That's a good idea. Another option: count how often each script is run before the first error.

Perhaps you'll be lucky and script x will be run exactly the same # of times before the failure each restart...