Multiplayer Growing Pains in Subterfuge

Back in December of last year (2014) our game, Subterfuge, went into private alpha. We had a bunch of people who signed up for the alpha so we just started inviting them every time we released a new feature or rule change and wanted to see how it plays out. That ended up being once every 3 weeks or so.

The results were promising. As expected, some people dropped out in their first game, some played one game never to return, but many were sticking around for multiple games:

The big spikes in the red graph (1/5, 1/22, 2/9) are dates on which we sent out a batch of alpha invites. The smaller bumps indicate organic growth, meaning players inviting their friends to play (once you’re in the alpha you can invite whoever you want in as well). Around 1/15 and 2/11 we actually saw pretty big organic growth, which we were super happy about.

Then, in February, we realized that we had enough people waitlisted for the alpha that we could add 100 new people each week until our expected launch date this summer — and that’s assuming no one else signs up for the alpha after that point. So yeah, we started inviting 100 new people every week.

For the first three weeks, things seemed to go great. We nearly doubled our active player base in that time period. But in the following four weeks we were seeing a substantial and steady decline.

In this period, there had been no meaningful changes in the game. We were mostly working on unreleased features and bug fixes. 

What Happened?

A while back I had lunch with Chris Hecker and we chatted about his game, SpyParty.

The discussion focused on how to make the game self sustaining (financially) without having to rush to release. It seemed obvious to me that putting it on Steam Early Access would be a good step in that direction. Chris was already selling SpyParty directly as beta through his website, and the extra exposure would certainly result in large growth, right?

That’s when Chris pointed out something fairly profound that hadn’t occurred to me: There are many ways to grow the player base, but for SpyParty, GROWING QUICKLY is not better than GROWING SLOWLY. The SpyParty community has a supportive, nurturing culture that has grown slowly over time.  A sudden influx of new players could seriously hurt (or even destroy) the culture of that community, and with it, the game’s long term prospects.

I’ve seen this kind of thing happen in non-game communities, so this concern made sense to me, even if there was no hard data to substantiate it.

Well, I think we now have at least one data point that substantiates this conern now.

In Subterfuge, a large influx of new players means the ratio of new players to veterans increases. We know for a fact that first time players are more likely to drop out mid game, either by explicitly resigning, or just abandoning the game.

When one player out of 8 drops out, it’s not a big deal. The game’s politics tend to compensate for the small imbalance this creates. When two out of 8 drop out, it starts to have a meaningful impact on the game’s balance. 

This makes the game experience worse for both new player and experienced players. New players then become even more likely to drop out or at least not return for a second game. More experienced players might not react as drastically, but it certainly can affect their desire to play another game after the current one ends.

We think this is what’s been happening these last 4 weeks.

So Now What?

This week we didn’t send out any new alpha invites. We’re going to slow down the pace and see if we stop “bleeding players”. If we start seeing organic growth again, that would be extremely encouraging.

But the central problem will remain, that a large influx of new players is likely to result in a poorer player experience for the entire player base, and it’s something we’re going to need to address soon because we want to have as large and established a community as possible before the game launches.

There are three things we plan to do about this in the near future:

We are not certain that these things will address this dynamic, but we plan to keep looking at the data to see if any of them have a meaningful impact.

If you have any thoughts on the matter, we'd love to hear them in the comments section!