The Recombination of Labor

A common theme amongst old-school scholars is to claim that the internet is making us dumber by focusing on shallow treatments of subjects. Of course, these people often are the same ones who enjoy writing long, boring, non-fiction books and want you to like them (here, the pronoun ‘them’ left intentionally ambiguous). But unless you’re planning a career at a political think-tank, reading a single blog post on neo-libertarian-socialist-relativism constitutes enough time spent on the subject before moving on to something else. The kind of dissemination of information the internet encourages is good for sharing basic ideas, but is not so good for sharing the deep expertise necessary to really apply state-of-the-art thinking to concrete problems. So don’t go overthrowing the government and installing a neo-libertarian-socialist-relativist regime after reading just one blog post.

Twitter is the extreme example of a communication channel for summarized information; Twitter forces writers to essentially condense their entire point to the length of a single sentence. Even if these paragraphettes were all extremely insightful, readers will reach a point where learning faster will only be possible with a greater fraction of their lives spent reading at a computer screen. We’re rapidly approaching this point today, where our natural language communication pipes are saturated, and the only way to get more information into our brains is to keep the floodgates open longer each day.

I have a theory: one reason so much investment goes into this kind of technology is that it resonates with the people who control the purse strings in the technology industry. Senior managers and venture capitalists look around and say, “I spend my days telling people to ‘bottom line it’ for me, and I’m super important, so obviously we should invest in technologies that help people be like me!”

But there’s a danger in organizing the world around the work patterns of these generalists that decide what to do but don’t actually do anything themselves. At its worst, an unstructured, constant flow of high-level information turns life into a poorly run meeting. On the internet, there is no agenda, no specifics are ever addressed, everyone gets to kick in their two cents on every subject even tangentially related to the topic, no conclusions are reached, and it’s never possible to really end a discussion as long as a single person wants to add one more point.

At my first startup after grad school we got a new gray-haired senior manager and we were discussing how to keep sales and engineering informed of what the other was doing. Several times, he returned to the point that “you can’t over-communicate.” Well, it’s 2011 and the technology industry is once again on the verge of achieving the impossible.

Unfortunately, the kinds of internet-enabled communication technologies developed over the last 20 years aren’t all that great for developing specialized expertise. Expertise is most often attained via mechanisms like advanced education, apprenticeships, and years of experience. Internet technology that lets people type in a text box and show it on someone else’s screen isn’t a lot of help here. In a way, this is just because natural language, which is easily shared on the internet, can only do so much to teach a person how to really do something. At some point it’s necessary, and much simpler, to advance a person’s expertise by having them use it on real-world problems rather than reading additional texts on the subject.

So, if technology that increases the breadth of knowledge is suffering from diminishing returns, what would technology that encourages greater specialization get us?

Free-market economic theory actually has a lot to say about specialization. A primary driver of economic progress is the increasing division of labor, which is another way of describing specialization of knowledge. First, a shopkeeper decides to specialize in making shoes, and before long he has someone working for him who is an expert in laces and someone else who is a master at making soles. As specialization increases, both the quality of the shoes and the quantity of shoes per worker goes up.

Actually, most technology is built for specialized needs. Consider those newfangled laser levels that carpenters can use to draw a straight line on a wall. The technology is probably as sophisticated as what’s needed to send an email (I mean, come on, household lasers!). But email is much more widely applicable. Tools for specialists get built frequently in all different areas of the economy, and cause significant benefits in aggregate. Technology advancements that are so widely useful that they move the needle of world productivity by themselves are quite rare. Industrial mechanization, electricity, the internet: these things don’t come about every day, but when they do they change the world.

What’s important is that there still can be technologies with wide applicability that help encourage specialization. Given that we’ve shifted to an information-based economy, the principle that currently can be exploited is that most of the benefits of specialization occur when people with differing expertise work on designing a single product or service. It’s often the integration of expertise that causes the greatest benefits. Unfortunately, it still requires lots of brute-force effort to make components “fit together.” Consider how much benefit comes from standardization efforts. From web browsers that understand a common form of HTML to engines that burn a common form of gasoline, a standard interface between components can free actors on each side of the interface to innovate in isolation while still creating pieces that fit together. Standardization is a technique, not a technology, but the important insight is that improving the way specialists work together is ripe for a major technology advance.

In the end, we have technologies for sharing lots of general knowledge due to the internet, and we build lots of specialized technologies for individual areas of expertise. We don’t have a general purpose technology to help apply diverse expertise to create new products and services. This integration of expertise is a fundamental part of the design process that creates better and cheaper products and drives the economy forward. Standardization of how components fit and work together has been an important principle in allowing the improvement of products that are more complex than a single person can understand. However, much more of the design process can be automated by technology that better addresses how expert knowledge is combined into new products and services. I call such an automation a Capitalist Algorithm.

In a future post, I’ll explain how Capitalist Algorithms work, why they are feasible thanks to modern computing capabilities, and how they can create economic value using far greater amounts of specialization in the workforce than current design techniques.

Comments are currently closed.