One of the first contributions Steve made to the game was also one of the most important. The design I was pursuing was really more of an archaeology game, with players traipsing around, investigating dusty old museums, interviewing locals about long-forgotten lore, etc. I think in retrospect, I wanted to design something that sounded similar to Mad Monks and Relics in scope, but with abstractions that probably eventually would have resulted in a game like Thebes. Steve correctly observed that you can't have an Indiana Jones game without the Nazis. He pushed for a more "cinematic" feel to the game, with more action, and a game flow that would be built more around players having to face Nazi challenges and to outpace them to discover a lost temple, wherein the relic was contained.
Steve also successfully advocated for several conceptual changes to the game. The most significant of these was to build the game's action around visiting "theme cards", cards that correspond to characters and items from a given film (eg "Henry Jones Sr.", "the Grail Diary", etc). We decided that each card would give you access to some particular clue in some sort of clue table, possibly a paragraph book like in Tales of the Arabian Nights.
But what do the clues tell you? Our first approach was to loosely emulate the story-line of The Last Crusade, and so the clues became centered around "where is the lost temple?", and "what challenges will I have to pass in order to get through the lost temple and find the grail?" Our first attempt involved a temple consisting of 6 steps, straight in a row, each requiring you to pass a challenge (spend some cards). If you looked up the clues corresponding to the challenges, you would know what cards to load up on, so there was a nice risk/reward balance, but nevertheless, we found that (a) generally, the first person in the temple would just run through the whole thing in one shot, and, more importantly, (b) it was kind of boring. It didn't feel like an exciting climax to the fast-paced information hunt.
We developed a more involved temple phase in which players had to actually enter the temple (represented variously by cards and tiles at various stages of development), and navigate their way through it to get to the objective. We began with a "structured" temple setup, in which there was a best path through the temple that was provided by the clues, but have since moved to a temple whose layout is random, and whose important features are governed by the clues. So, the game doesn't actually "know" where the grail is located, but it knows what "feature" contains the grail, and you go through the temple, testing different features until you find the right one. This is an abstraction from the game having perfect knowledge but we've found it helps considerably at promoting ease of setup and playability.