The Luck of the Irish
I don’t remember when I first encountered network economics, but I do remember when I first noticed that it had entered the interest of a broad public. M. Mitchell Waldrop’s Complexity: The Emerging Science at the Edge of Order and Chaos became a bestseller in 1993.5 Waldrop opened with the story of Brian Arthur, Catholic son of Belfast, sitting alone and dejected in a strange Silicon Valley bar on St.
Patrick’s Day 1987. But Arthur was not just any son of Eire; he was a chaired Professor of Economics at Stanford bemoaning the poor reception that his ideas had received from his colleagues at Berkeley. Those ideas laid the groundwork for a New Economics based on increasing returns, one of the central concepts of network economics.Waldrop was hardly the first one to bring these ideas beyond the temple walls; Arthur had done that himself a few years earlier in the pages of Scientific American.6 Arthur explained that “classic” economics derived from the insights of late nineteenth century thinkers. At that time, agriculture, mining, and basic manufacturing dominated the world economy, and the most advanced areas of scientific inquiry still lay primarily in Newtonian mechanics. Classical economics tend to reflect those sources, by considering, for example:
the competition between water power and coal to drive electrical generators. As hydroelectric plants take more of the market, engineers must exploit more costly dam sites, thus increasing the chance that a coal-fired plant will be cheaper. As coal plants take more of the market they bid up the price of coal... thus tipping the balance toward hydro. The two end up sharing the market in a predictable proportion that best exploits the potentials of each.7
In other words, if coal is expensive I’ll build a hydro plant, and if hydro is expensive I’ll go for coal.
Classical economics predicts that our energy supply should include some of each—and empirical experience bears that out.Stated more generally, the classical paradigm predicts that if two technologies perform the same basic task, and if each one appears to be superior under some but not all sets of circumstances, different people will make different choices and both technologies should survive. Real world experience with real world technologies and products validates that prediction. In my own home, I operate at least three lightbulb technologies (incandescent, fluorescent, and compact fluorescent), two telephone technologies (fixed and mobile), two shaving technologies (manual and electric), and three coffee-making technologies (drip, percolation, and French press).
But I also remember when I owned a Mac running Mac OS, a PC- clone running DOS, and a Sun workstation running Unix—and when my local video rental store stocked both VHS and Beta videocassettes. The classical predictions didn’t seem to work there. That bothered a small group of economists to which Arthur belonged. They noticed that some products, particularly though not exclusively new technology products, seemed to behave differently. In these settings, two roughly comparable products that perform more-or-less the same task, introduced around the same time, appeared to gain adherents in roughly equal numbers. But after a relatively brief introductory period, one emerged as the clear market leader. And then, suddenly and without any apparent explanation, the other dwindled to a niche market—or disappeared altogether. The rivalry between VHS and Beta illustrated the point nicely, and conventional wisdom concerning Beta’s technical superiority drove it home even further. If conventional wisdom were true, the market got it wrong! It tipped the wrong way, and locked consumers into the weaker VHS technology. And therein lies a problem, because capitalism’s basic rules taught us that markets don’t get things wrong!
But neither Arthur nor those economists who shared his concerns were heretics.
They weren’t interested in squelching economic research, but rather in expanding its horizons. They sought both additional examples and an explanatory theory. Arthur glommed onto the example of clocks running “clockwise” almost by accident. While “everyone knows” that clocks are numbered with the 12 at the top, the 1 to its right, and numbers falling sequentially to the right, there’s no clear reason why they shouldn’t be numbered sequentially to the left (or even, for that matter, with the 12 on the bottom). In response to a question after one of his lectures, Arthur casually predicted that other conventions should have existed at some point—and shortly thereafter, a listener sent him a photo of a clock numbered “counterclockwise.”8Perhaps an even simpler example comes in written languages. There’s no particular reason that my cursor traverses the screen from left to right. Ancient inscriptions didn’t necessarily adhere to any convention; Biblical era tombstones and coins arranged letters to form aesthetically pleasing shapes. Eventually, the Canaanites, the Judeans, the Israelites, the Arameans, and the Arabs all decided to orient their writing from right to left. The Greeks made the equally arbitrary decision to proceed from left to right. The Romans followed the Greeks’s lead with Latin, and we pretty much picked it up from there—hence my cursor. But historical precedent doesn’t make a decision any less arbitrary—or even irreversible. When Mustafa Kemal Ataturk founded modern Turkey, he switched written Turkish from the Arabic to the Latin alphabet—thereby reorienting the language. Ataturk thus chose a standard, imposed it on his people, and quickly drove out the old standard.
The most widely discussed example of this phenomenon, though, occurred with an important nineteenth century innovation: the typewriter. Our current QWERTY keyboard, named for the first six letters in the top row (counting arbitrarily from left to right) is hardly the most obvious organization of the alphabet. In fact, during the first few decades of typewriters, QWERTY competed with other keyboard designs.
How did it win? According to Waldrop:An engineer named Christopher Scholes designed the QWERTY layout in 1873 specifically to slow typists down; the typewriting machines of the day tended to jam if the typist went too fast. But then the Remington Sewing Machine Company mass-produced a typewriter using the QWERTY keyboard, which meant that lots of typists began to learn the system, which meant that other typewriter companies began to offer the QWERTY keyboard, which meant that still more typists began to learn it, et cetera, et cetera And now
that QWERTY is a standard used by millions of people, it’s essentially locked in forever.9
The pattern holds time after time after time. A vendor in a competitive market introduces a new product. Through some combination of product quality, outside events, and luck, the right decision-maker adopts it at the right time. Suddenly, it becomes the “must adopt” product, the de facto standard, and frequently the only player left in the game. Sometimes this victory occurs because the product is clearly superior to all of its competitors. Frequently its superiority (or lack thereof) is irrelevant. The market tips and the victor emerges.
Arthur proposed an explanation of such “increasing returns.” In an increasing returns world, the strong get stronger, the weak get weaker, and markets tip to a standard. This phenomenon arises infrequently— that is, it doesn’t explain most markets—but often enough to be of broad general interest. Of even greater interest, though, is where it tends to occur when it does occur. Arthur’s increasing returns tend to cluster in cutting-edge technologies, and in particular in knowledge-based industries. They’re thus critical to the information sector.
Of course, referring to the idea as “Arthur’s increasing returns” represents a severe injustice to the many other economists who explored the same ideas at the same time. Arthur himself noted that:
In the last few years I and other economic theorists at Stanford, the Santa Fe Institute, and elsewhere have been developing a view of the economy based on positive feedbacks.
Increasing-return economics has roots in economic thinking that go back for seventy or more years, but its application to the economy as a whole is largely new.10He even credited Alfred Marshall—among the most influential of those classical 1890s economists—with the critical underlying observation that “a firm that by good fortune gained a high proportion of the market early on would be best able to best its rivals; ‘whichever firm first gets off to a good start’ would corner the market.”11 Even the contemporary version of the ideas preceded St. Patrick’s Day 1987. Jeffrey Rohlfs (re?)introduced them at least as early as 1974;12 Paul David introduced the QWERTY story to the modern economic literature in 1985;13 and Michael Katz and Carl Shapiro published seminal articles in two of the most prestigious scholarly economics journals in the mid 1980s.14
More to the point, though, Arthur’s despondency over his poor reception at Berkeley notwithstanding, his was hardly a lone voice in the wilderness. True voices in the wilderness do not become chaired professors at Stanford or affiliates of the Santa Fe Institute—as Arthur already was back on that doleful St. Patrick’s Day. While one particular Berkeley seminar may have been a flop, economists were hardly shunning his ideas about network effects.
Few voices speak louder about a theory’s growing acceptance than those of its critics. About the time that Arthur revealed his ideas in Scientific American, Stan Liebowitz and Stephen Margolis decided to debunk the QWERTY story. In The Fable of the Keys, they reviewed the history of typewriters and explained that QWERTY’s emergence was little more than the normal functioning of the market.15 They contended that the QWERTY keyboard was simply the best of the various designs, and that it defeated its competitors the old-fashioned way: by being the superior product. Whether their historical analysis is right or wrong, their apparent need to attack both the example and its underlying ideas suggests that far from being obscure, network economics was well on its way into the mainstream by 1990.
So why did Waldrop choose to open his book with Arthur’s story? And why did I choose to incorporate his opening into my own introduction of network economics? As to Waldrop’s motivation, I can only guess. His book described some cutting-edge interdisciplinary work underway at the Santa Fe Institute—exciting work that unified seemingly anomalous observations in economics, physics, biology, politics, and a number of other fields. He described the scientists who brought that work together to create the emerging science of complexity. Arthur was the key economist in that group. I can only surmise that these concerns guided Waldrop to push Arthur beyond the temple walls and into his opening chapter. Then again, many decisions appear to be rational in hindsight when they were, in fact, arbitrary.
My own decision was entirely utilitarian. Waldrop had already performed the heavy lifting. He had laid out a wonderful story about an early thinker of network economics. I adopted his opening because it was accessible, and credited Arthur with these ideas because Waldrop had already given him that credit. In the words of one early network theorist: “For unto every one that hath shall be given, and he shall have abundance: but from him that hath not shall be taken away even that which he hath.”16 Stated somewhat more prosaically, my reference to “Arthur’s increasing returns” was positive feedback in action.