OS/2 is a lesson every ambitious game developer needs to know about. OS/2 in 1994 was so much better than any alternatives that it should have been the defacto OS. It ran DOS, Windows, and OS/2 programs side by side. It was very stable (rarely crashed). It had a very easy to use, object oriented interface (Similar to what we finally have in Windows XP) and it had preemptive multitasking and a DirectX capability called DIVE.
Yet, despite all that, despite being from IBM, Windows won. And it should be a lesson to developers that having the best product - be it a game or whatever, is not the whole story. Your ability to market and distribute your creation trumps the quality of the game in terms of real success. That is why a good publisher or distributor is worth their weight in gold.
If a publisher can get your $40 game into the stores and can pay you on time and allow you to sell lots of units, then getting $10 a unit is a great deal for the developer. Woe to the game developer who focuses on percentages instead of on how much they'll gross. Because it's better for your game to sell 100,000 units in which you make $10 a piece on then to sell 2,000 units in which you make 100% on each sale.