Measuring Innovation in an Accelerating World: Review of "A Possible Declining Trend for Worldwide Innovation," Jonathan Huebner, Technological Forecasting & Social Change, 72(8):988-995, © 2005 by John M. Smart
(This article may be reproduced for noncommercial purposes if it is copied in its entirety, including this notice.)

This article is also available in other languages:
Indonesian translation by Jordan Silaen
Serbo-Croatian translation by WHGeeks.
Ukrainian translation by Dmutro Nechuporyk.

Readers of this review may wish to read Huebner's seven page paper, A Possible Declining Trend for Worldwide Innovation, Technological Forecasting & Social Change 72(8):980-986, first (PDF here).


In a 2005 TF&SC paper Jonathan Huebner proposes that rates of global innovations which are judged significant to human beings have been declining in recent decades, since 1914 by an analysis of U.S. patents, which seems contradicted by independent data, and since 1873 by a subjective analysis of "important innovations," which may have greater general merit. I disagree with the author's analysis with regard to technological innovation as we might generally define it, which appears to be increasingly rapid, autonomous, and occurring more below the threshold of human perception with each passing year, while a number of objectively measurable technological capacities (Moore's law, etc.) continue to grow at exponential or slightly superexponential rates. It seems at least plausible that there has been a saturation (or less likely, decline) in human-significant rates of human-initiated innovation, as opposed to innovation significant to and initiated by our technologies, and thus in some subjective or apparent innovation rates, specifically, technological advances that are easily observable and classifiable by human beings. Two other factors that might be contributing to Huebner's observation of declining innovation in the human domain are an apparent saturation of fixed human needs by our accelerating technologies, and the abstract, higher-order, and incremental nature of innovation in today's increasingly virtual and human-surpassing digital environment. If replicable, this article's findings have important implications for better innovation metrics in a world of continuously accelerating change. In the context of other papers on innovation saturation, some also referenced here, Huebner's study may indicate a need for us to learn how to see and measure innovation signficance from a technological, not just a human perspective in coming years.


This is a fascinating article. Though some might seek to dismiss it by pointing out analytical and methodological shortcomings, it is a helpful early effort at understanding the interplay between technological innovation and human psychology, and suggests a number of future developments in innovation studies.

Johnathan Huebner, an independent scholar, proposes to show that the rate of human innovation has been steadily declining since the industrial revolution, and is headed toward an "economic limit" of very low apparent innovation that will be reached circa 2038. He makes his argument using two different methods: U.S. patent data and "important innovation" data, subjectively assessed by one account.

Let's deal first with his patent data, which I believe are less valid, and then move on to the other method of his argument, which is potentially more interesting. Huebner provides U.S. patent data which show that, when normalized to total U.S. population, there was a patenting peak in 1914, a significant drop from 1914-1985 to 50% of the 1914 value, and a recent rise between 1985 and 1999 back to 75% of the 1914 value. He suggests this distribution looks "most" like a bell curve, that the 1985-1999 spike is only a temporary anomaly, and that the per capita "innovation rate" of the U.S. has been declining since 1914. Huebner's curve is at right, reprinted by permission of Elsevier/TF&SC.

Looking for more recent data, I went to the same general sources ([1] U.S. PTO data for patents, but using different official US PTO tables at the site, and [2] U.S. Census data for population), and found that patents today, per capita, are back up to 95% of the 1914 peak (see [3] for my calculations). I do not know why Huebner's patents graph didn't have data more recent than an average from 1990-1999 as its most recent point. From my perspective, if 2003 data were included they would have refuted his argument that U.S. patents per capita fit a bell curve and are now in a declining trend. And when we take a longer view, issued utility patents increase 14 fold from 1870 to 2003, while U.S. population increased only 7 fold over the same period.

Huebner proposes that patents can be considered a "basic unit of technology," but I find them to be mostly a measure of the kind of technology innovation that humans consider defensible in particular socioeconomic and legal contexts, which is a crude abstraction of what technology is. Our judgments of importance are as much a measure of social custom as they are of assessed originality and value. The U.S. doesn't allow lots of basic process patents, for example, while Japan does, which makes for different patent climates in these two countries. The demographics of patenting in the U.S. have also changed dramatically as well. In 1901, four out of five U.S. patents were issued to individuals, but in 1999, more than four out of five were issued to corporations. Today, patenting frequency may be more a function of their perceived litigation value to U.S. corporations, which varies by industry and judicial context, rather than of perceived business utility to individual inventors, as may have occurred in the age of greater individual invention in the 19th and early 20th centuries. Recently, for example, we have seen a corporate-driven patent frenzy in the U.S. that may be attributable more to shortcomings in intellectual property law than to any genuine surge in innovation. In summary, patents seem to be a poor and problematic metric of accelerating technological innovation, and this is valuable to realize.

For his second set of data Huebner plots subjective "important innovation" data from Bryan Bunch and Alexander Hellemans' [4] survey work, The History of Science and Technology, 2004, involving 7,198 subjectively "significant innovations" they note from the end of the "Dark Ages" in 1453 A.D. to the present time. When normalized to total world population, these fit on a modified Gaussian (bell-shaped) curve with an innovation peak around 1873, early in the industrial revolution, and a roughly 66% drop (from 16 to 7) in "significant" events/year/1,000,000,000 people. Reprinted below by permission of Elsevier/TF&SC.

We know there's something odd about a measure of innovation that doesn't show a dramatic spike for all the advances that occurred, for example, between 1940 and 1945 during the feverishly innovative era of World War II. Recall the great strides made in computing, aviation, warfare, organizational methods, large scale engineering and manufacturing projects, new political structures, and so many other areas during this time, yet Huebner's curve, drawn from Bunch and Hellemans' data, shows a downtrend in global innovation per capita during this period. So we have a problem in definition or methodology here.

We already know Bunch and Hellemans are not independently counting the ability to make and deliver more of something at an affordable cost (e.g., innovation diffusion), which always involves additional and separate innovations beyond those culminating in the first prototype. In other words, they may be reporting some subset of innovation (the perception of mass utility?) rather than innovation in general. Are they also biasing against innovations that emerge during an era where 55 million human beings die as a consequence of their use?

With regard to Huebner's treatment of the Bunch and Helleman data, normalizing an innovation rate to total world population also has its problems. We might expect the global rate of perceived innovation to be overwhelmed, at least temporarily, by an exploding third world population. But Huebner argues that world GDP growth, university and student growth, and possibly education expenditure growth have all outstripped general population growth over the time period studied. So if he had normalized to more education-specific measures, for example, the innovation decline he reports would have been even worse. In other words, the world's economic and educational-technologic development infrastructure are already outstripping human population growth, yet the apparent pattern still persists. This makes his argument particularly interesting. If we assume for the sake of argument that Bunch and Hellemans perceptions have some replicablility whatever it is they are classifying, what then might Huebner be striving to clarify?

As one potential explanation, we must consider the possibility that human-initiated innovation, like energy consumption and population growth, is a process that naturally saturates with rising global income levels and technological intelligence. Shell International's 2001 report [5] "Energy Needs, Choices, and Possibilities: Scenarios to 2050," summarizes IMF and British Petroleum data which note that in every economy where per capita GDP goes above $15,000/year (e.g, the U.S., Europe, Japan, Australia), growth in energy use per capita, after rapidly increasing at lower income levels, begins to slow dramatically and then effectively stops. This saturation may be due to several factors: the increasingly service intensive, information intensive, and "virtual" nature of developed economies, the sharply fixed basic needs (transportation, housing, etc.) of human beings, the increasing sustainability politics of affluent nations, and perhaps most importantly, the incredibly rapidly advancing energy efficiencies of all our replicating machines (unlike the replicating bodies of their human users). At a GDP per capita of $25,000/year, energy growth per capita becomes so slow that it is effectively saturated. Europeans like to say that Americans are much less interested in energy conservation than they are, but the Shell report (see the graph on page 7) clearly shows that the U.S. has saturated in our energy consumption as well. The only difference is that our culture saturates at 350 Gigajoules/capita, while Europeans saturate at 150. This 2X difference seems almost trivial by comparison to the exponentiating capacities of our technological infrastructure.

Population follows a similar saturation with global economic and technological development. It is now well known that total population sizes, after immigration is factored out, are on the decline in every first world country irrespective of culture. Furthermore, the second derivative of world population growth went negative for the planet in the 1970's (this was the inflection point in the S-curve for global population) and even for India and Africa in the 1990's. Several independent estimates now project our total world population to hit a maximum circa 2050, followed by an accelerating decline thereafter, a time when even emerging nations will exhibit the "technological contraceptive" effect we now see in the first world, where non-immigrant birth rates (1.3, 1.5, 1.7 etc. for every two adults) are always consistently below replacement level (2.1 children for every two adults). There always seems to come a point in every nation's evolutionary development where the human interest in reproduction begins to conflict with our rapidly improving social, economic, and technological choices for personal and child advancement.

Furthermore, considering the rapid pace of globalization today, it seems plausible that the world as a whole will reach the lower echelons of the first world's current level of technological development within this century. Emerging nations increasingly employ "leapfrogging technologies" in information processing, communication, energy, transportation, agriculture, health, etc. which allow them to make GDP and technology advances using a fraction of the time and resources required by their first world predecessors. Who would have anticipated, for example, that Chile would already have 428 mobile phones per 1,000 people today, while the U.S. has 488 per 1,000? [6]

Such trends make it seem obvious to me, though it might not be so to others, that as technological progress increasingly satisfies current human needs, individuals become less concerned with technological development and turn more toward personal growth, unique experiences, and other activities which, while equally creative on an individual level, are less obvious examples of innovation in a technological sense.

The sociologist Ronald Inglehart [7] (The Silent Revolution, 1977; Culture Shift in Advanced Industrial Society, 1989; Modernization and Postmodernization, 1997) has extensively documented this predictable value shift in industrializing countries. As I interpret Inglehart's work, in addition to more tolerant ideologies and other predictable developments, the more industrialization we experience the more we become ready to take a long-deserved break from generations of toiling, including much of the traditional work of innovation, and the more we become willing to let our machines take over the task of supplying our very finite human needs.

The longstanding progressive improvement in and individualization of leisure in developed societies has been long identified by such forecasters as Herman Kahn [8] (with Anthony Wiener, The Year 2000, 1967), and recently Virginia Postrel [9] in The Substance of Style, 2003. Fortunately, new surveys like the BLS America Time-Use Survey will carefully track trends in the way we spend our leisure time, a poorly studied subject to date both nationally and globally. The 2003 ATUS [10] found that on an "average day," persons in the U.S. age 15 and over slept 8.6 hours, spent 5.1 hours doing leisure and sports activities, worked for 3.7 hours, spent 1.8 hours doing household activities, and divided the remaining 4.8 hours among a variety of other activities, including eating and drinking, attending school, and shopping. I would expect that even the recent disruptions of globalization would be unlikely to significantly affect these numbers, and such disruptions always disproportionately affect the developing world.

One measure of total environmental innovation, both human and machine-initiated, may be the number of choices available to the richest members of society, and the time and dollar value they place on those choices. One proxy for this may be the leisure our richest societies experience collectively, or perhaps the total number of hours in a day divided by the average hours worked by self assessment. By this measure we are living in an age of tremendous environmental innovation. But as the structure of my proposed metric would argue, increasingly less of this can be human-initiated if we only "work" 3.7 hours a day averaged across all our adult citizens (retirees included).

On one hand, we have more and smarter people on our planet, living longer than ever before, so we might expect more total human innovation than ever before. At the same time, it also seems plausible that human-generated innovation per capita may be trending down in recent generations, as technology-generated innovation rapidly increases. In other words, while it is reasonable to expect more innovation going forward from those who wish to innovate, and more total environmental innovation per capita, there may actually have been less human-initiated innovation per capita in recent generations if we were to carefully measure all the work being done by our increasingly clever and subtle machines. If it is also true that many classes of technological innovation are even harder to see than human innovation, this may be the main driver of the downturn Huebner is charting.

In the long run I would expect this to be a moot point if humans are also becoming increasingly intimately integrated with our machines, as several technology scholars (e.g., Ray Kurzweil, myself) propose. At some point, technology seems very likely to become an indistinguishable extension of our humanity. But it is possible that we'll see less human-initiated innovation per capita for a few more generations to come, and perhaps this is the trend Huebner is attempting to characterize. At the same time, as our leisure individualism increases (not "sovereign individualism," but a milder and more consumerist form), the kind of innovation that humans generate may also be changing, becoming increasingly higher-order and abstract (e.g., more psychosocial, health, and stylistic innovation), and perhaps also harder to perceive. This adds to the measurement problem.

Another critique of the Huebner article is that the innovations Bunch and Helleman chose to include in their introductory book were entirely subjective. One could argue their particular data may have been more a function of their research sources, procedures, assumptions, and biases than anything else. Furthermore, many systems scholars have put together alternative canonical innovation sets (Ray Kurzweil's uses a compilation from 14 different thinkers and reference works) and shown a clear trend of acceleration, not deceleration. Nevertheless, if several different subjective assessments all suggest innovation is decreasing, even if they differ substantially in the specifics of their analysis, there's something here worth better understanding. Increasingly, Huebner's argument has company, which I believe makes his perspective worthy at least of careful consideration.

As one example Ted Modis, [11] in "Forecasting the Growth of Complexity and Change," in the same journal, Technological Forecasting & Social Change, V69, No 4, 2002, using a different set of subjective data, also made the claim that important innovations have reached a past peak for human civilization and are presently declining. Modis' innovation peak was 1990, which might make his proposed downturn less plausible as a system change than as a recent fluctuation, but again we should look beyond the analytic particulars to ask whether there's something that is causing Modis to see saturation that deserves better understanding. In another example, systems theorists Tessaleno Devezas and George Modelski, [12] in Technological Forecasting & Social Change, V70, No 9, 2003, argue that world system change, while still upsloped, has been slowing for 1,000 years, with the inflection point at roughly 1000AD. Their model proposes that human social development is in a decelerating phase and is about "80% complete", and thus that the major features of human social organization are now in place. Francis Fukuyama [13] makes a similar point with regard to liberal democratic capitalism as a stable developmental attractor in The End of History and the Last Man, 1993, and John Horgan [14] also touches on these ideas in his thought-provoking The End of Science: Facing the Limits of (Human) Knowledge in the Twilight of the Scientific Age, 1997.

Such arguments seem plausible when we consider the fixed capacities of human biological systems relative to the accelerating technological systems rising all around us. I've written briefly about the Dvezas-Modelski paper in a previous issue [15] of the ASF newsletter Accelerating Times. Both systems theorist Kenneth Boulding and internet archivist Brewster Kahle have made a related point. They have independently suggested the era around the end of the 19th century, with the invention of the internal combustion engine and the commercialization of electricity, the era of Edison, and Tesla, was a far more innovative age than the one we live in today, as well as a time with significantly greater social impacts of accelerating technological change.

I think there are important psychological, perceptual, and developmental dynamics involved in these assessments of innovation saturation. Like the irrepressible anomaly in the orbital precession of the planet Mercury that aided the development of Einstein's new understanding of space and time, these anomalous models of change, should they persist, may help us develop a new paradigm for understanding technological change. In the process, we may also learn how to build better innovation metrics, so we can observe and predict the real accelerating changes occurring all around us.

It is my intuition, supported by today's crude exponential technology capacity growth metrics such as Moore's law (processing), Gilder's law (bandwidth) Poor's law (network node density), Cooper's law (wireless bandwidth), Kurzweil's law ([16] price performance of computation over 120 years) and many others, that technological capacity and technological innovation have always accelerated since the birth of human civilization, and that their growth remains exponential or gently superexponential today. Furthermore, there are a number of books, such as Carl Sagan's [17] The Dragons of Eden, 1977, Richard Coren's [18] The Evolutionary Trajectory, 1998, and an interdisciplinary book [19] by Laurent Nottale (an astrophysicist), Jean Chaline (a paleontologist), and Pierre Grou (an economist) Trees of Evolution, 2000, that have shown a developmental pattern of continuous acceleration on cosmic as well biological, cultural, and technological scales. Nevertheless, we now have Huebner and company's saturation perspective conflicting with these more numerous acceleration models. I think we will learn something in their reconciliation.

As another potential explanation of Huebner et. al.'s perspective, consider the observation that modern examples of innovation occur increasingly "under the hood" of the engine of change, below our threshold of easy perception. I've made this argument previously on my AccelerationWatch website [20] with regard to the "Dark Ages" after the fall of the Roman Empire. While many easily observable forms of innovation slowed in those politically repressive times (city sizes shrank, mega-projects fell into disrepair, etc.), scholars like Anne-R-J Turgot, [21] Reflections on the Formation and Distribution of Wealth, 1766, noted the "inevitable" march of technological progress that occurred even during this period, but on more local and smaller scales appropriate to the shrinking social structures in the West (not the East) at that time.

So while human social innovation may follow political and generational cycles of advance and regrouping, technological innovation may be becoming both smoother and subtler in its exponential growth the closer we get to the modern era. Perhaps this is because since the industrial revolution, innovation is being done increasingly by our machines, not by human brains. I believe it is increasingly going on below the perception of humans who are catalysts, not controllers, of our ever more autonomous technological world system.

Ask yourself, how many innovations were required to make a gasoline-electric hybrid automobile like the Toyota Prius, for example? This is just one of many systems that look the same "above the hood" as their predecessors, yet are radically more complex than previous versions. How many of the Prius innovations were a direct result of the computations done by the technological systems involved (CAD-CAM programs, infrastructures, supply chains, etc.) and how many are instead attributable to the computations of individual human minds? How many computations today have become so incremental and abstract that we no longer see them as innovations?

To his credit, Huebner speculates that the declining innovation he sees may be due to the "limits of the human brain." But I am not sure whether he would also agree that our brains are not only increasingly unable to engage in truly different classes of innovation, they seem to be increasingly unable to perceive the technology-driven innovation occurring all around us. I believe that creates an opportunity for us to develop substantially better models of our accelerating future.

As yet another interesting possible explanation, certain types of innovation saturation might now appear to be occurring because our accelerating technological productivity is beginning to intersect with an effectively fixed number of human needs. Humans have a very finite set of physical needs, and even when considering psychological needs and desires, our biocomputing systems operate on scales that are multi-millionfold slower than those of our emerging technological successors. For a good analogy, I suggest you think of the entire human species on earth like a large collection of plants, slowly extending ourselves over the planet's surface, and then think of our emerging computer infrastructures like human beings, able learn, think and move so fast (using electricity rather than chemical diffusion as their rate-limiting computational process) that human cognitive systems are effectively rooted in space and time, like a plant, by comparison. How many physical needs does a plant have in comparison to those of a human? How rapidly can a human saturate a plant's needs, as long as it remains a plant?

As a final proposed explanation of the article's findings, we may observe that as the world develops and we all climb higher on Maslow's hierarchy of relatively fixed needs, those who already have sufficient housing, transportation, etc., are now pursuing innovations on the most abstract, virtual, and difficult-to-quantify levels, like social interaction, status, entertainment, and self-esteem. All this may be a direct result of the leisure individualism discussed earlier. Would Bunch and Hellemans' innovation metric treat psychological profiling internet dating websites like [22] as an "important" innovation for their list? Or new, network-enabled modes of innovation such as the open source software movement [23] or the graphical and socioeconomic constructs now emerging in persistent virtual worlds like Second Life [24]? If such emergences aren't counted we will have difficulty seeing the accelerating innovations occurring in our environment going forward, because they are increasingly higher-order, virtual and abstract.

In short, there are a number of opportunities for us to improve our innovation measures in coming years, to reflect possible saturations in human-initiated vs. technology-initiated innovation, in human awareness of innovation, and in human physical and psychological needs, as well as the increasingly abstract, higher-order, and incremental nature of innovation in today's ever more virtual and human-surpassing digital environment.

Are you a researcher who would like to work with us to apply for grants to do studies on these topics? Would you like to speak on this or a related topic at our next conference? Send us an email [mail(at)] and let's consider how we can work together to advance the dialog on these fascinating and future-important issues.


Thanks to Robert Adler, Jef Allbright, Patricia Bacon, Iveta Brigis, Troy Gardner, Norman Gilmore, Alex Jacobson, Ray Kurzweil, Hal Linstone, and Vernor Vinge for helpful feedback.

References and Footnotes

[1] U.S. Patent Statistics Chart, Calendar Years 1963-2003,
[2] U.S. National Population Estimates,
[3] Total U.S. patents in 1995 were 113,834, with 0.57 of U.S. origin. U.S. population was 263 million, giving 247 U.S. patents/year/million population. Total U.S. patents in 2003 were 187,017, with 0.53 of U.S. origin. U.S. population was 291 million, giving 340 U.S. patents/year/million population. Huebner's data show 355 U.S. patents/year/million population in 1914.
[4] B. Bunch, A. Hellemans, The History of Science and Technology, Houghton Mifflin Co., New York, 2004.
[5] "Energy Needs, Choices, and Possibilities: Scenarios to 2050,"
Shell International, 2001
[6] "World Changing Ideas," Technology Review, April 2005, p. 46
[7] R. Inglehart, The Silent Revolution, Princeton University Press, 1977; Culture Shift in Advanced Industrial Society, Princeton University Press, 1989; Modernization and Postmodernization, Princeton University Press, 1997)
[8] H. Kahn, A. Wiener, The Year 2000, Macmillan, 1967
[9] V. Postrel, The Substance of Style, HarperCollins, 2003
[10] BLS America Time-Use Survey, 2003
[11] T. Modis, "Forecasting the Growth of Complexity and Change," Technological Forecasting & Social Change, V69, No. 4, Elsevier, 2000
[12] T. Dvezas, G. Modelski, Technological Forecasting & Social Change, V70, No 9, Elsevier, 2003
[13] F. Fukuyama, The End of History and the Last Man, Perennial, 1993
[14] J. Horgan, The End of Science, Broadway, 1997.
[15] J. Smart, Ed., Accelerating Times, 1.18.2005.
[16] R. Kurzweil, "The Law of Accelerating Returns," 2001.
[17] C. Sagan The Dragons of Eden, Ballantine, 1977/86
[18] R. Coren, The Evolutionary Trajectory, CRC Press, 1998
[19] L. Nottale, J. Chaline, P. Grou, Trees of Evolution, Hachette, 2000
[21] A. Turgot, Reflections on the Formation and Distribution of Wealth, Othila Press 1766/1999

About the Author

John Smart is developmental systems theorist who studies accelerating change, computational autonomy, evolutionary development, and the technological singularity hypothesis (the possibility of progressively human-surpassing technological intelligence). He is president of the Acceleration Studies Foundation (, a 501c3 nonprofit engaged in research, education, and selective advocacy of communities and technologies of accelerating change. His personal website is Acceleration Watch (, e-mail: johnsmart(at)
Permalink for this article: