GDC Summer 2020: Quantic’s 400K-gamer motivation model

    
1
GDC Summer 2020: Quantic’s 400K-gamer motivation model

Quantic Foundry’s Nick Yee gave a lecture at GDC Summer 2020 yesterday showing how the analytics firm’s massive amount of data can be used in various ways to parse out players into different gamer categorizations. But in truth, it’s not always that helpful. I don’t say that to pass judgement but rather to repeat what Yee himself said before diving into various ways his team categorized players in their data for various reasons.

(I will say much of Yee’s presentation was spent explaining what various categorizations meant, but that was helpful chiefly only to highlight theoretic granular differences you might find within data trying to stereotype players. Watching the full lecture whenever it’s publicly available) might make this more clear. The whole slide show is available right now, though it lacks commentary.)

MMO gamers will recall that Yee and Quantic pioneered the Gamer Motivation Model a few years ago. While it’s fun to look at some of its categorizations, it’s not much like the Bartle Test as just groupings Yee did to show that the data from all the profiles we’ve filled out can be organized in different ways.

For example, while he started out with about nine player segments taken from the whole data set, he also creates five player segments based on players who mentioned enjoying Civilization 5. The idea is that these are ways a data scientist could create clean-appearing segmentations, but the practical application is a bit dubious. Yee says that demographic segmentation may not help when trying to figure out what features would appeal to each player type, but they may help with acquiring them from social media. He also warns that Quantic recommends that its clients not “include gender as a segment input because it’s often not doing what they think it’s doing. It’s exerting an inappropriate splitting point.”

In fact, a lot of Yee’s anecdotes make the data feel more useful in hindsight than for making predictions. For example, a group of players who were often dying, missing shots, and performing poorly weren’t returning to a certain shooting game. The scientists’ initial thought was that the tutorial system needed to help support these players, but further research determined that the players were actually FPS vets who preferred aggressive run-and-gun shooters, not tactical shooters that rely on cover play.

It might be fun if Quantic adds some of these titles to the website for fun, but as Yee noted, these arbitrary titles work on specific contexts. For example, you can segment the Civ 5 population, but those may not carry over well into another 4X game like Masters of Orion. It’s not simply a matter of trying to match genres but remembering that genres can get blurry or contain sub-genres.

No posts to display

1
LEAVE A COMMENT

Please Login to comment
  Subscribe  
newest oldest most liked
Subscribe to:
Reader
memitim

Personally I don’t think this kind of data is useful. At all.
Take this bit:

the players were actually FPS vets who preferred aggressive run-and-gun shooters, not tactical shooters that rely on cover play.

It’s fair to say that’s true of me however it doesn’t mean I won’t play a tactical cover shooter…
Do you know what type of games I like? Good ones.

Obviously it’s nothing more than my opinion whether or not a game is ‘good’ but this whole ‘this player likes these games’ and ‘this player plays like this’ is just trying to put people in boxes for the sake of putting them in boxes.

In the above example they thought they needed a better tutorial but it turns out those people just didn’t like that game, this doesn’t necessarily even mean it’s a bad game, maybe they did like that game but just didn’t feel like playing a cover shooter today and figured they’d come back to it later, in other words it’s all so nebulous it means almost nothing.

What I actually play and how I actually play can change radically from one week to the next and I suspect I’m not alone in this. This data won’t actually tell you anything about whether I will like a game or not and it definitely won’t tell you how to make a good game.

I just don’t see any value here…I’m sure the PR/marketing departments of large publishers will lap this stuff up though and I think that says a lot of things, none of them good…