Firstly the Moskowitz Prego research was in 1986 so it’s possible TSR knew about it and it confirmed that their D&D (B/X and BECMI) and AD&D split wasn’t hurting them (although the reasons for the split were likely royalty related). Some played AD&D, some B/X then BECM, but for the most part and many just bought both types of adventures.
When 2e rolled out and the redo of BECMI, I think the philosophy may have been: 2 varieties, don’t mess with success. So we got some minor clean-up, and a few common house rules made official. Based on some of the history of 2e D&D from David Cook and Kim Mohan, there was a strong “don’t change much” mandate from the top. I think there was less buying of both here because they became more different, and gamers were starting to be more selective because of the release schedules got much more bloated.
Since TSR management felt like they had already created the 2 best choice versions of D&D they let it go unchanged for long periods of time although twice they released revisions/new options to the game which saved them from the brink of bankruptcy (UA and the 1993 reprints/Options books).
When the company did go under it was more related to bad business practices and not D&D per se (returned inventory of books and dice, over-production, and making products that cost too much to make).
WotC did some market research after the acquisition in the Coke/classic vein. New Coke was in 1985. Some blind taste tests, but primarily surveys asking what people wanted. Now in the Moskowitz model that doesn’t account for players not really knowing what they want. For example in 1999 I thought I wanted 3.0 and then later 3.5. But what some wanted was essentially the “chunky spaghetti” version of D&D which didn’t exist yet. Not chunky as in heavy crunch, but just a variant that some wanted but hadn’t existed in the “D&D” family as a major choice.
The TED talk was in 2004 which would have been after 3.5 came out. But I think it may have informed 4e (providing an unmet flavor for the 20-35% that didn't know what they wanted).
When I go back and read the posts from 2009 and earlier I see a strong trend to compare 4e to the New Coke fiasco. D&D was no longer D&D, maybe it was a good game, but it wasn’t Dungeons & Draqons.
It certainly informs some of the decisions we're seeing now.
1. No edition warring per GAMA presentation
2. Keep 4e in print
3. Not badmouthing Pathfinder
4. Reprint White Box, 1e, 2e, and 3.5
5. DDN as Core plus dials, modules, and options
6. DnDClassics.com - "Every Edition Available Again"
I originally watched the video because Mike Mearls referred it in one of his posts (I couldn't find it again - likely between 2004 and 2008).
I'd posit that 4e itself came from the uncontrolled 'test group' of the 'internet'. Many of the problems and solutions came from the forums (toss in a little "distance from the OGL" mandate per the higher ups). But until the research (i.e. sales) data comes back you don't really know how big that group really is.
Certainly 250,000 downloading the rules plus a sizable but unconfirmed survey response would let them generate some sort of grouping for preferences. The DDI itself generates a trove of minable data.
I believe they are using the data and they know what the groupings are (given the customization of an RPG each grouping can be flexible).
I think WotC can do the multiple editions from their end (essentially they are already doing it - premium reprints, DnDClassics, DDN, keeping DDI & 4e in print).
The problem will rest in individual Retailers where of 6 tables for RPG: 2x5e, 1x4e, 2xPathfinder and 1xOSR is seen as a problem. Or trying to stock the Premiums, 4e, and Next will be too much.
Amazon and the other e-Commerce sites won't care.
And the bigger problem from the players themselves going tribal on the other groups to validate their own preferences. Edition warring on the forums and in the stores.
A thread from 2009 discusses this – with some excellent predictions of what is happening now.
http://www.enworld.org/forum/showthread.php?269298-Spaghetti-Sauce