10 print, making 64 MB of flash memory the digital equivalent of a 36-shot roll of film. As the typical camera moves up to 6-megapixel resolution, storage requirements will double, but flash cards with 256 MB capacity, which can store more than 80 6-megapixel JPG files plus other associated data (timestamps, for instance) are already priced in the $75 range and falling. Currently, 35mm film from name brand manufacturers costs $3 to $8 per roll (with the higher priced types offering outstanding technical attributes). It doesn’t take all that many photos before a reusable flash card is cheaper than film, and that’s without the cost or processing, which puts film at an even greater economic disadvantage. The lower the resolution of the camera, the faster a flash card gains an economic advantage over film because the card holds more pictures.

Photo costs aside, there’s no contest between digital and film when photos have to get into databases. Insurance adjusters, doctors, dentists, cops and real estate agents have long since gone digital. Photojournalists with a fortune invested in Leicas are laggards, but those who send in their work via satphones or email are not. Tradition maintains its grip on artistic types who say, with some justification, that digital photography just ain’t film yet . . . and may never be. Ordinary consumers know film still has some benefits that are hard to find in affordable digital cameras, but they are willing to accept the limitations of digital cameras in return for convenience. Nobody is tossing away his or her film cameras, however, so it’s a slow fade to black.

One factor militating against digital photography is its complexity. Digital cameras have an intimidating number of features and some confusing incompatibilities. In particular, a camera may use any of several types of flash cards, and each type of card has a unique physical size, interface, and range of capacities.

Despite the differences among various packaged formats, flash card image storage all depends on the FAT file system used in Windows (and DOS before that). This turns out to be good for Microsoft, which is starting to charge patent royalties of $0.25 per card, capped at $250,000. To link the file system to various devices in addition to the camera, most flash cards also have a controller chip. The upshot is that smart flash cards can be plugged into a computer’s USB port while in a camera or via a very cheap and simple reader. To the computer, they look just like a removable disk. PDAs also use flash cards and can be used to manipulate photos. The type of flash card used by all PDAs except Sony models is called Secure Digital; Sony has stuck with the Memory Stick format it developed.

Flash cards are not only a negative for film; they’re quickly replacing other storage media, too

While cameras are big business for flash makers, industry watchers say the largest market for flash memory is the mobile phone business. Mobile phones don’t use as much memory per unit (yet) but they sell like hotcakes. On the other hand, digital shutterbugs end up buying extra flash cards, so the camera segment of the market could at some point overtake phones. Solid-state portable storage devices that plug into USB ports are the third largest consumer market for flash memory, but unit volume in that segment is comparatively small.

New markets are likely to emerge. Very soon, quite a few PCs (laptops and desktops) will routinely include flash card reader slots. Some PCs and printers already do, putting flash memory in direct competition with other removable media. Even now, flash devices that plug into USB connectors are quite popular. But visible applications of flash memory are only a part of the growing market. The technology is built into all kinds of electronic gadgets from network routers to servers, where it stores firmware and other information.

It all adds up to a market that is currently about half the size of that for DRAM, $7.8 billion compared to $15.3 billion in 2003, according to the Semiconductor Industry Association. Flash looks to be growing faster, but it’s hard say it will become as big a business. The PC and server markets, which consume the lion’s share of DRAM, have been in a slump that now looks to be ending, so DRAM sales volume could surge upwards even as the flash market continues to leap ahead.

Semiconductor pundits argue about the limits of the flash market. One point of contention involves flash versus disk. To most observers it might seem as if flash could never replace hard drives. It’s much more expensive per unit of storage. Current flash wears out faster than hard drives do, after a couple hundred thousand cycles rather than millions. But flash fans argue that there is no telling what markets the technology might eat next. Flash is faster than a disk and uses a lot less power. It’s also more tolerant of physical shock and temperature change. The step from today’s largest high volume flash modules, which have 4 GB of capacity, to devices that can store 40 GB, a capacity that suits many laptop users, might not be far in the future.

In fact, it could all happen in a flash.