A lot of responses in this thread are concerned with narrative or aesthetic issues, especially romance. But Sawyer's odyssey has little to do with that and more about the nitty-gritty of systems and RPG design. Sawyer is borderline traumatized by his stint as lead of both Pillars of Eternity games. Both were games where he attempted to re-design many of the core assumption of the CRPG subgenre. Which, up until the 2010s renaissance, had mostly comprised of people creating ever more faithful adaptations of D&D.

Pillars 1 saw a very contentious design period, with people opposing many if not most of Sawyer's ideas on the obsidian forums. A lot of it was knee jerk reactions to changes from Baldur's Gate 1 and 2, but Pillars was still experimental and Josh was not a genius developer who could outpace all the playtesting involved with original tabletop gaming just like that. The first game was a success (arguably on the back of kickstarter and the general CRPG renaissance) while Deadfire otoh was a demoralizing flop at release (arguably on the back of terrible marketing combined with a less familiar setting). Deadfire got a long tail, with good sales after the fact. But it still didn't justify the work and finance put behind it. That's why Sawyer thinks he doesn't have the pulse of the CRPG audience. At least not any more.

The tragedy with Josh, I think, is that he and Obsidian became lost on the growing pains of creating a new ruleset for a CRPG and don't realize how unfocused these games were. Pillars of Eternity 1 was trying to be Baldur's Gate 1/2, Planescape Torment and also Icewind Dale. All into one. Deadfire was no better with a story that also didn't know wether it wanted to do Big Metaphysical Stuff or delve into the issues surrounding colonial warfare and settlement. Sawyer in particular is a good lead when you have a deadline to finish and want to deliver a mostly bug-less product. But he's old, experienced and unlikely to take risks. Compare that to the Pathfinder developers who deliver insane projects that are more bugs than code. Hell, compare that to Sawyer's attempts to improve on what was a disastrous game under him: the spirit meter from the Mask of the Betrayer DLC was pretty much universally well received. Sawyer himself said he wouldn't have implemented it today.

More successful developers like Larian and Owlcat Games simply adapted 5e and Pathfinder, while focusing on their own strengths. Sawyer was fully occupied creating a whole ruleset for Pillars, only to see mixed to negative reactions from anything but a core audience. All the while forgetting that Obsidian's real strength is writing these little stories self contained stories. You see that in New Vegas, as well as in Pillars' DLCs. You're unlikely to see it in Avowed, just as you had to strain your eyes to see it in the main storylines of Deadfire and so on.