Monday, May 19, 2008

Moore's Wall

Advances in computer performance, size, price, and capacity have continued to astound. What can we realistically expect in the next twenty years?

Moore's Law continues unabated, having passed several "fundamental limits" on size or performance improvements that briefly appeared to stop or at least slow the rate of improvement of computer technology. In fifteen years, we'll likely reap the benefits of yet another thousand-fold increase in price-performance. In thirty years, a million-fold improvement over today.

Computer memory (high-speed main memory, Flash memory, hard drives) have seen even greater advances than CPU technology, with yet still greater advances on the horizon, such as memristor technology that has achieved 100 gigabits per square centimeter in the lab. To date, our implementations remain two-dimensional, so I believe that many orders of magnitude of improvement can be achieved. However, I don't foresee thousand- or million-fold improvements over the next 15 or 30 years, simply because I don't see the demand for ever more storage continuing. Once you have a copy of every song, movie, and book ever recorded on your PDA, what next? Add the entire Wikipedia, and every thing you've ever seen, heard, or said--in your entire life. All of that is within reach, possibly within a decade.

We've already reached demand constraints on the advance of telecommunications technology. I projected continuing improvements in Optical Ethernet back just before the telecom and dot com collapse in 2001. While the technologies have reached no physical limits, the rate of improvement has slammed to a crawl. Improvements can and will continue, but only once a market large enough to pay for new infrastructure proves itself.

Back to the point of this ramble.

Much of our wealth and technology can be directly related not only to the astounding advances in computers, memories, and telecommunications, but to the rate of improvements in these technologies. We throw away phones, televisions, and computers because it's cheaper to buy the latest and greatest. We put up with incredibly inefficient software because next year's computer will make it bearable. We invest in Intel (etc.) because of the certainty of next year's advances and tomorrow's new markets.

But much of our society is built upon a foundation of continuing growth and will collapse when the limits of new markets and new technologies are reached. This includes fundamentally non-self-sustaining things like Social Security, the stock market, the concept of investing for the future itself, and retirement. Capital investment only works when there is an incentive of reaping more than you sow. When markets cease to grow, the economy will stagnate and collapse as a domino chain reaction of events undermines our way of life.

Remember that everyone not working (ie, retirees) is living on the current productivity of active workers. The only way to sustain a growing population of retirees is to maintain an equally growing population of workers (or in the short term, their growing productivity). This translates to a need for a growing population, or  a need to drastically reduce the number of retirees. Remember that mass starvation and pandemic diseases are nature's way of dealing with overpopulation.

I am writing a science fiction story exploring these problems, called "Moore's Wall".

I can see only one long-term solution: continued growth in technology, markets, and population. And the only sustainable population growth will be into space.

And lastly, I fear the unchecked growth of computer technology. See my post on the Technological Singularity.

No comments: