The past few weeks, in between watching Twitter fall apart, I’ve been paying very close attention to how various communities are handling the whole situation. The differences are ridiculously stark, and I think it makes a very interesting case study about communities, product adoption, and change management.
For context, I’ve been around on the internet for a solid 25 years now, and I’ve seen a lot of online communities across many platforms grow, flourish, and then ultimately die for various reasons. It stretches from AOL chatrooms, to forums on websites, subreddits, to MMOs, and the depths of IRC. Most of those, when faced with a massively disruptive event like a forum closure, or a moderator dispute that breaks the community, actually do not survive the chaos. There’s often half-hearted efforts to rally people of a certain faction to move to another platform — and the vast majority of them fail miserably.
There’s a lot of stuff that goes into these failures to evacuate.
Probably the biggest is that the group couldn’t overcome the switching costs. If your favorite online MMORPG was shutting down, how many friends and guild members are going to collectively find a new place to hang out together when the thing that originally brought them together, the game itself, is going away? Who’s going to change deep ingrained habits and log into a different website forum to talk about stuff with a smaller group of familiar people? Is that relationship really worth the energy to change behavior?
For many, it’s not.
When building product, convincing users to switch from a competitor’s product to your own is usually a very high bar. You not only have to offer similar enough features so that users are even able to switch to begin with, but you also have to offer incentives on top of mere feature parity to overcome the significant changing cost of a user having to move their stuff over and learn your product. If learning to use chopsticks merely let you shove food into your mouth as a fork did, with no differentiating benefits, why would you spend time being hungry learning?
So when you see people on Twitter suggesting people flee to Tumblr, Mastodon, or something else, products that are clearly not even the same and are generally less usable than Twitter itself, you can see why most people don’t switch. The cost/benefit equation just doesn’t work out for them. It’s the default response that I’ve seen countless times.
Tightly connected groups seem to do better
But what amazes me this time is that this equation actually worked out for certain groups! They collectively weighed the cost/benefit analysis and decided to switch. I’ve only seen this succeed in a handful of other situations.
So who are these strange folk that seem to have moved in a big cluster? I’ve seen large groups of academics across many disciplines and data scientists. I’ve seen bits and fragments of legal commentary Twitter, journalist Twitter, and people who contribute to open source and developer communities. Meanwhile, in the other communities I hang out in, I see practically no movement at all. Artists, indie game developers, are all just trucking along posting things seemingly as normal. The contrast in tone is pretty striking.
What’s the pattern here? I think it’s primarily a difference in what these various groups are. I’d be willing to bet a lot of money that if we did a network graph analysis of the groups that successfully evacuate versus groups that that aren’t going anywhere, we’d find that they structurally look very different.
Groups that seemed to move over well are groups that talk TO each other. You see this in how data twitter moved over, a few popular people migrated and talked about fleeing to Mastodon, and because everyone talks with everyone else, word spread and it snowballed into critical mass. If you were to draw a graph of Data Twitter, there’s a ton of central figures who frequently talk to all the other central figures on a regular basis. The network graph is extremely well connected with lots of “Strong ties”.
Academic twitter did a very similar thing by curating large lists of accounts from various disciplines to make it easier to find one another on the new platform. Clearly, they valued the connections that existed in the network they had, and spent considerable amounts of energy to preserve them.
Meanwhile, the groups that aren’t moving seem to be much weaker in terms of tie strength. The main mode of communication of these groups more closely resembles talking AT each other than having conversations. It’s everyone sharing work and news in a broadcast model. These accounts typically have larger followings of fans who are there to happily receive information, but I haven’t seen a mass exodus of gamedevs to any particular location (and I’ve been looking). While the individual artists and devs might want to move, they know very well that their audience isn’t so invested in them personally to pay the cost of moving. So they’re hanging on while trying to put together as many backup plans as possible. I’ve seen all sorts of discussion about whether they should move to Tumblr, or even Facebook.
But you can’t overlook urgency
If product adoption were merely as simple as “Groups with this property will adopt our product”, I wouldn’t have any work left to do. Luckily it doesn’t seem that simple. The fact that Twitter is very publicly on fire, for everyone to see, seems to have contributed to evacuation success.
Even if we humans know we MUST pay a switching cost and are willing to do it, we still don’t want to volunteer to pay that cost immediately. It’s so much easier to do the easy thing, use the old platform, and procrastinate on paying the cost. In slower-moving situations, I’ve seen people just procrastinate about moving so long that everyone else leaves, they’re left in a quiet space, and those people just leave the community due to inactivity instead of following the moving group.
Urgency, the threat of the eminent loss of something valuable, needs to exist to give people the kick in the butt needed to switch without procrastinating. If the whole dumpster fire had played out slowly over the course of 6 months, I very much suspect that we wouldn’t have had the pickup rate. Big blazing fires are great!
As an industry quant, this whole situation is forever outside my grasp
Looking over all these attributes and situations needed to create the conditions for a mass platform exodus, the thing that annoys me the most is that as someone who is tasked with measuring and predicting user behavior, this is something that feels like it’ll be forever outside my reach.
Sure, I can use network analyses to describe various subgroups and whether they successfully evacuate or not. Maybe if I’m lucky there’s a regression I can run that relates some features against success. I can do text analysis to try to figure out where they’re going. I can maybe even see if there are tipping points like minimal group sizes needed for success. I’m sure there’s academics studying social networks who will be trying to run similar analyses in the near future if they haven’t already.
Then there’s all the construct validation that would need to be done. How would I measure “urgency”? How approximate the cost/benefit analysis going on in every user’s head? How do I even prove all that is even a valid way of thinking about everything? I’m sure that somewhere in the product adoption literature these topics are covered to various extents.
But none of all that stuff really will flesh out the whole theory of “Why and when do people decide to GTFO?” that I need to be able to do product work. Most importantly, how far can I foresee all this stuff so that I can either prevent this from happening to my own product, or capitalize on it when it happens to a competitor. If I can’t manage to squeeze some baseline value out it, then all the potential research I could do to explore the space just winds up being something of a curiosity.
Blech.