The Economy of Cities, chapter 2: where does this all come from, anyway?
In this chapter, Jacobs presents a number of examples – some based on the fictional example city but most from modern industrial economies – of how old work creates new work, and discusses some ways that the structure of businesses can encourage or discourage this process. In short: if you want your economy to stay energized and develop in new directions, big companies are your enemy.
The core dynamic is that every task someone is working on persistently will produce byproducts, inefficiencies, and useful subtasks, some of which can be developed into entirely new useful tasks. The inventor of the bra was a dressmaker who wanted to improve the way her dresses fit. 3M started out trying to make sandpaper from their sand mining business, and in the process of making it, they created new adhesives and then a wide variety of novel and useful types of tape.
Jacobs’s insight is that the further work is subdivided, the more tasks there are where someone is spending their time focused on that task and how it could be improved; the inputs changed, the outputs made better, or the byproducts and side effects used for other purposes. Specialization leads to more opportunities, and also to more domain knowledge; someone who just sorts sand might have the idea to turn the different grains into sandpaper where someone who oversaw the entire digging, crushing, and sorting process might not have the same focused expertise.
(Yes, it seems a little silly to talk about sand-sorting expertise. But hey, you could be a connoisseur of Joe Biden eating sandwiches, why not sand?)
This ties back to Jacobs’s earlier work in zoning briefly; she points out that when new work splinters off old work, it rarely will be confined to the same industrial category. Heavy industrial manufacturing might lead to ideas about a better lubricant (chemical manufacturing) or retail sales can lead to making more of the product you sell (light manufacturing). This is also a problem with guild systems and credentialing; there is a historical example where silver-plated knives and forks were created in the London guild of knife-makers and fought bitterly by the goldsmiths, who had historic rights to all silversmithing. New work arises unpredictably, so regulating what companies are allowed to do risks preventing them from exploring new ideas.
Her most famous example, the Japanese bicycle industry, also shows up here; surprisingly, it goes along with the origins of Ford Motor Company. Both manufacturing processes started with an ecosystem of parts suppliers; in Japan this was bicycle repair shops, for Ford it was the machine shops of Detroit. They both then began building new vehicles from the parts purchased elsewhere, then gradually moved their suppliers to dedicated manufacturing for their own products; Ford expanded his facility and hired more workers to create parts in-house, while in Japan it became the zaibatsu system of vertical monopoly, where the former repair shops were now dedicated suppliers of specific parts for the bicycle manufacturers.
(In another example of new work drifting far from old: A Japanese playing card manufacturer eventually moved into electronic games, and then arcade games and home video game consoles. They are now known as the creators of Mario.)
Critically, large companies are structurally opposed to this process. Even when operations are split into small teams, for the sake of avoiding chaos a large company can’t let each of them investigate every fruitful-looking side project they see. For one thing, this would give any marketing or branding team headaches; for another, bookkeeping would be a nightmare. For certain, most of the benefits of scale would be lost.
At one point the University of California system attempted to engage in this, letting all of its departments and professors contract directly with different agencies and companies to do whatever research they needed or whatever specific tasks they might want that an academic lab might do. (Dr. Clark Kerr’s ‘multi-versity’ concept.) Predictably, it was a mess, and eventually abandoned. Before it was, one suggestion was that they farm out the independent research teams to separate institutions (like Caltech does with the JPL or Stanford did with the Stanford Research Institute). This is what is known in the UK business world as a ‘breakaway’.
Jacobs is very fond of breakaways as a means for large companies to create productive work, though this does not seem to be particularly fruitful in terms of implications. Clearly she would be very much in favor of Stanford and the SRI producing Silicon Valley and similar phenomena around other universities, but as far as finding benefits for the company that allows the breakaway or observations on what environments encourage them, she is silent. It suggests to me that if you want to have a developing country benefit from importing a factory, the right thing to do is to insist on it training its employees in ways that encourage and allow breakaways, but there are a lot of natural obstacles for that going anywhere.
She has further bad things to say about bigness; it is a fairly easy observation that large companies grow now, as then, almost exclusively by buying smaller companies and merging with other large ones; they do not create new markets or products. (Since she wrote, expanding existing brands into foreign markets has become popular, probably because it is much easier for a brand to carry perceived value after decades of global television and a decade of global internet; her point remains.) R&D divisions, as is increasingly recognized since then, are not only artificial but also frequently ineffective or even counterproductive.
One example of what it takes for a large organization to create new work, which serves as a rule-proving exception, is IBM. (A little less inspiring these days, sure, but still relevant.) It grew when small by buying a patent on electric typewriters existing companies didn’t want, and then stagnated. When it added an early computer to its lineup, it had to restructure radically to provide training, programmers, and places to rent mainframe time, none of which matched its original skillset. That done, it grew rapidly and then stagnated, staying basically the same through to the present day and losing ground to newer innovators.
There are two main conclusions here:
One, divisions of labor is not inherently a good thing. It is useful to some degree for creating jumping-off points for new work, and specializing can make an existing process more efficient, but it’s not good in its own right, and splitting it up for its own sake is more likely to be harmful than helpful.
Two, growth is an explore/exploit tradeoff. It is useful to grow to do old work faster, but inhibits the discovery and creation of new work and new techniques.
Also, one semi-prescient direct quote:
When large organizations actively try to add new goods or services to those they already produce, they create, like special reproductive organs, special divisions of labor for that purpose called research and development departments. These are substitutes, or surrogates, for the great body of sterile divisions of labor. But by definition, the parent work on which R & D can build, in comparison with the organization’s total work, is exceedingly limited. And even within these limitations, the new work that the researchers find it logical to develop frequently turns out to be irrelevant or hostile to the interests of the organization as a whole. Hence we have the paradox of useful inventions neglected by the very organizations that have “taken the trouble” to develop them.