Automating the network away

Internet networks used to render brick-and-mortar businesses obsolete. Arguably, automation businesses may soon render networks obsolete.

January 26, 2016
Jonathan Libov

As the term “social media” has become somewhat quaint, it’s easy to forget that Facebook and the like actually represent the networked form of Media businesses. Whereas Old Media required lots and lots of editors, writers, photographers and videographers, Facebook leveraged the network to create and curate content. In doing so they grew to a scale that far exceeded all the Old Media companies, and could do so with a fraction of the staff.

This is, of course, why networks were so attractive to investors. Offloading work to a network of people you don’t pay salaries to is a hell of a way to be capital efficient.

What happens, though, when machines can actually produce the media? In the realm of sports, for example, one could cobble together Automated Insights to write game recaps, WSC Sports to pick out and produce highlight reels. In music, computers can already autonomously compose background music; how long will it be until machines can produce popular songs?

Automation is happening all around us. It’s happening for things that seem relatively perfunctory, like human resources and lead generation. But it’s also happening for things that not too long ago seemed like only a human could do: Yousician is automating away music teachers, and Lemonaid Health is automating away doctors, one of the most erudite professions on Earth.

It’s not as if HR or sales staff, music teachers or doctors will be out of work anytime soon, but it does sort of change how we think about delivering services. Whereas the old way of doing things was to build a network that outsourced the work to the people in the network, the new way might be to put a machine at the top of the funnel and then farm out the work that the machine can’t handle to the network.

As a venture capitalist at a firm that has always invested in networks, this presents something of a challenge: If the path to delivering services goes through things which are not networks at first, you need to figure out who’s going to execute best on the automation that sits at the top of the funnel. This looks much more like investing in gaming companies—your job there (and pardon my oversimplification, as I’ve never been a part of investing in games—is to figure out who’s likely to execute a game that people really like.

Moreover, it’s not even clear if or when these automators will ever build networks. None of the automators I listed above appear to have built one, and perhaps for good reason: Much as the new networks looked at the incumbent, non-networks and thought, “Ha! All that work you have to pay salaries for, I can offload to my network!”, the automators might look over at their networked counterparts and think, “Ha! All that work that you have to do building a network, I can do with the two people on staff who can spin up a new instance of automation!”

This is sort of a good problem to have as an analyst at a venture capital firm. At least until a tool that automates your work adds a network which makes your job obsolete.