Rat race to the Steam store

“Boy, that escalated quickly!” (c)


A few days ago Sergey Galyonkin tweeted the graph of the Steam game releases, showing a steady growth in the last couple of years.

Seems like a good fit, at a first glance :)

Its message should be clear — the amount of games released on steam every month is rapidly growing to the point when no sane human being will be able to catch up with all of them.

I was only slightly confused by that dent on the right. It could have been just a bad parameter fit due to the “noise” of the actual data, but I though, what if the fitting curve cannot even represent the correct behavior?

Fitting the second order model, to compare with the SteamSpy data

So, as a person, who’s jumping on every opportunity to procrastinate, I ventured right off to the Steam store API and scrapped the release dates myself. Though some data was missing, the end result looks almost the same.

Even better, now the function seems monotone! But does it actually work? Or is it just a luckier fit of the still inaccurate model?

Let us look at the games released before 2013

Boy, oh boy, does this not fit at all :)

The red part of the parabola is the part missing from the previous figure. So, the second order curve fits only the most recent releases, and it doesn’t explain the years of data before that.

Bring in the fun (and math)!

The polynomial fitting model is normally used to describe a slow growth. The model you’ve seen thus far estimates the amount of games released based on the date of the release and the square of it.

An alternative law, describing the growth of game releases, can be derived from the model of the spread of information. According to it, the growth rate can be viewed proportional to the number of possible connections between all the elements in a set, or in nerdy terms, the amount of edges in a complete graph. It is roughly the square of the size of the set.

Dynamic model of the information growth

In its formula t stands for the date of the release, x represents the amount of games released, and dotted x shows the first derivative, or the speed at which the amount of new releases grow. However, this formula doesn’t tell much unless you find the solution, meaning the rule of how x depends on t.

Solution of the information spread model

The solution ends up looking like this, where the amount of releases is inversely proportional to the coefficient a and the amount of time left till some time point t0.

Does it even fit?

Here’s how the fit looks for the recent releases

Again, comparing with the initial graph, this one looks pretty similar, though you can notice a more pronounced uptake of the releases. The more interesting, though, is the overall figure:

Hyperbolic fitting, taking into account the complete history of Steam games

Now the model seems to reflect the fact, that the rate of the releases picks up faster and faster. If you think of the percentage of games released on steam vs. released at all on PC, the model seems to be even more precise.

And here comes the plot twist!

That parameter t0 we glanced over. It’s greater than the current date, and the closer we get to it, the faster we are approaching the division by zero.

Lo and behold, the #indiepocalypse is nigh!

A time point like this is regarded by some in similar models as the singularity, transcendence with Johnny Depp, or simply that time when shit goes down. I would call it an #indiepocalypse, because I’m a romantic like that. According to my fitting model, this time point is April 3rd 2016, 12:54, though because of the mentioned errata it might be off by a day or two :) And that means…

We only got about half a year before Steam will quite literally drown in games!

“So Tony, do you actually believe there’s a singularity point on the horizon?” Of course not. Why did I write this post then? Well, my goal was to raise awareness of the changes that have happened. And, of course, of the changes that are still to come. The rate new games are being produces is not going to go up indefinitely, but how high will it climb — kind of hard to even guess from the current point of view. One thing is clear, sometime in the very near future the growth rate of new releases will grind to a halt, constrained by the factors, that do not seem relevant right now. So let us put on the Nostradamus hat and guesstimate some of the possible scenarios, that might actually realize.

Logistic growth model

According to this model the growth of a biological population is self-regulated around the point, when there is an indication of a possible scarcity of resources. In clearer terms, the population (monthly game releases) will grow unconstrained for a while, and then gradually stabilize at some, probably still overwhelmingly high, number. The measure of the available resources with respect to Steam games is, of course, the overall sales figures. Let’s look at another graph provided by Sergey.

Yes, it does compare the lifetime sales, which of course are higher for a game that’s been around. However, even taking that into account, I really see no evidence that the median of the ~300 games released last month is ever going to climb as high as the 2013 figures.

This means, the market is not as unconstrained as some think. It is true that Steam is selling more and more games. But every new game is not only competing against the growing pile of the contemporary releases, but also against an even larger back catalog.

The closest we got to a scenario like this is the AppStore or the Google Play Store — there is no shortage of the games released each day, but you don’t ever hear of the majority of them, since it’s kinda overwhelming. And even though it might not seem like that yet, this model is what I’d call an optimistic one :) Which brings me, of course, to the other one…

“Predator-prey” model

This model, also known as Lotka-Volterra model, describes the dynamics of the two populations, where one is quite literally consuming the other. The growth behavior of the populations in this model is cyclic for both the predators and the prey.

I like this model because it not only illustrates what’s going on (the cyclic behavior), but also why it is happening. The abundance of prey leads to an unconstrained growth of the predators, that in turn eat up the majority of the prey. This leads to the situation, when the prey is so scarce, the predators start dying out of hunger. And the following scarcity of the predators give prey the green light (no pun intended) to breed uncontrollably, which closes the cycle.

At first I thought the games are the prey, and was wondering what entity might take the role of the predator in this scenario? Maybe it’s marketing analysts, that jump on the bandwagon of every new hit and grind it to the ground? Or maybe it’s the shareholders of large publishing companies that demand profits quarter after quarter?

But the more I thought of it, the stronger I lean to the hypothesis that the games may in fact be the predators in this model. And the games released are feeding on gamer’s willingness to buy them. Some years ago, along with the Kickstarter hype, and the floodgates of Greenlight opening, the gamers were ecstatic about the opportunity to have a game made tailored to their desires. The prospect of games arising in such niche genres, that would satisfy even the pickiest of gamers, was titillating. And then we learned, that even though there will always be fans of your game, there might just be under 32000 of them in the whole world, and the chances are, they’re already playing a game, that’s like yours, but came out earlier.

What would this model imply then? I guess if the perceived hype and eagerness for the new frontiers of gaming has faded, we’ll see a delayed, but a very sharp drop in the amount of new releases. This situation seems entirely plausible, and that’s the pessimistic scenario.

One important example of a sudden fall of the games market is, of course, the infamous “Crash of ‘83”, which seemingly was rooted in different types of problems. That time the crash occurred mainly as a result of bad practices of console manufacturers and overly optimistic predictions of the distributors’ abilities to sell the physical copies. The customers did wise up and stopped buying bad games, and by proximity they became wary of the console games in general, possibly driving the rising third-party developers out of business.

Interestingly enough, the video game industry crashes were pretty common in their early years. As Mirko Ernkvist said in his article Down Many Times, but Still Playing the Game

Despite the familiarity of the 1983 crash, no study has elaborated on the fact that crashes and shake-outs were a recurrent structural phenomenon of the video game industry during its first 15 years, with a number of severe crashes or major firm shakeouts occurring in every game platform after short periods of high growth.

In the article (that I highly recommend to read in its entirety) Ernkvist determines the three sources, that joined together lead to an industry-wide crash and a mass-closing of the development firms. He labels them the “three D-s” (shortened, to my confusion, to 3D), namely

  • Disruptive technologies,
  • Delimited differentiation,
  • Decreased entry barriers and destructive liabilities of newness and smallness.

And it doesn’t take much to see that all three of those characteristics are conveniently introduced and wide-spread recently via the accessibility of free game engines, such as Unity or Unreal Engine. Indeed, these tools are on the bleeding edge of the technological progress, incorporating latest disruptive technologies at the speed, that just could not have been matched by the small development firms with the in-house engines. And though the popularity and availability of these engines greatly decreases the entry barriers for new players in the market, it unifies the look and feel of every game produced with them, i.e. delimiting the differentiation. Together with the simplified self-publishing process on Steam, the entry barriers are probably at the new all-time low.

So, if by now you’re not convinced of the inevitability of the impending crash, I hope I at least made you consider it highly plausible.


Will either of these scenarios come to life? Is there a more optimistic scenario, when suddenly Steam populace grows tenfold and to be named the “softcore gamer” it wouldn’t be enough to own just 4 games? Or will we witness the games market crash so hard, that we’ll enter a new dark age of gaming? Or maybe the #indiepocalypse will turn out to be a thing after all? To the majority of those question my only response is

God, I hope not… but in a way, wouldn’t it be fun to watch?