I think a decent argument can be made Carmack is becoming a failure. It's a bit weird. I saw a gaffer more or less say something to that effect over the keynote, and that he was sad about it.
I completely disagree with that. id is certainly having problems but I don't see how or why that would be Carmack's failure.
Let's get a few things straight first though.
id has never really been a studio with huge sales, their best selling title to date is Doom 3 with something like 3-4 million units plus whatever this BFG edition has made. But their self-financing and small studio model has made them extremely profitable, and their technological prowess made them highly influential too. Also, their shareware approach has allowed Doom to multiply that effect.
Realtime 3D graphics is converging to a singular approach and it's very close to become a solved problem. I've posted about it a lot, but just take a look at current presentations to verify it - HDR gamma correct rendering, physically correct shading, shadow buffers and so on.
Most of the technology at this time is some approximation of the real thing (which would be to raytrace every pixel within the above mentioned framework), and there is some competitive edge in how someone can do these approximations (like global illumination or area lights), but faster hardware is going to eventually take care of that too.
All in all it's absolutely different to the age when hardware limitations made it necessary to be extremely clever in doing even simple stuff like sorting polygons for rendering. At that time it was much easier for a genius like Carmack to come up with solutions that put his tech way ahead of everyone else. Today it's much harder to stand out from a crowd of games that all look extremely good, just as it's much harder to move the general gaming experience ahead the way the move to 3D did.
On the other hand, the complexity of games has moved forward significantly, and a simple straightforward FPS like Doom can not even be considered at the arcade level, the audience expects more. So games need a huge team of designers, artists, and engineers to build gameplay systems and physics engines and such. Standing up to these expectations has required id to abandon the small studio approach, and the amount of capital necessary has also meant that they had to fit into the publisher system. This transition was apparently not managed well enough and this is what lead to id's current problems.
Also, most of the original team at id has left the studio or was fired (and Paul Steed has actually passed away about a year ago...). Romero was much publicized, Tom Hall not so much, and they both faded away from the front lines of game development long ago. Graeme Devine, Sandy Petersen, American McGee are more or less in the same situation and they weren't that influential at id anyway. Adrian Carmack was fired back in 2005. Kenneth Scott and Sparth (Nicolas Bouvier) has left during Rage's development to take over Halo's visuals at 343. So I think only Tim Willits and Kevin Cloud are still there from that Doom-Quake team by now.
So, one could argue about Carmack's role in id's management issues over the past year mostly because he was one of the original owners before the Zenimax sale. He probably had some level of say in the larger decisions but I don't think it had too much weight in the end. Blaming him for the Doom 4 issue would definitely be too much.
The other argument could be that Rage was delayed because of its extremely work intensive unique virtual texturing approach. However that wasn't the main factor in its relatively modest sales, which I frankly consider to be unfair as the game is quite good. I also believe it looks amazing, especially considering the 60Hz framerate.
He explained the Armadillo issue in several articles, it's mostly that space exploration is hard and needs a LOT more money than he had. Once again I don't see how one could consider it him failing.
And lastly, the keynote and the second talk has demonstrated how well informed he is about the state of not only the games industry, but also offline graphics, consumer electronics devices, technologies and market trends, and a couple of other things too.
So, all in all I find your statement completely unjustified. And Carmack has moved to Oculus to be part of that next possible great step, VR, now that the graphics problems are practically solved; and probably also because it's a small group of people and he misses the old days of id. I fully expect him to contribute to VR greatly, even if it might not be that easy to see his part in the final results.
