When we were crunching on Horizons, I was in the strange position of being pretty damn sure the game would fail because it was not fun and had major issues, both design and technical. But I had also never shipped a commercial game before, and management was saying that everything would be ok, and I had to be open to the idea that I was wrong. So I ran it like a science experiment, expecting the worst but trying to work as though it would be a success.
In the end my instinct turned out to be right, but I also know that this was only one data point. Still, there are some warning signs that I know to fear now.