Alpha Protocol is one of my favorite games. I own it on PC, (which I bought off Steam for 3 dollars), and from Amazon, which I bought new for ten dollars. (I bought it new in the hopes that some money would flow to Obsidian.) I actually got one of the pre-order bonuses from it too.
The biggest problem with the game is summed up by another game: Deus Ex: Human Revolution.
The problem with the branching structure of AP is that it means a lot of work that wasn't true during Planescape:Torment or KoTR or Fallout I & II. When you branch now, you have to have developers re-program the entire scene. Voice actors have to retell the lines. Animation has to be done for facial movements and actions. If they stealth or do any movements, those animations have to be added in as well. The branching gives huge replayability to a game, but for a developer, they want to know how to all this extra time/money translates into a higher ROI. It reminds me of something Sol Stein said.
He's an author, but he spent most of his time as an editor and as the main guy you have to get approval for when you want to publish a book. He breaks down authors he works with into two categories: Literary authors and commercial authors. Literary authors are writing something which will stand the test of time and will be praised years down the road. The book needs to be its absolute best and they may take years working on it. Commercial authors are only making their book for the here and now, and they do not invest any more effort than would be required to make the book sell a certain amount. After all, that's time lost from writing the book after that, and that means publishers get mad, less bonuses, etc.
I tend to think of Obsidian as the kind of company that makes the literary author equivalent of games. However, the game publishers will generally not like this attitude and want to see a return as quick as possible.
Back to Deus Ex, outside from the boss fights, which weren't done by Eidos, you could go pure stealth or pure violence, your choice. In AP, you get stuck in awkward zones where some missions clearly prefer you stealth and some clearly prefer that you engage in firefights and everyone in the area is hostile and looking for you. This makes specialization, especially if you picked "rookie/novice" as your class, difficult. These sorts of forced decisions go against the otherwise open nature of the game. Likewise with the skills in Deus Ex, they made it so certain skill enhanced the game greatly, but didn't effect core gameplay. In A.P., it greatly effected core gameplay which skills you choose. That's typical of an RPG, but not of an action game. The problem though is that unlike a game like Fallout: NV, which has a steady progression of difficulty, you can be thrown into some difficult missions at any time and not having the right skill set can make you useless. Another distinction between the action/RPG-style game.
So it had a lot of moments where the game wasn't sure which direction it wanted to go and the game suffered for it. Deus Ex solved this by making the game linear, which means once you beat it, there's not much incentive to go back and play it again. So that's the hurdle that Obsidian or any other game team would have to overcome in making another AP. Personally, I'd love another AP.