2013-10-11

I like thinking about technological disruptions that take place over really long periods of time. This is because the older a technology being disrupted, the more profound the social impact. In my disruption of bronze post, I speculated about one that probably took a few thousand years (iron disrupting bronze) and made spaghetti of the prevailing world order.

I just thought of a potential example that spans 10,000+ years: as a technology, computing disrupts natural language in the thinking and communications market.  That would make computing the mother of all disruptions in terms of time scales involved.  Well, maybe electricity disrupting fire in the heat and light markets is a contender too. Here is the disruption, speculatively mapped out in the form of the familiar intersecting-S-curves visualization used in disruption analysis.

 



Here’s my reasoning. I am convinced it hangs together.

Soft versus Hard Technologies

The argument hinges on the idea that electronic computing is only the second truly distinct soft technology invented by humans, the first being natural language (within which I include mathematics and other general symbolic representation systems used by humans and organizations to think and communicate).

By soft technology, I mean a technology that cannot by itself do anything to the world of atoms, but can be realized within the world of atoms in many ways. So natural language can be carved on rock or written or paper. Software can be stored on magnetic disks or punch cards.

Soft technologies can only affect the world through hard technologies, by controlling flows of energy.

I tried hard to think of one, but I couldn’t think of a single soft technology besides language and computing. Music and platonic geometry (in the sense of the language of architecture and other forms of visual design) are possible distinct candidates, but they seem like cousins of natural language.

Now, here’s the thing about disruption and the soft/hard distinction: like disrupts like. So only soft technologies can disrupt other soft technologies. Only hard technologies can disrupt other hard technologies.

Of course, there are no pure soft or hard technologies. Soft technologies need a hard substrate (neurons, paper, vacuum tubes, silicon) to actually function.  Hard technologies, to be more than natural raw materials, need to embody a design, a construct within a soft technology, if only by accident (as in the case of a rock used by a caveman without modification to kill a rabbit).

But the idea that like disrupts like works even for realistic, non-pure technologies. A change that is purely at the level of hard technologies cannot disrupt soft technologies, and vice versa.

So email disrupting paper mail was a case of typing skills (soft) plus keyboards, chips, cables and CRT screens (hard) disrupting hand-writing skills (soft) plus paper, pens, mail sacks and mail vans (hard). Hand-writing and typing both represent points within the evolution of natural language, so this would be a case of self-disruption within a single soft technology.

There are some subtleties here involving cases where the evolution of hard and soft technologies within an artifact are not synchronized, and one lags the other, but loosely speaking, the like-disrupts-like proposition seems to work.

This means the only true candidate for “what does computing disrupt?” is language.

The Disruption Pattern

In Clayton Christensen’s original notion of a disruption, a new product offers an under-served marginal market a little more by offering  the over-served core market a lot less. It thereby manages to carve out a niche based on a much simpler product. The incumbent retreats upmarket to the high-end core. If the disruption has broader potential, it may eventually marginalize or eliminate the incumbent altogether.

Viewed as a soft technology, human language serves many needs, from high-end to low-end:

Poetry

Postmodern works that only French speakers with 200+ IQ  can understand

Logic and computation (with a specialized vocabulary expansion)

Routine news

Corporate contracts

Legal briefs

Everyday financial transactions

Instructions in instruction manuals

Over time, human language, especially languages that have been co-evolving rapidly with modernity, such as English and French, evolve in complexity and are driven by the most complex use cases.

But it remains clumsy for the relatively simple cases, which are increasingly marginalized as the demands of the most sophisticated customers (poets, postmodern scholars, lawyers, stand-up comics, political orators and mathematicians say) drive the evolution of the technology.

It is not exactly clear what either language or computing is, but it is clear that both serve very similar needs in thinking and communication for autonomous agents.  The difference is that computing can as yet only handle the simpler cases covered by natural language. But it serves those cases much better than natural language.

To apply Christensen’s definition, we also need to identify the core and marginal markets in question. The answer is surprisingly simple: the over-served core market is humans, especially the highly civilized ones. The under-served marginal market is machines and organizations ( the two other entity types in our world for which agency can plausibly be claimed).

This is the only breakdown that makes sense.

For the first 150 years after the industrial revolution, machines and organizations had to make do with human language (and employ human translators) for their thinking and communication needs.

Now they’ve found a technology that serves them better, and they’re switching.

Implications

There are three major implications to treating computing as a disruption of language that carves away large parts of the machine and organizational markets.

Three-agent economics: First, a lot of technological, social and economic analysis in the future will only hang together coherently if we treat machines conceptually as economic agents (i.e., they can represent “markets”). If you were troubled by the economically inevitable idea that organizations are people too in a legal sense, things are about to get a lot messier. Once machines are able to own property autonomously and spend money, like organizations, the fun will begin.

Competition among agent types: Second, this has implications for humans in an age of mixed human/organizations/smart-machine populations: there is no reason the three populations of economic agents have to stick to their historical roles. There will be machines that do high-end things comparable to poetry, and humans who do low-end things comparable to CPUs chattering to each other within a data-center. And there will be organizations that do things we cannot imagine.

The rise of uber-organizations: Third, the society of organizations, the most complex agent species in our world, is going to get really weird, since they can be composed in very flexible ways using machines and humans. Amazon is an example of this sort of uber-organization (by analogy to ubermensch in the Nietzchean sense) that runs on two soft technologies employed in a powerful combination.

So now that we have two soft technologies (language and computing) and three kinds of economic agents (organizations, humans, smart machines) thinking and communicating in our world, things are going to get very messy indeed.  If you thought things were confusing enough with B2B, B2C, C2B and C2C markets, you can add the combinatorics of machine markets in there. So soon, we’ll inhabit a world with five additional types of markets: B2M, M2B, C2M, M2C, M2M. Your refrigerator might buy its own replacement compressor. Your vacuum might rent an attachment from the neighbor’s vacuum without telling you. Your friendly neighborhood snack machine might own itself and literally sell you a can of coke (M2C) and order more when it runs out from Coca-Cola (B2M).

But wait, there’s more.

Even though computing is disrupting language, the two soft technologies are also already blending in complex ways. Human-readable source code employs both soft technologies for instance. Amazon’s Mechanical Turk is a meta soft technology, whose substrate is a mix of computers and humans. We should think of HITs (Human Intelligence Tasks) as a hybrid soft technology that blends natural language and computing. The same holds for the soft technologies used to run crowdsourcing models, flash mobs and the like.

I’ve changed my mind. Electricity disrupting fire is not a contender. Computing disrupting language really is the mother of all disruptions.

Share/Bookmark

Show more