PC/Mac only you are right. I haven't worked in the industry in a while but I vaguely recall one had to purchase production time and essentially predict your sales for Sony/Nintendo if the system was hot in the marketplace. If you aimed too low, one had customers without product for an extended period (most traditional game sales are in the first weeks), if you aimed too low, you had to eat those DVDs/carts at a loss after the retailers returned them. Also with consoles, the devkits for each were priced ridiculously at something like 3000 for PS3 and more for XBox360 (and those things red-ringed like crazy, went through three in three years), and there were test versions and special development versions one had to use to debug, I remember there was one special PS3 that was I want to say 10000 that was the size of a desktop computer that our resident graphics guru got. So deciding to add on console support for a PC game like they did is a huge commitment for a AAA game, it is not simply a matter of using an engine that works on all the platforms, all the QA has to be repeated and there are different extra steps for every platform but they didn't even get to that step.

$3,000 dev kit is pretty cheap. Just think of the cash a N64 developer had to drop on an SGI...$100k+

True, I remember there was also something of a shortage of Wii kits for a long time at least for my developer shop we were low on the priority list so we only had on Wii developer box for the US branch for at least a year after the Wii came out.

Publishers front money (and having money often leads to needing more money) and provide a lot of expertise to get a game over the finish line.

Publishers give game developers money. They need a publisher to fund the project if the budget grows...

Plus provide support with all the non-game-development things involved in a game development project. Just look at Obsidian who partnered with Paradox Interactive for Pillars of Eternity. They didn't need more money, but they wanted their expertise in publishing.

Cause they overscoped the promised project to something needed a Publisher, wasting months of time with an engine switch which make only sense in the case of preparing for a publisher pitch.

I think this is what I was getting at: they made some bad decisions. If I understand the original project/proposal (maybe I've got this very wrong), but they essentially just had to take an existing concept and build it in a new engine? That shouldn't be too much work, should it? Still a lot of work, but not something from nothing?