What Games People Are Talking About: A Pie Chart

You may have seen me talking about EN World's HOT GAMES TRACKER recently. It asks the question: What's the current zeitgeist? What are the hottest games being played right now? This isn't a list of sales figures; it tracks what's currently being talked about using a top secret algorithm. Each game is also conveniently linked to a search for discussion about it right here on EN World, should you want to find out more. The spotlight list changes from time to time. The red and green arrows show a game's general trend over the last 90 days - is it being discussed more or less than it was in the previous 90 days?


This page tracks discussion of over a quarter of a million forum members and approaching a thousand blogs on a selection of major independent RPG discussion forums to create an overall sample from a list including EN World, RPGnet, UK Roleplayers, RPG Geek, the RPG Bloggers network of nearly 300 blogs and the RPG Blog Alliance of nearly 600 blogs.

I've extracted some data from that page and turned it into a couple of nifty pie charts. I've presented them below. A couple of caveats:

1. The pie charts aren't really the the thing. The raw data on the linked page is. If you disagree with the way the data is presented here, I encourage you to look at the actual numbers instead and derive your own conclusions.
2. This is NOT sales data; it's also NOT what games folks are playing at home. It's exactly what it says it is: a large representative sample of the game folks are talking about online. So be careful what conclusions you extrapolate from that.
3. The final D&D Next playtest packet was just released. This gave DDN a huge boost. I looked at this data this time last week (sadly, I didn't think to graph it) and Pathfinder was leading D&D Next by nearly 5%. I'll look again at it in a few weeks to see if D&D Next holds its current lead or drops back down to second place again.
4. The two graphs were compiled a couple of days apart, so the figures changed slightly between them.
5. I was asked yesterday why 13TH AGE was considered D&D in one graph, and not part of "Other D&D" in the second. That was simply because it was the largest item in "Other D&D" so I slipped it out separately as a clear visual point of comparison. It's not meant to imply the "D&D-ness" or "lack-of-D&D-ness" of it or anything else.
6. It's ROLEPLAYING GAMES sites and blogs only. I'm sure if we were looking at tabletop wargaming sites, things like Warhammer 40K would be vastly more popular; as it is, that game refers here to the line of RPGs, not the wargame.
7. This is not a qualitative judgement. Your favourite game is the best.

So, without further ado:

​

​

 

[Morrus]

It's not just increasing non-company site portions, by going to company websites, you're adding a lot of bias to the sample. Naturally, a preponderance of discussion on sites like WotC and Paizo will be on their own products, making them particularly biased sources of data. Sites with a particular focus in their history, like ENWorld with its original D&D focus, already add bias to the sample despite being open to discussion of any games because the historical focus generated a sample of gamers with a systematic bias. A company website would be like injecting that sort of sample bias on steroids.

Absolutely. Nobody will deny that a company's website will - nearly always - be the largest source of discussion about their own products. So the largest source of D&D discussion by far is wizards.com, the largest source of Pathfinder discussion by far is paizo.com, the largest source of Green Ronin's stuff is their website(s), and so on.

What I would LOVE to do (but I don't believe I can) is somehow separately measure ALL of those and see if the proportions match up to the independent site data. So if Game X is discussed 3x as much on independent sites as Game Y, is the official site/forum also 3x as busy? That would be fascinating to find out (and if it were the case, it would indicate that including/not including all of official sites would make no difference overall; but including only some of them would distort the results heavily).

So in that particular case "more data" or "a larger sample" would actually be less representative, and less useful, than less data. The statisticians in this thread no doubt have special terms for that! Sometimes a smaller sample is better than a larger distorted sample - although our sample in this case is pretty darn large!

 

[Morrus]

Sorry, I should have made it clear I was aiming at your post #21 ("statistically significant sample" and whether there's any need to bother with confidence intervals) and not at the actual HG page. You can't get much more careful in a description than saying "top secret algorithm"!

To secret algorothm is just a Google page rankings joke. Probably funnier to those of us who run websites! Basicially, nobody knows Google's algorithm, and SEO experts get paid a lot of money to try to game it - all the while, Google keeps changing it to stop them.

 

[dd.stevenson]

It's not just increasing non-company site portions, by going to company websites, you're adding a lot of bias to the sample.

Emphatically no. If Morrus' goal is to answer the question "what rpgs are people talking about on the internet?" then introducing company sites into the survey cannot introduce any bias that cannot be counterbalanced by increasing the non-company site portions. Not unless you define "internet" to exclude company sites.

 

[Mishihari Lord]

Really interesting, Morrus. Thanks for sharing. The one point that kind of shocked me is that only 13% of the discussions counted were non-D&D. Considering the number of games out there, I expected D&D to come in at less than 50% for all versions combined.

 

[Morrus]

Really interesting, Morrus. Thanks for sharing. The one point that kind of shocked me is that only 13% of the discussions counted were non-D&D. Considering the number of games out there, I expected D&D to come in at less than 50% for all versions combined.

Pathfinder is massive; the DDN stuff I imagine will start to fall off again once the latest playtest package wears off. So that percentage will change over the next month or so (that's just a guess though).