Just Enough Governance: Introducing Steward Governance for DASL

When we started working on DASL, it was little more than an experiment. We talked to as many people as we could who had been interested in content addressing but couldn't make it work for their needs. The result was radically focused on interoperability and adoption, on a mission to make content addressing something so basic and simple that you'd adopt it without having to think about it much.

That's not the kind of project that calls for much collective governance. It's: listen, do, see what works, and go back to square one until it either works or fails. Or as philosophers of science put it: fuck around, find out, then do it again.

But things are changing. DASL has grown in usage, in community, and in the number of its specs. A growing share of the spec-writing is done by a wider group of people. At this point, maybe, just maybe, it shouldn't be a secretive coterie calling the shots.

And so we're adapting: we're introducing a governance model for DASL. There's a tendency when people all of a sudden decide to Do Governance: they'll switch into a formal, pseudo-legalistic tone register and put together a process that individually addresses every problem they can think of in ways as over-engineered as they're incoherent. So we thought "hey, how about we don't do that?"

We built DASL's governance in accordance with the approach we've taken across the entire project: it's written in plain English, it's short and doesn't YAGNI (You Aren't Gonna Need It), it only addresses the needs at hand and can be upgraded or complemented later with something more involved if and when needed.

We call it Steward Governance. It's more democratic than having a BDFL but less than, say, a stakeholder cooperative. A small core group looks after the project’s direction. A wider set of editors owns individual specs. The bar for changing a spec grows with its maturity level: experimental specs change at will, more stable ones need wider review and agreement. (The full model is itself, of course, a DASL spec.) Essentially, it turns a secretive coterie into a public-facing coterie, but without creating the procedural overhead associated with more democratic forms. The idea is that this is 1) good enough 2) for now. Good enough because it's a group of people well embedded in the impacted community, such that we know that if we screw up we'll hear about it from the people we hang out with. It keeps the actual labour of governing at very reasonable levels that match the level of disagreement in the community (i.e. very limited). And for now because it's not meant to be a forever solution, just the right option for where DASL is now.

In the long run, as I've written elsewhere, I'm worried about the systematic emergence of oligarchies in peer production projects like DASL. Perhaps paradoxically, that concerns argues against a more complex arrangement than what the community has bandwidth for today. When organisations go for complex governance while they still have a comparatively small community in terms of its engagement in the governance process itself, it creates a system in which the complicated rules become entrenched behind the small priesthood that understands them. Others find it hard to challenge them because they immediately get bombarded with a cloud of references to process and precedent which they can't readily challenge.

If — when! — DASL becomes big enough that it calls for more complex governance, then we'll cross that bridge then, with gusto and terrible acronyms. If it so happens that, when that time comes, we don't make the changes, then you lot know where we live and something tells me that you won't be shy in letting us know what you think.

In the meantime, I know I speak for my friendly neighbourhood stewards in the DASL Core group when I say: bring us your questions, your specs, your concerns and together let's keep DASLing the internet.