Strategic reactiveness // Atomic networks and anti-network effects // Bootstrapping community

March 15, 2022 · reading

📘 The Cold Start Problem by Andrew Chen is exceptional. Strongly recommended for anyone building startups or communities.

First, the startling urgency of the early days of building network effects:

*[Uber’s founder] TK grew visibly frustrated [and] paced around the room again. 

“No, no! Look guys. Our network is collapsing. We need to stop the bleeding . . . now! Let’s do the other stuff and get it on the road map, but let’s get this email out over the weekend.”  

[…] We committed to ship the changes before Monday.*

This sort of “strategic reactiveness” stands in harsh contrast to the conventional startup wisdom of focus, focus. And also the convention wisdom of being proactive, not reactive.

But it makes sense given Chen’s observation that until a network has reached a stable size, it will naturally backslide into a ghost town:

When your primary (and arguably only) goal is to achieve a stable and self-sustaining network, then remaining ready to react immediately to any network disruptions is a rare case where you are being proactive by being reactive.

Chen points out that the strength of your network effect is measurable via the user experience. When most Uber users are able to get a cab within 5 minutes, the network is working. When it takes more than 15 minutes, both riders and drivers become frustrated and the network is at risk of backsliding to zero. (Hence TK’s urgency above.)

If users don’t find who or what they want, they’ll churn. This leads to a self-reinforcing destructive loop. 

[…] I call these “anti-network effects.”

In the context of an outcome-oriented community (OOC), although network effects don’t matter for everything, they absolutely do matter for: member introductions/discussions and many events/rituals.

I’d put the backsliding number for discussion at something like getting a good response within 18-24 hours; if it takes much longer than that, you stop coming back. And I’d put the “healthy, stable network” indicator at somewhere like answers in >4 hours (and >5 people at most events).

(Which means that, in the early days, community conversation can be concierged by checking the forum at the start and end of your workday and responding to anything that hasn’t already been convincingly engaged with.)

To grow something stable, Chen suggests thinking of a “network of networks.” For Uber, each city (and sometimes neighborhood) can be stable (or not) independent of the rest of the Uber network. For Slack, a team of just three active users is stable indefinitely, regardless of what’s happening with Slack in other businesses. “Atomic networks,” Chen calls them.

I’ve noticed this with my community’s live Writing Accountability Groups, where the “regulars” at each group have grown into a stronger, tighter, more resilient mini-network within the overall group. (And I think that even if the overall group collapsed, those mini-networks could still survive.)

Two subtle takeaways from the above:

First, as Chen emphasizes, you can grow a network without risk of backsliding by intentionally adding one atomic network at a time. In the case of my authors community, that would mean starting with the writing accountability groups. Under our current setup, people join the overall network and then (hopefully) trickle down into a smaller atomic network within it. Under atom-first model, we would create a new writing group (for example, at a new timezone) and then hustle it hard until it’s reached the point of stability (i.e., >5 attendees each week).

Second, you can reduce your own urgency by picking (or designing) smaller atomic networks. Slack’s atomic networks (of three coworkers) are much less stressful than Uber’s atomic networks (of a neighborhood).

The smaller your atomic network, the easier it will be to build, especially if you’re bootstrapping.

(All quotes are from 📘 The Cold Start Problem by Andrew Chen.)

← all writing