Claims

Factual · AI timelines and forecasting

“Whatever the approach, we find–at least–double exponential growth.”

Ray Kurzweil, essay author · 23 October 2000 · Promise + Peril (Kurzweil Library)

Examining technology data in many ways and time scales shows at least double exponential growth.

Mostly true True in substance, wrong in a detail.

The second judge would not resolve this claim; the verdict shown is the first judge's.

What would count as true or false

Judge as of 2000-10-23. The claim is that technology data, looked at in various ways and over various time scales, shows at least double exponential growth (the exponential growth rate itself increasing exponentially). True: technology datasets available by 2000 (e.g. computations per second per dollar over the 20th century, and other price-performance series) consistently show an accelerating growth rate that fits double exponential or faster, in independent analyses as well as Kurzweil's own. Mostly true: the main series (e.g. computing price-performance) shows a growth rate that clearly accelerates and fits double exponential, but other series fit plain exponentials. Mixed: about half the representative series show acceleration beyond exponential and the others show plain exponential growth. Mostly false: technology data generally show exponential growth but not double exponential; the claim takes a single or cherry-picked series and states it as general, or the fit is unproven or depends on the method. False: analyses show the data are clearly not double exponential (plain exponential or slower) in the cases cited.

Evidence

1. Exponential and non-exponential trends in information technology - Machine Intelligence Research Institute Machine Intelligence Research Institute · 12 May 2014

   “Notice that clock speed stopped improving exponentially in 2004, but transistors per chip has continued to increase exponentially via the jump from single-core to multicore processors.”

2. the Kurzweil Library Tracking breakthroughs in tech, science, and world progress. KurzweilAI.net (reprint of Scientific American article by Ray Kurzweil) · 1 September 1999 Own record

   “The graph is a study in exponential growth: computer speed per unit cost doubled every three years between 1910 and 1950 and every two years between 1950 and 1966 and is now doubling every year.”

3. Technological singularity Wikipedia · 30 May 2002

   “On the other hand, it has been argued that the global acceleration pattern having a 21st-century singularity as its parameter should be characterized as hyperbolic rather than exponential.”

4. Two Centuries of Productivity Growth in Computing | The Journal of Economic History | Cambridge Core Cambridge University Press (The Journal of Economic History) · 30 March 2007

   “For the period 1850 (which I take as the birth of modern computing) to 2006, the compound logarithmic growth rate is around 20 percent per year.”

5. Two Centuries of Productivity Growth in Computing | The Journal of Economic History | Cambridge Core Cambridge University Press (The Journal of Economic History) · 30 March 2007

   “For the period 1945–1980, cost per computation declined at 37 percent per year, as compared to 64 percent per year after 1980.”

6. the Kurzweil Library Tracking breakthroughs in tech, science, and world progress. The Kurzweil Library / KurzweilAI.net (reproducing Scientific American) · 1 September 1999 Own record

   “The graph is a study in exponential growth: computer speed per unit cost doubled every three years between 1910 and 1950 and every two years between 1950 and 1966 and is now doubling every year.”

7. Technological singularity Wikipedia

   “argue that the rate of technological innovation has not only ceased to rise but is actually now declining.”

8. Technological singularity Wikipedia

   “it has been argued that the global acceleration pattern having a 21st-century singularity as its parameter should be characterized as hyperbolic rather than exponential.”

Reasoning

An independent study of computing productivity [E5] finds that the decline in cost per computation sped up, from 37% a year in 1945–1980 to 64% a year after 1980. So the main series, computing price-performance, has a growth rate that is itself rising, as Kurzweil's own figures also describe [E2]. The evidence does not show the same pattern across many technology series. Other sources cut against the general claim: some writers argue the rate of innovation is falling [E7], and others say the pattern is better described as hyperbolic rather than exponential [E8]. The main series fits the claim, but its broader statement that every approach shows it is not established, which matches the 'mostly true' criterion.

Cite this claim

Ray Kurzweil, “Examining technology data in many ways and time scales shows at least double exponential growth.” (said 23 October 2000). Verdict: mostly true. Reliability API, claim c_d42d9d5bb00c, method 0.1.0+28da144b. https://epistacklabs.com/claims/c_d42d9d5bb00c