At Rhizome we have collected a reasonable number of twine projects and other web pages that use client-side generated randomness to display different things each time a page is accessed or “roll dice” in a game.
Many of these projects are self-contained, running fully client-side, and are not using the random generator to generate, for instance, random user IDs that would be used in URLs to load ad banners etc.
Webrecorder’s Wombat is replacing the browser’s built-in random number generator with its own that always generates the same sequence of random numbers. As far as I understand wombat doesn’t guarantee that each CALL of the random function would produce the same result because scripts might still load and execute in a different order each time when accessed via a web archive, which in many cases would also be different from capture time. But overall the predictable random generator is used because it improves general replay.
Now in the projects mentioned above, the predictable random generator is effective in relatively simple, not concurrently running JavaScripts, producing the same game or the same algorithmic poem or graphic every time.
This test collection demonstrates the effect.
I wonder if there is a way to prevent Wombat from overwriting the browser’s random number generator for select collections in their configuration? Would there be any side-effects I’m not aware of?