I have been working for the past long while on (and just about completed) a proprietary game engine to play works of interactive fiction of my creation. To say that the task has been more difficult than I had imagined would be accurate...perhaps because the traditional plaforms have had many additions to them from ingenious and creative users over the years.
There is something to be said for using a common programming environment (or simply improving upon one) as opposed to starting a new one from scratch. And since my personal understanding of the other environments (TADS, Inform, ALAN, HUGO, etc.) was little when I began this project, it is probable that had I known about how the platforms could be modified to fit the needs of a particular story, I might never have bothered starting on a new one.
But once starting upon a task, why not finish it? And the benefit of having done it is that I have a system of my own design and am intimately familiar with. It is manipulated with external files, so stories can be loaded and unloaded as with any platform. The language with which you address the engine isn't as graceful as with Inform, but my goal isn't to release the engine to the community so I don't have as much concern about that (someday I'll probably write an interface to create some of the code necessary to create a story and allow for the manual coding that is needed but that will be for another time).
My next goals are to have any willing beta-testers help me polish the engine itself and to find someone to create a story with.
My only advice to those who wish to create their own game engine: Think hard about everything that needs to be done. It is a long road, but if you can finish it you will have learned much. Thanks for your time, d