There is a small "Japantown" near my house - a cluster of noodle shops, sushi bars, groceries, and Japanese bookstores. Most though not all of the shops are in Porter Exchange, a grand old building that used to be the Cambridge Sears outlet. The building, with its high ceilings and open floor plan, was unable to find a good use for an extended period after Sears moved out, but turns out to be quite well suited to use as a sort of mall. The location also turns out to be excellent: because many of the patrons are public-transit-using students at MIT and other area schools, it benefits from its proximity to the Porter Square stop on the Red Line. The cluster seems to be growing; perhaps one day Porter Exchange will be only the nucleus of a large and distinctive shopping district.

How did that building become available? Sears discovered that a location with poor access to highways and limited parking could not compete with stores in suburban shopping malls. So why did they build it there in the first place? I don't know this for sure, but I would guess that the location was actually a very good one when the building first went up, far enough away from downtown to offer inexpensive land and room to build, but still accessible because of the trolley line that ran up Massachusetts Avenue. And Mass. Ave., as we all know, began as a track for cattle drovers ...

I think Henry Ford once said that history is just one damn thing after another. Maybe not, but explanations of economic location are almost always historical, and the history does tend to have a "one damn thing after another" character. If you try to explain why a particular region is home to a particular industry, you usually end up explaining it largely by describing the sequence of events that caused the industry to be there. A Georgia teenager makes a tufted bedspread as a wedding gift; 60 years later, that event makes Dalton, Georgia America's carpet capital.

The essential driving force behind this historicity is, of course, the tendency of an industry to stay where it has been established, thanks to locational external economies. It remains hard to improve on Marshall's threefold classification of these externalities, as arising from the ability of producers to share specialized providers of inputs; the advantages to both employers and workers of a thick labor market; and localized spillovers of knowledge, especially through personal interaction. They apply just as well to Boston's complex of world-class hospitals as they did to the Sheffield cutlery district that inspired Marshall. Now as then such localized externalities provide a virtuous circle that tends to keep an industrial cluster locked in place once established.

But how do such clusters get established in the first place? You might think that the process is largely random: an accidental event, or a remarkable individual, creates the seed around which an industry grows. Reading Browne and Sass, however, suggests that the process is not so much random as chaotic. That is, the history of New England's changing industrial structure often seems to be a story of events that were more or less deterministic, but whose outcome nonetheless could not have been predicted.

Consider, for example, the Boston mutual fund industry. Its existence - and the resulting ability of metropolitan Boston to benefit disproportionately from 90s stock market mania - is clearly not accidental: Browne and Sass suggest that the financing needs of high-technology startups built up a special sort of financial competence in the area. The high-technology industries of Massachusetts, equally clearly, did not emerge by accident; they were there largely as a legacy of World War II and Cold War research. And this research was concentrated in Cambridge because of the presence of universities, which was ultimately the product of the need of Puritans for someplace to train preachers. At each stage there was a clear reason for what happened; but nobody could possibly have predicted the sequence.

The process seems to work mainly as follows. Start with a region that has a particular industrial base, itself the product of a long historical evolution. If the environment were unchanging, that industrial base would tend to persist; but things do change. Most important, probably, is the rise of new technologies that make old advantages irrelevant but offer new opportunities. However, the past is not completely irrelevant: the special characteristics of regions, the consequences of their old industrial mix, determine which new industries find them congenial soil. Machine shops set up to serve textile mills can turn to the production of components for aircraft engines; or - an example I didn't find in Browne and Sass, but one mentioned to me by computer industry acquaintances - experience gained from time-shared minicomputers, having become temporarily devalued with the rise of PCs, turns out to be relevant again with the popularity of local area networks. In other words, the regional industry structure at time t determines the structure at time t+n in a non-random way, but it does so through quirky linkages that nobody could have foreseen.

Because the linkages are so hard to foresee, of course, the rising and falling fortunes of regions inevitably involve a large degree of luck (even though they may not involve very much true randomness - chaos theory is funny that way). And over the past 50 years New England has clearly gotten lucky. As Browne and Sass point out, at the end of World War II the region's industrial base was clearly destined for decline: its textile and shoe industries could no longer compete with lower-wage producers in the South. So one might reasonably have expected New England to become a permanently depressed area. Instead, however, the legacies of the region's past turned out to give it an unexpected advantage in snaring some of the industries of the future. Some of this involved conventionally Marshallian linkages - the support network that had once served the textile industry could be transferred over to new uses - but much of it also involved novel linkages, especially those that ran through the region's universities. Again, there was not really that much true randomness; taking World War II and the Cold War as given, it was in retrospect entirely natural that MIT and some lesser universities up the river would have played a crucial role in developing military technologies and their civilian spinoffs, and that this would give rise to a new high-technology economic base. But half a century ago nobody realized just how economically significant the kinds of things done at the Radiation Lab would turn out to be.

The interesting question looking forward is, of course, whether New England will stay lucky. This turns out to be more than a question about whether New England will retain its lead in the sectors it now dominates, or whether it will manage to seize enough new sectors to replace old advantages as they erode. For even a casual look at the current New England economy suggests that this time things are somewhat different - a difference documented at fascinating length by Browne and Sass.

Put it this way: what exactly does New England do for a living these days? As recently as the 1980s, it was not that hard to give a first-pass answer: the dominant "export" sectors, those that drove the region's growth, were minicomputers plus a few other high-technology manufacturing industries. Digital Equipment was both the largest area firm and the emblem of what the new New England was all about. But now we have what Browne and Sass call an "amorphous economy", one in which it is hard to find any focus. New England is clearly doing very well selling something; but what?

I have to admit that I am myself of two minds about the putative amorphousness of the economy. One simple interpretation is that the fuzziness is not in the real economy but in our statistics. As measured, New England seems to be virtually indistinguishable from the national average employment structure; well, maybe that is simply because industrial-era statistics cannot keep up with a service economy. It is easy to think of examples. Browne and Sass themselves point to the rapid growth of "business services" as a statistical illusion: in large part it reflects the growing preference of companies for outsourcing of previously internal activities, and even for temporary over permanent employees. Yet all business services are lumped together, so that firms selling technical support for software companies in Cambridge end up in the same category as those selling technical support for oil exploration in Houston. Or consider the health care industry: the world-famous brain surgeon at Mass General, who attracts clients from all over the globe, gets lumped in with general practitioners; the result is that even though Boston probably has an almost uniquely large sector exporting medical services, it is lumped in with the nontradeable general health care sector, and the region's distinctiveness is obscured.

And yet it is hard to avoid the sense that the economy may have become truly more amorphous as well. To travel the cities of the United States today is to be struck by the sameness of their working worlds: where once Chicago was a city of hog butchers, Pittsburgh one of steel workers, and so on, today every metropolitan area is populated largely by office workers who look the same and do the same kind of jobs. Some labor economists I know actually suggest that the surprising recent ability of the U.S. economy to achieve low unemployment without inflation is partly the result of a homogenization of job requirements. Once upon a time, they suggest, being a hog butcher or steelworker was a special skill, not easily acquired; so hog butchers could push up wages even in the face of general unemployment, but would themselves become unemployed if the demand for their services declined. Now, however, many jobs require the same set of skills (literacy and basic computer knowledge) so that firms find it easy to replace workers (reducing the market power of current employees) and workers find it easy to switch jobs (reducing frictional unemployment).

But if regions really are all pretty much the same these days, why aren't they more self-sufficient? In particular, why are there hundreds of thousands of businesspeople taking plane flights every day, to make presentations or attend meetings somewhere else in this apparently homogeneous economic landscape? The people they are going to see must be doing something different, or the business could be concluded locally.

At this point I believe one could tell two quite different stories. One is that despite all that I have said on behalf of amorphousness, in the end a region does have a set of core competences that give it a hard-to-measure but real distinctiveness. Most residents of Southern California are not beautiful, but nonetheless the core of beautiful people gives the region a special competence in the entertainment field; most people in greater New York are not unusually avaricious, but the core of superlatively avaricious people gives it a special competence in deal-making; most people in Greater Boston are not brilliant, but the core of brilliant people gives it a special competence in the knowledge industries. (New England, as Browne and Sass point out, has a higher share of college-educated workers than other regions. What really stands out, however, is the overrepresentation of the region in the tallies of top-ranked universities - by the US News and World Report rankings, New England, with only about 3 percent of the nation's population, has four of the top five schools, eight of the top 25). And some rarefied version of the Marshallian trinity sustains these clusters.

Alternatively, regions really are pretty much the same, and all the business between them is a matter of micro-level, almost accidental advantages and connections. An Atlanta firm needs some specialized service; it just so happens that the firm that can supply that service is in Boston, when it could just as easily have been in Atlanta. Or perhaps there even is a firm with that competence in Atlanta, but it is easier to go with the firm you know about a couple of hours' plane flight away than to search for a local supplier.

If we believe my first story, then the way to think about New England's future is to try to figure out what the true common denominator of its success really is, then ask about the prospects for that newly defined "industry". If we believe the second, regional fortunes will henceforth be determined by some sort of generalized economic quality, not by the specifics of what they do and know. And we really don't know which is true.

But I guess I would argue on general grounds that it is unlikely that the historical importance of regional specialization has truly vanished. Probably New England is indeed still driven by a few self-sustaining clusters of activities, which do not appear in our data only because we collect the data badly. Indeed, one might guess that the area's investment activities, its remaining computer industry, its exportable business services and so on all do reflect a common set of Marshallian external economies.

In that case, however, the question is whether the region's advantages will persist.

Well, it's chaotic - so whatever we say is almost sure to be wrong. Still, reading Browne and Sass did make me feel a bit uneasy about the future value of my house, because it struck me that many of New England's recent successes depend on the costs of transportation and communication being in a sort of Goldilocks range - not too high, but not too low, either.

Let me explain. It is a familiar point from the "new economic geography" that the impact of transportation costs on agglomeration tends to have an inverted U shape. At very high transport costs, there cannot be agglomeration: the world consists of self-sufficient peasants. At very low transport and communication costs, there is little incentive for agglomeration: necessary inputs can be delivered to wherever the factor costs are lowest. (This is what happened to the textile industry: improved transportation made it unnecessary for mills to remain in the established centers, and allowed them to move to lower-wage locations). It is only in an intermediate range that agglomeration is both possible and necessary.

Now what struck me about many of the examples of recent New England successes is that they depend on transport or communication costs having fallen enough that previously nontraded activities become possible export sectors, but not so low that localized external economies fade away. Thus specialized computer services can now be delivered at long distance via email or a few plane flights, but the personalized interaction that gives New England firms their expertise still depends on face-to-face contact. People can fly to Boston for specialized medical procedures where they would once have had to rely on whoever their local hospital could provide, but the core of highly skilled doctors and staff must be concentrated in a small geographical area. And the particular state of modern transportation and communication technology seems to be particularly congenial for the kinds of things New England is currently good at.

Suppose, however, that videoconferencing becomes cheap and routine, so that a working cluster of experts can form without any need for physical contact. Might the computer consulting firm of the future have only a tiny central office, consisting mainly of a coalition of geographically dispersed people who meet around virtual conference tables now and then? And if so, wouldn't it be likely that many of the experts would choose to live somewhere cheaper and/or warmer than Boston? Similarly, the time is clearly not too distant when remote surgeons will guide operations - perhaps even operate robot machinery - from long distance. In that case, why can't the great surgeon do it all from his seaside estate, with no need to deal with Boston traffic?

As you might guess, this observation is partly based on personal experience. Over the past 20 years I have watched the physical location of an academic researcher steadily diminish in importance. One's important colleagues in a specialty are increasingly likely to be scattered across the world, rather than concentrated in one's own universities: Cambridge has far more experts in, to take a non-random example, international finance than anyplace else on earth, but most of the IF community nonetheless is someplace else; intellectual interaction takes place mainly at conferences, via email, or - especially since the onset of the Asian crisis - via Internet postings. When I recently spent a couple of weeks in a rented house several hours' flight from Cambridge, I found it possible to do most of what I normally do quite well - and I suspect that most of the people who correspond with me via email, or look for my work in cyberspace, had no idea I was away. And perhaps the same is true of many "knowledge workers", which means that the kind of cluster of such workers New England now offers will soon dissolve.

Or maybe not. For the one overwhelming lesson of New England's economic history is that while each successive stage reflects forces that are obvious in retrospect, attempts at prediction nearly always get it completely wrong.