While I've played neither of the Dead Rising titles, I've seen plenty of games in which exploration and discovery were both keys to victory, to the point that if I find myself depending too much on my tactical skills (and am not playing an expressed tactical simulator), I'll take a step back and look around for elements I've missed.
I would list Thief: The Dark Project and Freedom Fighter as two good ones, in which there was almost always a way around guards and goons whose defensive position seemed impenetrable. Indeed, both games were rife with back-alley routs that allowed such pinches to be flanked or circumvented entirely.
But the idea of mixing and matching items to create superior weapons or better healing brews does make DR worth a look. I'm not sure DR1 will ever make it to the PC but DR2 already has, and I will certainly play it at some point.
When developing a fantasy game some time ago, I remember wanting to create a ro-sham-bo [http://en.wikipedia.org/wiki/Ro-Sham-Bo] system that compared opposing weapons and maneuvers, giving one side an advantage over the other. Part of the play was in discovering which maneuvers beat what, and which weapon was best suited to a personal fighting style. The maneuver tree would be generated at the beginning of the adventure, requiring the discovery process to be repeated each time the adventure was begun.
This was a common device used in early Sid Meier designs for adventure games (the 2004 remake of Pirates! [http://en.wikipedia.org/wiki/Sid_Meier's_Pirates!_(2004_video_game)] is probably the most recent example), since games during the early '90s were often built for replayability. In Pirates!, nuances such as relations between nations, the availability of maritable daughters, the locations of treasures and so on were generated with each iteration.
(This was also attempted in Escape From Monkey Island with the Monkey Combat finale, but it was poorly executed, requiring too intricate a charting system for too little good effect.)
There is the advantage of, as Yahtzee called it, water cooler conversation where a sizably extravagant model is used instead of a generated one, which makes it consistent between games. This allows players to share their discoveries in fora and catalogue them in FAQs and Wikis. A happy medium might be one in which the end results are generated, but the paths to discovering are consistent and fodder for sharing vectors.
Still, the mainstream game industry seems, with few exceptions, to shy away from replayability, since a good game that entertains well for many hours might deny a market for other games of a similar ilk.
U.