Just for the heck of it, Dave Babcock is writing an IBM 1620 emulator for his PalmPilot. The 1620 was an $80K machine back in the 1960s. It had to load arithmetic tables into memory before doing calculations, so Babcock has simulated that. The PalmPilot is still a lot faster, though. Also a lot smaller and cheaper. [Chris Nolan, SJM, 04May98, 1E.]

Molecules are capable of much greater diversity and complexity than we have yet tapped with wires and silicon surfaces. Genetic engines can build tiny electrical and mechanical systems from components with rich interaction modes. (Those interactions are problematic for us now, but will ultimately be useful.) Indeed, biological life itself can be so engineered, as far as we know. (No word yet on the creation of souls, although cloned sheep and test-tube babies show difficulties to be unlikely.)

Biotechnology companies are on the verge of breakthrough technologies. It cost $2.5M to sequence a gene in 1974; now the cost is $150. Monsanto used to try about 60 new potato genes per year in the 1980s; now it can try 10K/year. Gene data libraries are doubling every 12-24 months, in the same range as Moore's Law for semiconductors. This knowledge, plus corporate acquisitions and mergers, may soon produce biotech companies larger than Microsoft and Intel. Leaders include Monsanto, DuPont, and Novartis (in Switzerland, formerly Sandoz and Ciba Geigy). [NYT. SJM, 02May98, 1C.]

The Drake Equation is a way of estimating how many intelligent civilizations our galaxy may currently host, to within a few orders of magnitude. It includes estimates such as ten stars forming per year, half of all stars having planetary systems, a couple of those planets being suitable for life, etc. If life commonly evolves on such planets and if 1% develop to intelligence, we can estimate that the number of technically advanced civilizations in the galaxy is about 0.1 times the average lifetime of such a civilization. Thus if an average civilization lasts a million years, there could be 100K such civilizations in our galaxy.

Now consider the pace of evolution. Human intelligence depends on a complex brain, which has quadrupled in size (although probably not in connection complexity) over the past 3M years. Call that a doubling time of 1.5M years, which is astonishingly rapid development for such a structure. However, note that our microprocessors and other chips are doubling in transistor density (or "complexity") every 1.5 years, or a million times as fast. It's harder to measure our progress in algorithms and intelligent systems, but certainly their improvement is also spectacular. (Look at speech recognition, for instance. A PC can now respond to more commands than can a dog or chimp.) By the year 2000, Intel expects 100M-transistors chips able to execute 2B instructions per second. It seems likely that we will soon -- in 100 to 1,000 years? -- create artificial intelligences. A significant fraction of other intelligent species will do likewise, given enough time.

We can further suppose that artificial intelligences will help their biological creators survive, or will at least endure and spread beyond the limits of their creators. Hence civilizations that develop artificial intelligence are likely to be the longest-lasting and most wide-spread in the galaxy. They greatly increase the number of civilizations we could encounter. When we do meet our neighbors, it is likely that they will be very old civilizations, and most likely they will be non-biological. [Efram E. Goldstein , comp.ai, 30Apr98.] (Perhaps they will be biomechanical hybrids: cyborgs. Or we may find the distinction irrelevant, as molecular components replace our bulk technologies.)

-- Ken