default header

Theory

Why Steam Curation Algorithms Are Doomed To Eternal Failure

Moderator: JC Denton

Why Steam Curation Algorithms Are Doomed To Eternal Failure

Unread postby icycalm » 06 Apr 2018 17:49

https://www.patreon.com/posts/18007867

A heads-up to everyone who's already read the article: SriK posted a lengthy and interesting reply just now.
Image
User avatar
icycalm
Hyperborean
 
Joined: 28 Mar 2006 00:08
Location: Tenerife, Canary Islands

Unread postby SriK » 10 Apr 2018 01:27

icycalm (on Patreon) wrote:The $100 per game that Steam charges, however, is nothing compared to the 30% they get per sale. How many Far Cry 5s at $60 a pop do they have to sell to make the $100 that one "indie" game pays them to get listed? Five or six? And there are what, 10,000 "indie" games on Steam? So Valve makes more on one real game than from the fees of all "indie" games they've ever listed put together. I would guess that their income from sales of those "indie" games is negligible too. I am sure they make a significant amount from borderline real (but still bad) games like Braid and The Witness, so I think they could keep those but ditch the true garbage without losing much profit, and thus solving the Steam-garbage issue that so many of their customers hate.


I was curious about this and spent an hour running numbers, using a site called SteamSpy which tracks sales numbers and price history. The results seem to tentatively support your conclusion. I figured I would post the analysis here, since its length seems better suited to the forum than to Patreon comments.

For reference, first we can look at the sales of Far Cry 4, which comes out to about 1.2 million copies at $30, for an estimated gross of $36 million.

And then these random games I picked from the 50th page of the "indie" top sellers (about 1200 games in, of what seems like ~4200 games total):

Influent - 140k copies at $10, est. gross about $1.4m
CASE: Animatronics - 67k copies at $10, est. gross about $670k
Heroes Rise: The Hero Project - 20k copies at $7, est. gross about $140k

So even at this level of obscurity a fourth down the list, you still have titles making a very surprising amount of money. Not nearly as much as Far Cry 4, but nothing to sneeze at either (although who the hell is hearing about and buying this stuff?), and I mean there are ostensibly pages and pages of better-selling titles above these.

("Ostensibly" because these rankings seem a bit fishy; the game that grossed $1.4m was ranked below the game that grossed $140k, which was a text adventure that looked like it was originally programmed to run in a web browser. My first guess was that this "top sellers" ranking might only take into account sales within the past week or something, but then what's the point of having the "new and trending" category right next to it?)

Things thin out by the halfway point on the 100th page, kind of (it's still weird that the first one isn't on a higher page):

SWORDY - 4800 copies at $15 price, est. gross about $72k
BoX -containment- - 2300 copies at $2 price, est. gross about $4600
Until the last - 770 copies at $0.8, est. gross about $616

And then if you look at some high-profile titles, they're making money in line with FC4:

Stardew Valley = 3.6 million copies at $15, est. gross about $54m
Undertale - 3 million copies at $10, est. gross about $30m
The Witness - 600,000 copies at $40, est. gross about $24m

What this all suggests to me is that Valve could indeed eliminate a lot of the dreck without taking a huge loss (50% of it?), but maybe not as much as we'd like. What it also suggests is that the high-profile titles that make it to this site's reviews section do make up a significant portion of the "indie" market, but perhaps less than previously thought. Of course, this is a relatively simple analysis with very little data, and it doesn't take into account bundles, price changes, and so on. But at least it points towards something interesting.

I'd create a real model with a ton more data, but I don't have the time or the economics background, so maybe someone who does can step in and figure out the nuances of this.
User avatar
SriK
 
Joined: 05 Nov 2011 15:12

Unread postby icycalm » 10 Apr 2018 02:42

I can think of a reason why they wouldn't want to eliminate any games, no matter how bad.

If you eliminate some games, the people that make them will still want to publish their stuff somewhere, and the more stuff you eliminate, the more games will move to another service that will allow them. I don't see much if any dreck on GOG, and of course Uplay and the like would lol at the idea, but basically by vacuuming up everything what Valve is doing is preventing any rivals from even forming in the first place. If you put 10,000 dreck titles on a brand-new dreck-only service, you will still attract hundreds of thousands of moronic players there, and then from that it's just a short step to start selling them real games too, and attracting even more buyers, on a service that will now offer a greater number of games, in the long run, because it won't turn down anybody.

So even if you ignore the extra pocket money that Valve is making from the dreck, it probably still makes good business sense to allow it on their service.

It's harder to judge the business sense of having crappy automated curation.
Image
User avatar
icycalm
Hyperborean
 
Joined: 28 Mar 2006 00:08
Location: Tenerife, Canary Islands


Return to Theory