I don't think you need to rate them all from 1 to 100, my preference would be to order them, like I did on my old toppy 5k- epg tap (can't remember the TAPs name). Initially the order (or rank) is created as the timers are created. Then you move the timers up or down the list to 'rank' them as you want. From experience with the toppy, most recordings don't conflict, however when they do, you have a pre-arranged preference for your tuners.
I'm not sure how you display and manipulate the set of disjoint
partial orderings that this implies. Clearly the TAP did something useful, but I can't work out what from the description. It will be a set of disjoint partial orderings if all you see in one conflict resolution is the set of recordings that are currently actually in conflict. It will reduce to a single strict total order if at each conflict the whole set of recordings on IceTV is listed. For me that's about 150 recordings, and I don't want to have to deal with that.
You get a partial ordering when you have not specified pair-wise preferences between all possible pairs of items. You get a strict total order when all pair-wise preference have been defined. A total order can be mapped one-to-one to a subset of the integers (that would imply that each recording has a unique preference ranking applied to it), and ordering items in a single list implies a strict total order between all of them (because each item is mapped to its position in the list).
A single preference order list may be workable for the list of manually created timers that are typically on a PVR at one time, but I don't think it's well suited to something like IceTV.
Anyway, I would only want a scheme like this to be applied
after some sort of preference scheme of
where the recording would be made was applied - I'd want things recorded with my DP-Lite as first preference, then as second preference on my DP-H1, and only if that failed, some sort of automated decision about not recording something. The scheme would also need to take into account whether padding in sequential recordings on the same service is better preserved by running the recordings on the same PVR (and catching the end of the first program in the second recording), or moving one of the recordings to another PVR and getting post-padding on both.
I have a feeling that it would be hard to make something automatic that I could rely on as much as sorting out these relatively infrequent conflicts manually.