D&D General As of 1998, 4,007,685 people played AD&D in the US, as estimated by Ben Riggs.


Did nobody else send in the DMG warranty registration card back then? That is how they tracked all the members of the cult. My number was 2,345,666. ;)

log in or register to remove this ad


Scion of Murgen (He/Him)
"Welcome to the most speculative and controversial post in my series thus far!

I hypothesize that as of 1998, 4,007,685 people played AD&D."

Yep, but he's talking circa end of the '90s.

Were there actually 4,007,685 people playing AD&D in 1988? Almost certainly not. But on the other hand, it’s similarly unlikely that there were fewer than 40,0768 people or more than 40,076,850 people playing.

There was another discussion a few years back that posited that in the late 90s there were only about half a million active D&D players (that's for that time period, not all of the players ever).

I think the 4 million is total players (which I think is a bit low, 8 million is still on the low side of what I'd think ,but it falls within the 8-10 million AD&D players of my own thoughts).
As a clarification (in part because @darjr phrased the thread title inaccurately; UPDATE: thanks for the edit!), note that Ben's estimate is that roughly four million people HAD PLAYED AD&D AS OF 1998. Not that there were four million active players IN 1998.

Further he is not counting D&D or other RPGs. Only AD&D, and only counting North American sales.

Just so we're clear on what he's talking about.

I tend to concur that his baseline assumptions are dubious. His estimate may not be far off, as he's within a couple of million of WotC's old estimate, but I'm really not sure on what basis he's establishing several of his premises.
Last edited:


Well with the WotC 5.5-6 million number late 90's/early 2000's I assume they guesstimated an attach rate to PHB sold via surveys I suppose.

There's also a difference between active players vs everyone who ever tried it.
I reckon thst they trues a similar exercise to this calculation, but with more established data than assumptions. And it is still spitting distance.


Follower of the Way
I wasn't 100% clear on what your concerns were. Correct me if I'm wrong, but I think you were concerned that the proportion of 1st and 2nd ed. players is underestimated compared the 50 million figure. Is that correct?

Comparing the 4 million with the 50 million figure is going to be problematic:
  • the 4 million has a big margin for error because of the Fermi estimation
  • we may have introduced unconscious bias that systematically skew the estimates (possibly based on our own play experiences)
  • we don't know the methodology used to produce the 50 million. For all we know, there could have used Fermi estimation as well
I see your point about 4 million would still part of the 50 million and could be better represented in the graph.

BTW, I the original slide (I think) and it was in 2020:

I assume that is "TTRPG" players, but I can't be sure. Does it include CRPG players?
Yes, that's my concern here. Two different Fermi estimates cannot be meaningfully compared because even in justified use cases, the relative sizes are specifically in doubt. If we cut that over 50 million figure in half, it would still be a good Fermi estimate, and likewise if we tripled the 4 million figure it would still be a good Fermi estimate, yet that would make the change go from increasing by more than a factor of 10 to increasing by just over a factor of 2. Hence, the graph is purporting to display information that we don't have by giving the values implicit precision they do not possess, much like specifing the values down to the single digits when that is clearly way below the precision of the estimates involved.


Well.... that's not entirely true. There is a claim being made here that depends on the value being pretty close to the listed value, and not merely within an order of magnitude of the actual value.

That claim is implied in the graph shown which compares the expected number of players in different eras. Because, as it stands, it is communicating the idea "there are ten times as many D&D players now as there were in 1998." That claim is...dubious. Given some of the assumptions that are questionable (such as only 4 players per DM and not 6 or more, and assuming a very high re-buy rate, among other things), it seems this is specifically aiming to he a low-ball estimate of yesteryear's player count and a rather inflated estimate of current player counts, leading to that dramatic comparison when the actual comparison is likely to be much closer together.

Given further things, we expect high retention and low attrition. That is, US population growth (e.g. we've gained about 60 million new bodies since the late 90s and about another 60 million between 1974 and 1998) means there are literally millions of young people who could join up. Further we expect a comparatively slow rate of losing old players by death: someone who was in their 20s in 1998 is in their 40s-50s in 2022, attrition by death is going to be fairly low. Further, major growth is likely with three new product lines (or four, it you count 3.5, or five, if you count PF1e, both of which I do). That means whatever numbers we claim for 1998 should mostly stick around and be part of the current numbers today. If the actual values are closer to 8 million and 20 million, that would be more than double, but far far less than over 10x as much.

That graph is really where the claim is being made that we should take this estimate as actually representative and not simply a plausible guess.

I have seen these "irrefutable" claims. They suffer from the same faults that tell us that the "Doomsday Argument" does, which asserts that, by the Copernican Principle (that is, "you are probably not a special observer") and a few other things, humanity is essentially guaranteed to go extinct within the next ten thousand years, purely based on average human lifespans and the estimated number of humans that have ever existed. The "irrefutable" nature of these claims is not nearly as ironclad as you imply. Merely choosing the wrong kind of prior (one of the main objections to DA-type arguments) is enough to trash the entire approach.

There are also issues like addressing the "grabby alien" hypothesis which can explain the divergence between the Drake equation estimated value and the actual amount of intelligent life observed, among other concerns.
WotCown numbers put the number of everyone who ever played an RPG at the end of the 90's at 5.5 million, but the number of active D&D fans in 2020 at 50 million. So a nearly 10 fold increase is the case. And note that a huge portio of those 50 million people weren't even born yet in 1998!


Follower of the Way
Somebody has to be.

Or, we are a much more extreme outlier than generally thought, somehow.
The Copernican Principle is useful, but by definition it has its limits. As you say, some civilization has to be the first one. The problem of basing new statistical reasoning on the basis of prior statistical reasoning. You get a telescoping chain of "if X holds, then Y has a 90% chance of being true" etc. And that's great, but when you stack together seven different 90% if-then statements, you've only got a 47.8% chance that the final result cashes out, and it just gets worse from there.


Follower of the Way
WotCown numbers put the number of everyone who ever played an RPG at the end of the 90's at 5.5 million, but the number of active D&D fans in 2020 at 50 million. So a nearly 10 fold increase is the case. And note that a huge portio of those 50 million people weren't even born yet in 1998!
Sure. I'm saying, without knowing their data or method for determining this, these look like Fermi estimates. You cannot make a direct size comparison between two different Fermi estimates. Their magnitude is, in fact, exactly the thing being left open. These numbers are accurate, in a certain technical sense, to zero significant figures, while avoiding being useless.

It would be like making a Fermi estimate of the number of piano tuners in Chicago, which could be ten times too small or too large, and also separately making a Fermi estimate of the carpenters in Chicago, which could be ten times too small or too large, and trying to conclude from your estimate of there being 1246 piano tuners and only 105 carpenters that piano tuning is a massively more popular career than carpentry. You don't know that! It could be anything between a hundredfold increase (~11000 piano tuners vs ~11 carpenters), or no increase at all (~1100 piano tuners, ~1100 carpenters), or exactly the reverse conclusion (~110 piano tuners, ~1100 carpenters), or anything in between.

That is, in this case, this Fermi estimate of the late 90s player count would be perfectly consistent with 40 million players. And, if it is also a Fermi estimate, the 2020 number would be consistent with there being only 5 million players, a drop of about 88%. But it would also be consistent with there having been only 400,000 players in 1998 and 500 million players in 2020, meaning a thousandfold increase.

That's why you cannot do the comparison that that graph entails. Because it illustrates a relationship which only holds if the values are reasonably precise. These values are not precise, at all, they are (at least for the 4 million figure, and likely the 50 mil as well) useful but extremely loose estimates. The error bars on that graph should be in the stratosphere because of how wide the range of validity is.


Goblin Queen (She/Her/Hers)
I'm saying anything evolving at the same rate could theoretically have evolved 9 billion years ago. And gone extinct in a few million years.
That’s absolutely true, but it is still statistically very unlikely that we aren’t among the first intelligent species in the universe - however many there may have been before us, there will be a lot more than that after us. And that’s weird, because it makes us an exceptional case, which is by definition unlikely. Maybe it’s a coincidence, but more likely there’s something off in our assumptions. Colonialist aliens is a pretty good explanation because if it were true, it would make the observation that we’re early exactly what we would expect.
Last edited:

Remove ads


Remove ads

Upcoming Releases