Anyway...
I've been thinking a bit about Regency Collection (RC) and perhaps some of my pondering can be of use to others.
The original way RC was done in 2nd ed worked fine due to the way multiclassing and classes in general was set up. It wasn't the fluid concept we deal with today.
3rd edition changed that and in 3rd ed, basing RC purely on class seems problematic. Consider a 4th level Fighter and a 4th level Fighter/Wizard/Rogue/Cleric. They are equally good rulers, but the latter collects every kind of regency fully and can cast spells to boot. In actual combat encounters the single-classed character is clearly stronger, but we don't really use the 3rd edition system for adventures either. Adventures are more down to an ECL comparison, creativity, writing skills and role-playing.
So basing it on having the class(es) or not having the classes does not seem viable. Requiring a certain number of classes or levels to collect regency is problematic because it punishes low-level rulers and it weakens composite domains. Basing it on the degree of multiclassing is complicated and weird (oh, too many fighter classes compared to rogue classes, no more guild regency for you...)
So what to do about regency collection?
I see a few options...
a) Find a default regency collection "magnitude" and stick with it. The default could be that you collect full regency from provinces and 2 holding types, combined any way you want. So a Wizard might have full provinces, full sources, half law and half manors. Perhaps 3 full holding types, in addition to provinces, would better... then composite domains like Ilien and Dhoesone can work fairly well.
b) Level-based regency collection, like Brandon is suggesting here. As mentioned above, this potentially hurts composite domains and possibly Realm domains (Law, Manor and Provinces). It does mean that a high-level regent is an important asset to any domain, which I think is kinda cool... but it might tip the balance a little too much.
c) Some combination of a) and b). For example, every characters starts by being able to collect full province and 2 full holding types (or a combination of halves). Then at level X (maybe 10 in D&D, 3 in Brandon's system) you add another holding type, and a fourth holding type at lvl X x 2.
d) Everyone collects full regency from everything, possibly with the exception of sources and temples. This makes BS the only cap. In this model I like that classes allow you to exceed you regency collection maximum - holding levels of your classes "favoured type" does not count fully towards your max, depending on the number of levels. Fx. for each level of Fighter you have, 1 Law holding level does not counter toward your maximum collection. In Brandon's proposed system it should probably be 2/level. That way a Guild master can collect full regency from X + BS holdings, where X depends on his Guilder/Rogue levels.
Personally I lean towards option a). We figure out what the standard regency collection rate should be and then that is what everyone gets. For some reason I like provinces + 2½ holding types... I think it makes composite domains strong enough in that compartment. The could possible be an option of taking a Regent Action and waiting for some time to pass, in order to change your focus.
So a fighter/cleric with a realm and a temple might collect province, law, temple and ½manors.
A rogue/cleric with a temple/guild might collect province, temple, guild, ½manor.
A fighter with a realm might collect province, law, manor and ½guild.
And so on.