ClaimsRay Kurzweil

“it took ninety years to achieve the first MIPS (million instructions per second) per thousand dollars, now we add one MIPS per thousand dollars every day.”

Ray Kurzweil, essay author January 2001 The Law of Accelerating Returns (Kurzweil Library) archived copy

Factual AI compute, hardware and data

It took ninety years to achieve the first MIPS per thousand dollars, and as of 2001 one MIPS per thousand dollars is added every day.

Unresolvable The public record could not settle it.

Factual: Statements about the present or the past, checked against the record: did they hold up?

What would count as true or false

Judge as of January 2001. The claim has two parts: (a) computing price-performance took about ninety years (from about 1900) to reach 1 MIPS per $1,000; (b) by about 2001 price-performance was rising by about 1 MIPS per $1,000 per day, meaning roughly 365 MIPS per $1,000 per year. True: published price-performance data put (a) at about 80–100 years and (b) at about 0.5–2 MIPS per $1,000 per day. Mostly true: both hold in substance, with one figure modestly outside those bands (for example 70 or 110 years, or a daily gain within a factor of about 3). Mixed: one part holds and the other is off by more than that. Mostly false: both are distorted, such as an order-of-magnitude error in one with the other only roughly right. False: both are wrong by large margins.

Set before the research began.

Evidence (8)

  1. E1 The Coming Merging of Mind and Machine the Kurzweil Library (reprinting Scientific American) · 1 September 1999 Own record

    “It took 90 years to achieve the first $1,000 computer capable of executing one million instructions per second (MIPS). Now we add an additional MIPS to a $1,000 computer every day.”
  2. E2 The Progress of Computing Cowles Foundation Discussion Papers, Yale University (EliScholar) · 2001

    “Performance in constant dollars or in terms of labor units has improved since 1900 by a factor in the order of 1 trillion to 5 trillion, which represent compound growth rates of over 30 percent per year for a century.”
  3. E3 Trends in the cost of computing [AI Impacts Wiki] AI Impacts · 10 March 2015

    “Since the 1940s, MIPS/$ have grown by a factor of ten roughly every five years, and FLOPS/$ roughly every 7.7 years.”
  4. E4 Trends in the cost of computing [AI Impacts Wiki] AI Impacts · 10 March 2015

    “Prior to the 1940s, growth appears to be ambiguous and small.”
  5. E5 The Coming Merging of Mind and Machine the Kurzweil Library (reprint of Scientific American article by Ray Kurzweil) · 1 September 1999 Own record

    “It took 90 years to achieve the first $1,000 computer capable of executing one million instructions per second (MIPS). Now we add an additional MIPS to a $1,000 computer every day.”
  6. E6 Chapter 10 The Emergence and Impact of Intelligent ... NASA NTRS (chapter by Raymond Kurzweil, Kurzweil Technologies) Own record

    “it took 90 years to achieve the first MIPS (million instructions per second) per thousand dollars; now we add one MIPS per thousand dollars every day.”
  7. E7 [PDF] Thematic investing - docfinder.bnpparibas-am.com BNP Paribas Asset Management · September 2022

    “it took 90 years to achieve the first MIPS per 1000 dollars; now we add 1.2 MIPS per 1000 dollars every hour.”
  8. E8 Trends in GPU price-performance Samaritan Research · 27 June 2022

    “| Nordhaus, 2001 | CPU-based | MIPS/$ | 1.6 years | 5.3 years | 0.189 OOMs/year |”

Reasoning

The ninety-year and one-MIPS-per-day figures appear only in Kurzweil's own writings (E1, E5, E6), which do not verify them; a later report (E7) repeats the ninety-year figure but gives a different rate. The independent sources give growth rates: over 30% per year since 1900 (E2), a factor of ten roughly every five years since the 1940s (E3), and a Nordhaus doubling time of about 1.6 years (E8). None of them states when 1 MIPS per $1,000 was first reached or what the price-performance level was in 2001, so neither the ninety-year span nor the daily gain can be checked against the criteria's bands.

Cite this claim

Ray Kurzweil, “It took ninety years to achieve the first MIPS per thousand dollars, and as of 2001 one MIPS per thousand dollars is added every day.” (said January 2001). Verdict: unresolvable. Reliability API, claim c_155246115434, method 0.1.0+28da144b, dataset release v1.1.2. https://epistacklabs.com/claims/c_155246115434

This claim is in dataset release v1.1.2, with the same verdict.

Reply to this claim

For Ray Kurzweil, or someone speaking for them: a correction or counter-evidence. Replies are read before anything is shown; an approved reply shows your name, role, date, statement and link. Your email address is never shown and is deleted 30 days after the decision (privacy).

More factual claims by Ray Kurzweil

  1. Mixed Factual March 2025
    “But I got to 10 billion. It started to do things.”

    Kurzweil said that at about 10 billion connections neural networks started to do things.

  2. Mixed Factual 9 July 2024
    “the cloud right now is expanding exponentially. It gets twice as fast every year for the same money.”

    Cloud computing is expanding exponentially, doubling in speed every year for the same money.

  3. Mostly false Factual 29 March 2024
    “a 20 quadrillion–fold increase over 80 years.”

    Computation/AI capability has increased 20 quadrillion-fold over 80 years.

  4. Resolved true Factual 12 December 2002
    “It’s remarkable how long it’s lasted, but it wasn’t the first, but the fifth paradigm to provide exponential growth to computing.”

    Moore's Law (shrinking transistors on integrated circuits) was the fifth paradigm, not the first, to provide exponential growth in computing.

  5. Mostly true Factual January 2001
    “A computer can also remember billions or even trillions of facts perfectly, while we are hard pressed to remember a handful of phone numbers.”

    A computer can remember billions or trillions of facts perfectly, while humans struggle to remember a handful of phone numbers.

All of them

Verdicts are made automatically; how verdicts are made.