One Decade Out

July 12, 1999

I like to keep an eye on what's in the technology pipe. For some, this means reading the Financial Times. For me, this means reading Scientific American. In essence, you can have a damn good guess at what will be driving our technology a decade hence by looking at what the boffins are doing now. A good peer-reviewed science journal is still by far the best way of doing this. I love SciAm, mostly because of their careful in-depth articles, where too many other mags just gloss the 'gee-wiz' surface. SciAm gives you the principles behind it all. And they have this peculiar sense of humor that comes from knowing in their bones that everything they write now will be laughed at 50 years hence. They even have a special column to chuckle at old issues.

Predicting the future can be made a science (but, probably not a technology :-) if you look for the principles. Here's Orinoco's Principles of Technology Transfer:

1. Scientists and Technologists are people too.
2. The things scientists are really excited about are 25 years (or more) from widespread application. The things they're regularly working on are 10 years from application. The things they're bored with are just crossing over.
3. The application that becomes most important is never the one that drove development in the first place.
4. Developments are usually replicated in parallel by many groups if the knowlege isn't limited.
5. The greatest ideas are the most incomprehensible at the time.
6. A Technology only really 'exists' when everyone can access it. Until then, it's not technology yet. It's still science.
7. Technology changes things.

Principle 2 most concerns us here. In the short term, it means that the next great device is sitting in a disused corner of some lab. When you ask the whitecoat about it, he'll probably say something like "Oh, that old thing..." Fortunately, principle 4 kicks in, and someone else, somewhere, will slap one together and say "Hey. This is pretty cool..."

One Recent example I can think of is a mob called Radiata. One of the guys there worked for the CSIRO in their telecoms research centre, playing around with low power 5Ghz microwave trancievers, and hacked up a little prototype 30Mbit/s link across the lab bench using handy parts and a new encoding scheme called OFDM. (Orthgonal Frequency Division Multiplexing. Essentially a wireless-tuned form of DMT (Discrete Multi-Tone. Also known as ADSL)) Thankfully someone saw the potential, and they're working on commercial prototypes now. The IEEE 802.11 draft standard for the 5Ghz band is based on their work. Their web page is pretty sparse, though.

One of the weird things scientists are routinely doing these days are Bose-Einsteinien Condensates. BECs are bloody strange. They are macroscopic quantum objects, millimetre-sized blobs composed of the overlapping wave functions of a hundred thousand atoms. In one sense, they're like a super-atom. In other ways, they're like black holes. (The quantum states of everything 'inside' have merged, and the whole ensemble only has a very few quantum properties) Only 3 kilometres from where I sit, physicists at the University of Queensland build their atom traps with lasers removed from CD players, RF coils and vacuum chambers. It's cheap to do.

BEC's are almost incomprehensible. Principle 5. Then you rattle off a couple of possible 'far-out' applications, and people's eyes light up. "Replicators" is the main one. You can use BEC's as a 'matter laser' to 'paint' matter in incredible ways, like use interference patterns from intersecting BEC's to create a precise grid of atoms. BEC's might be the tool we use to build the first nano-assemblers, kicking off the much-desired Nanotech revolution. Or the available control over quantum states might make BEC's the starting point for quantum computing. They're even doing a primitive form of quantum alchemy with spin states. Of course, principle 3 suggests that the eventual killer app will be completely outside the box, something no-one has imagined yet.

Holographic storage is slowly coming. It's not quite a technology yet, by right of principle 6. There are a few niche commercial systems, but no general solutions. (That I know of.) Holo is another example of a lab setup which made it's slow way out into the world. You can always tell first-generation technologies. They look and act just like the lab prototype, except the wires aren't showing quite as much. They're a good example of principle 7 too. Holo storage isn't just another point along the storage curve. Sure, they're bigger and faster (terabytes of storage with nanosecond access times and gigabit transfer speeds) but they have one extra ability which changes everything:

Holographic storage works like this: A hologram looks three dimensional because your eyes see different images, because your eyes are viewing the holo at different angles. The holo has 'recorded' a slightly different view for each angle. Some cute holos even change the image over this range, so a frowning Franklin morphs into a smiling one as you jiggle your head. Holographic storage uses this angle-dependant view ability to store a page of data in each angle. Generally, you use a reference beam at a particular angle to 'select' a page of data for reading or writing. You write a page by turning the reference beam and the imaging beam on. When you present just the reference beam, the crystal reconstructs the page beam. Cool.

The little-known ability of Holo storage is that if you present a page beam, the crystal will reconstruct the reference beam. If you present the image of a page you're looking for, you'll find out where (if) the image is stored. Or, the closest match. Or matches. In the same time it takes to read a page. A database lookup, in other words. Without an index, in constant time, regardless of the amount of stored data. Wow. That will change a lot of search algorithms. This is why the first niche uses are fingerprint recognition systems. It's this 'seeking/matching' ability of the holo which makes it economical. I'm still waiting for my Jazz Holo drive though, with data crystals for a buck each. Yeah!

As for computers, what I see more often every day are groups playing with superclusters. Beowulf machines. A hundred off-the-shelf PC's on a heavy network backbone, pretending to be one big parallel computer. Of course, researchers can't get enough computing power, so they turn to duct-tape and spit to get the most bang for their buck. And in the process, they're learning how to make these messy machines tick. My favorite is the "Stone SouperComputer", a Borg-like ensemble of old PC's which keeps assimilating anything that gets near it. Imagine that it's the seed of a new internet-like computing space, (the web is a document space) which will gradually absorb every other PC and supercomputer until it finally becomes Cyberspace, the real thing. Hey, who knows. It just might.