To the Limitations of Network Production and the Convergence of Passions

This week, my friend Mitch sent me a copy of the excellent paper, Working Wikily, by authors at the Social Innovation group at the Standford Business School. I highly recommend it to anyone interested in issues of collaborative production or social action generally. As I said to him, these guys definitely get it. They do a decent job selling the potential of network production, but the real thing I'd say they do, which a lot of folks who get excited about this stuff forget to, is describe the limitations. Network production is incredibly powerful in tons and tons of situations -- the majority of the time I'd say exponentially more so than command and control structures. But almost always, network production needs to be directed by (or at least come in conjunction with) more traditional forms of organization.

The paper reminded me of a long, interesting discussion I had with another friend earlier in the week about a dinner party I hosted the week before. I'd spent a good bit of effort at the time trying to decide whether we should do it potluck style and everybody bring something (ie network production), or whether I should just collect $15 from everybody and do it myself. It ended up being a bit of a hybrid, and basically worked out, but I realized in the future, I'd definitely opt for command and control. The reason for this is that network production has huge advantages. Often it can rally resources far in excess of what any centralized effort could produce, tapping the latent pool of distributed investments in both physical and human capital members of the network have already made that are simply sitting around idle. In so doing, it often allocates production far more efficiently, optimizing for the different talents and interests of the various members of the network. And in a case like this, it would also optimize for the diminishing utility of time (20 mins of labor would have trivial impact for everyone, whereas the 3 hours it would take for me to do it myself would clearly have an impact). Network production meanwhile generally produces much wider diversity of outputs, allowing people to find things more customized to their tastes, even if the average output were of lower quality than the more regulated, more limited output of a command and control structure.

But the problem is that for these parties, I need to be able to offer quality of service (QoS) -- otherwise I can't justify the risk of inviting certain people, friends who this is perhaps the first time I'm hanging with and therefor need to give a good show for. Networks aren't great at this, and can generally overcome the problem only either with strong cultural pressures (eg the original Northwest Native American potlucks) or with tremendous redundancy (eg the web, where 95% of things are crap, but there's such a huge abundance that the remaining 5% is more than enough to compete with centrally organized products). Neither of these conditions would have been met in the case of the party, as you'd have needed near 100% participation and it would have been within an ad hoc group amongst which there was limited social pressure (I'd sent out invites only the night prior, and it was a mix of people from different places, such that most people didn't know each other). Meanwhile, there are always efficiencies of centralization -- if it takes 15 people 20 mins each, that's 5 hrs total, vs the 3 hrs it took me alone given I could make just 1 trip to Eately, 1 trip to the wine store, 1 trip to get plates/cups etc. And the benefits of distributed production need always be weighed against the transaction costs, which for this situation would still be high... I (or a platform) would have to somehow know who's good at what, who lives near what speciality stores, dole out assignments without redundancy, etc. Emerging platforms might facilitate this sort of coordination (why for instance I'm pushing friends to join services like Latitude), but nothing at present I think would bring those transaction costs low enough to make the equation work for the level of quality assurance I was looking for.

Earlier this month, I posted the following on Facebook:

"I've said the thesis explains a lot of my personal prefs. Turns out, actually quite central. The idea is that upfront, fixed investments (in production assets, in employees v contracted services, in internalized data, in who you hang out with on any given night) are only a way to solve the problem of difficult on demand access. As transaction costs lower, these fixed structures become both unnecessary and inefficient."

I love this idea of my two great passions in the world coming together, digital media and good times with friends (which any of you who know me know what that's partially code for ;). When one is lucky enough to have this occur, the two amplify one another. Leveraging the mechanics of distributed production I'm coming increasingly to understand I feel is making my life infinitely better and easier... it's helping me understand and channel the way I work, helping me build networks of people I care about, helping me plan dinner parties. And then it runs in the other direction... I think of that scene in A Beautiful Mind where John Nash discovers non-cooperative equilibria because he and his buddies are deciding how best to chase some girls in bar. It's these situations I think, the ones we really understand because we've lived them, that help make clear some of the fundamental issues, with the added benefit of helping others understand.

And so a toast... to the limitations of network production and the convergence of passions.