I'm not sure how much Chris Roberts had to do with the final product but I don't think Freelancer was a failure when it was released.
I dunno if you've played it recently btw, but it hasn't aged that well either.
Ambition and wild ideas are fine, but you shouldn't make promises you can't deliver on.
Of course "Freelancer" (hyped like "Star Citizen" today, and basically the very same game idea) came out late and most announced features were missing - actually the same features are now announced again by Chris Roberts for Star Citizen.
And another game of "Digital Anvil" never got released thanks to Microsoft, it was called "Loose Cannon". It was announced in 1998 and it would have been the first 3D open world game in the very same sense of Grand Theft Auto 3, just released 1-2 years before GTA 3 !! Nowadays we could be playing Loose Cannon V and no one would play GTA. Anyway "Loose Cannon" was very good looking for it's time, and the actual gameplay video footage back then looked really promising. Barely any media can be found online today, but much more is on CD-ROMs of old PC Game mags (1998-2001 era):
https://www.unseen64.net/2008/04/15/loose-cannon-xbox-cancel...
http://www.ign.com/games/loose-cannon
https://en.wikipedia.org/wiki/Digital_Anvil
Like the poster above me, I say the same: "Fool me once..." to both Chris and Bill.
Fool me once...
I've heard people say "the technology didn't exist back then!", which is true. But the technology doesn't exist today either. Games have gotten a lot better at drawing convincing pictures, but in terms of simulating worlds and generating content, we've mostly stood still.
No Man's Sky already showed that, with its demo scene derived trickery. You can't entertain people with random seeds for very long. You need intent, storytelling and reactive, interacting systems.