Thomas Friedman and the Fallacies of Moore’s Law

Thomas Friedman’s recent New York Times column, “If I Had a Hammer,” typifies the stubborn persistence of deterministic thinking about the implications of new technologies. The column is an uncritical gloss on a recent popular book by two MIT business economists, ‪Erik Brynjolfsson and ‪Andrew McAfee, who herald the coming transformation of the “second machine age,” the rise of smart machines, supposedly as a result of something called Moore’s law.

As explained by Friedman, Moore’s law establishes “the relentless doubling of digital computing power every two years.” However, this is not what Moore’s law states. Moore’s law is actually a rather narrow prediction about the exponential growth in density of components (specifically, “gates”) on an integrated circuit. Gordon E. Moore, one of the founders of Intel, came up with this prediction in 1964. In its current form, Moore’s law predicts that chip density will double roughly every two years.

Graph of Moore's law

Graph of Moore’s law, displayed on a semi-log scale to make exponential functions appear linear. Source: Neil J. Gunther, http://www.cmg.org (click on link for original context).

This exponential rate of growth has held steady since the 1970s, making it seem like a law of nature, a relentless force driving technology and society forward. But, as STS scholars Harro van Lente and Arie Rip have noted, Moore’s law functions as a self-fulfilling prophecy, establishing expectations for the rate of improvement in computer chips. Major chip producers use the expectations created by Moore’s law to decide how much they need to invest in better technologies to stay abreast of their competitors. The cost of these investments are staggering. A single cutting-edge chip plant requires billions of dollars. Put simply, Moore’s law remains valid largely because people believe it to be valid, at least the people who plan R&D investments in the semiconductor industry.

Techno-utopians like Friedman place great stress Moore’s law, using it to portray technology as an autonomous, deterministic force creating radical social change. But Moore’s law hardly represents technology in general. No other major technologies experience comparable high rates of exponential growth. New pharmaceuticals and patents don’t grow exponentially. There is no exponential growth in the efficiency and cost of engines, or in communication speeds, as these technologies are limited by laws of nature. Even Internet traffic cannot grow exponentially, as it is constrained by the physical capacity of fiber-optic cables.

Moore’s law really only applies to digital technologies, but even in here it doesn’t imply exponential growth in effectiveness. Doubling your computer’s RAM does not double its power, nor does doubling the number of gates on a microprocessor automatically double its performance. And even if the power of hardware did increase exponentially, hardware is useless without effective software. But software emphatically does not obey Moore’s law. In fact, the greater the size of software, the less efficient it becomes to create, because problems of coordination increase exponentially with size, especially for large projects that require hundreds of programmers. Complex software is very difficult to produce, as the recent problems with the health.gov website show.

Ultimately, Moore’s law is a law of man, not a law of nature. It is the result of human choices, and it will cease to be a law if humans choose differently. But Friedman refuses to acknowledge that human choices drive technological change. Instead, he insists that we need to “rethink deeply our social contracts” to cope with this “technological hurricane reshaping the workplace,” a hurricane that, due to Moore’s law, “just keeps doubling.” Does Friedman ever imagine that we might choose different technologies if we were to subordinate the supposed imperatives of technology to human values? Ultimately, Friedman practices idolatry of technology, making it into a false god whose demands trump everything else.

Advertisements

4 thoughts on “Thomas Friedman and the Fallacies of Moore’s Law

  1. I have been searching for the secret behind the exponential growth in the semi-conductor industry for the past months. Trying to understand why the same “principles” could not be applied to other enabling technologies like the composites. I totally agree with your point. Moore’s law was kept almost like a religious belief, it was the symbolic meaning that united the whole sector and inspired those huge investments and open road mapping exercises. But is there anything to learn from this story, rather than the fact that self-fulling prophecies can create the wave of the future? Why it is so hard to replicate similar beliefs in other sectors?

  2. Thanks for this post, Eric. It addresses many of the concerns that I had with Friedman’s recent article and with Brynjolfsson/McAfee’s book. Regarding this book, I am also curious to know more about the evolution and writing of it, which seems to in large part offer contradictory claims to those made in Brynjolfsson’s earlier book, which was, I believe, self published.

  3. Moore’s law is descriptive – as you rightly point out. However, the human choices you refer to are simply high level causal consequences. We all like to think we’re “choosing”, and we can just arbitrarily “on a whim” choose otherwise at any point. This is certainly how we feel, but this doesn’t describe our actual predicament – which is quite a bit more constrained by causality than you seem to let on (particularly in your final paragraph). We shouldn’t forget that human brains and our decisions are constrained by the laws of nature as well. Although individuals may make silly decisions or just “sell their assets” on a whim, the aggregate result is quite a bit more stable than that.

    Having said this – none of this hand waving gets you a *prescriptive* law – in fact, I know of none. Everything we would say we “know” about nature (including ourselves – we are wholly contained by that category) follows from repeated empirical observations that are descriptive in nature. So – these facts cannot tell us whether Moore’s law will continue on past the integrated circuit – or what the yet unexplored boundaries of physics and technology look like, but whatever DOES end up happening is quite a bit more constrained and less arbitrary than saying it is merely a “law of man”. It’s not a law of man, it’s a law of nature that pre-supposes the existence of a technology creating species. If it were not man, then someone or something else. For all we know the universe could be full of technological species. We don’t even know what the odds of success are or otherwise have any conclusive answer to Fermi’s paradox.

    One possible answer (other than the staple and frankly depressing outlook for our own chances that they all “blew themselves up”) is that there are other ways to communicate that don’t involve sending EM radiation everywhere in space (some entanglement variant perhaps? Apologies for jumping on the funky QM effects technological panacea bandwagon). Perhaps any technological species ahead of us learned that quickly keeping us “out of sync” in time and space of their brief “light cone shell” period of communicating in a fashion as we do now. So physical constraints as we know them can stay as they are – and I have no delusions of exponentials that can grow without end. I’m not saying *anything* is possible, but whatever is possible is something that life (including us) is systematically exploring. Whether we realize it or not, we really have no choice to do otherwise. There is a thread connecting the “survival of the fittest” idea with the desire for economic growth – the expression on the surface can change drastically, but the same algorithm is still at work at the core – it’s all part of one fractal.

    • I just don’t buy this argument. The “laws” of economics are not like the law of gravity, even though classical and neoclassical economists would like to believe they are. Of course human decisions are constrained by the material world, and of course they are shaped by neurological processes that we don’t fully understand, but unless you want to be a complete determinist, people still choose. The consequences of other people’s choices often appear to us as an objective reality, like the number of transistors on a chip or the national highway system, and in fact they are objective realities that we as individuals cannot make otherwise. But these technologies are always the product of human choices. Sure there are regularities in human behaviors, like the prices of commodities, at least until human start behaving differently and everything falls apart.

      On the “for al we know” bit, for all we know our world is just a massive simulation on a quantum computer in the asshole of a perverse deity.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s