I realise that, but until that person is paying £21.99/month or more, why should they get a better service than someone who is a "Premier" user?
If you are a premier customer, ask yourself why you are not on a PAYG account. If your average monthly use would cost you more than the £21.99 you pay now, the "premier" package is not as "expensive" as the PAYG one.
On the other hand, you may be asking "why should
specifically P2P traffic be better for PAYG on-peak?" The answer to that lies in whatever your beliefs are about whether traffic type prioritisation is acceptable. Most of the arguments in favour of it (i.e. peak time capacity cannot be as big as it needs to be so something has to slow down) lose some of their merit in a PAYG model since traffic used is, to a larger extent, being paid for and wont be used at some other time. A price model with a cap above the real cost relies on some of that use being "free", either by you not using it or by using at at a time when the network would have been idle. The prioritisation would therefore encourage you to follow that desired behaviour.
Personally I'd like to see sufficient bandwidth in place such that when the shaping kicks in it isn't dreadful on any protocol (say 256kbps at absolute worst), and the packages sold with clear peak/off-peak limits that make this viable. If a package is going to get a smaller fraction of the pot (eg PAYG P2P might temporarily get as bad as 512k at peak, whereas BB+ might go to 256k) then this should be clearly indicated as a "network priority" thing and priced accordingly. This whole thing with hoping customers statistically balance out just doesn't work nowadays where too many "average" users sit with a P2P server running (legal or otherwise) and streaming video from websites. Net habits have changed.
