Polymathica Home

Polymathica

Refined.. Erudite.. Visionary

The Singularity is Not Near

If you kick around the intellectual community for any time at all, you will undoubtedly encounter the Transhumanists and Singularitiarians. On any list of suggested reading for the Polymath, Ray Kurzweil’s book, ‘The Singularity is Near’ should be prominent. That is because the book is influential, not because it is particularly accurate. Kurzweil accepts that technologies develop and mature along a path described by a logistic or S-curve. He then makes the claim that, in cases such as computing capacity, S-curves follow one another sequentially to form a grand geometric curve. (page 43) In this, he is mistaken. The same forces that cause an individual technology to follow an S-curve applies to the function, such as computing capacity per constant dollar, itself. Consequently, it will follow a super S-Curve. We believe that there is strong evidence that we currently are at the point of flexation in the growth of computing technologies.

Technologies follow logistic curves because at the beginning of their development, improvements are easy and generally have high returns on R&D dollars. This is Kurzweil’s ‘Law of Accelerating Returns.’ The growth is geometric because the high ROI’s attract more investments. However, there comes a point when the ‘easy pickings’ are over and it begins to become more difficult to improve upon the technologies. Then the investment in the technology begins to experience the more well known, ‘Law of Diminishing Returns.’

It seems that when looking at a function, such as computing, that experiences several technologies sequentially, that the Law of Accelerating Returns will eventually be replaced by a Law of Diminshing Returns. Kurzweil identifies five different technologyies so far in computing capacity. First, was electromechanical, then the relay, then the vacuum tube, then the transistor and lastly the integrated circuit.

When electromechanical computation began to reach its maturity, it was relatively easy to find a technology that had more potential in the relay. When the relay began to reach its maturity, it was a little more difficult and costly to find a new technology with greater potential. It was, however, found in the vacuum tube. When the vacuum tube began to reach its maturity it was even more difficult and costly to develop a new technology. It was the transistor.

The current technology, the silicon based integrated circuit, is approaching its limits. The number of components that can be put on a chip may continue to increase at least until 2019. However, some knowledgeable industry insiders are saying that the meaning of Moore’s Law, that computing capacity that could be fit on a chip would double every year, has already reached its end. The most signficant measure, the amount of computing power that can be purchased for a constant dollar amount, appears to be reaching its limit no later then by 2014. In other words, after 2014, even if you will be able to get a more powerful silicaon based CPU, it is going to cost you more, not less.

This is critical to the Singularity. At its core is the assumption of ever cheaper computing capacity. It now appears, unless a whole new technology emerges, that, instead of one thousand dollars of computer buying you a human brain equivalent by 2030, this price-performance measure will stall out soon and below the intelligence of a mouse. This is the difference between the Singularity and the Transformation. While robots and AI are likely to invade every aspect of our lives by 2030, they will remain comfortably less intelligent than us.

There is a replacement technology, carbon nanotubes, that appears capable of exceeding the performance of silicon based chips. However, the most enthusiastic proponents do not claim that the cross-over point will occur before 2020. So, at a bare minimum, it appears that the point of flexation has been reached in the Super S Curve and the incorrect Singularitarian assumption of a ‘law of accelerating returns’ is incorrect. Additionally, there is currently no strong evidence that even when they do become an economically significant technology that their price performance can exceed current technology. If the growth I computer power that can be purchases by a fixed real dollar amount follows and S Curve, the asymptote will likely be about triple that of 2010. If it follows a Gompertz curve, it may be closer to five times today.

There are several components of the Singularity, including nanotechnology and brain-computer technology that, while significant as science, would not result in a change in society fitting the definition of a Singularity, to wit, that the human experience would be transformed beyond our ability to comprehend today. One other prediction, however, does approach this criterion, that of functional immortality. This prediction, however, falls into the category of unknowable. We do know that science is getting close to understanding some of the reasons why the human body deteriorates with time and why that deterioration, sometime prior to 123 years old, invariably has led to death. However, even if we can eliminate the factors of aging that manifest themselves prior to age 123, we have absolutely no idea what problems may await us at, say, 150 or 200 years old.

In the Transformation, we make the prediction that life expectancies are likely to increase over the next twenty years at a rate faster than the historical rate of approximately 1.5 years per decade for a 60 year old. We expect that science will continue to unravel the problems of heart disease and cancer with age and that new scientific approaches, quite likely involving mitochondrial and telomeric therapies, will increase gerontological vigor up to 100 years old and, perhaps, beyond. However, further predictions are nothing more than speculations, since we don’t know what we don’t know about the medical problems of, say, a 150 year old. Furthermore, it is unlikely that any life extension past 150 years will be economically or culturally significant until after the projected date of the Singularity at 2045. Put another way, anyone who is 150 years old in 2045 would need to be 115 today. Dealing with super-extended lifespans belongs to the next Transformation not this one.

In the end, The Singularity is Near is a wonderful exercise in imagining the possible. However, it is bad Futurism. Everyone is familiar with Murphy’s Law, ‘Anything that can go wrong, will go wrong.’ There seems to be imbedded in the Singularity what we call Kurzweil’s Law, ‘EVERYTHING that can go right, will go right.’ When one reads the book, as every serious intellectual should, one should always have in the back of their mind, ‘I may not know which ones, but some of these things won’t happen.’ Murphy was wrong. Everything doesn’t go wrong. However, everything doesn’t go right, either.

In the end, the Singularity is really about a future of immortal humans and god-like computers. It is probable some day. However, it is not likely in the 21st Century. In reality, for the reader, it is the next twenty years that are going to be extraordinarily exciting. The Singularity emphasizes some events that are unlikely and focuses on a date well into the future. However, it ignores some very important and imminent factors. We call it The Transformation. These include an explosion in incomes that will radically change the very fabric of civilization. Internet radio and television are emerging right now and will put the final nails in the coffin of geographically defined cultural hegemonies. This will cause the concept of nation state to fade slowly into irrelevance in favor of administrative districts and networked cultural regions.

The Singularitarians and Transhumanists have wonderful imaginations. However, they lack reasonable perspective. Some of their predictions will come true, though generally later rather than sooner. Some will end up in the same heap with flying cars and nuclear home furnaces.