Yeah, gotta say the "no rules" thing doesn't make much sense if you believe "form follows function." I agree there should be some exceptions, always, but at some point, "if it feels good do it" takes you from decent golf with variety across a line to "goofy golf."
I do consider USGA Slope stats, Broadie and others stats, etc. to at least make a preliminary "minimum" size determination for the green. I may not stick to it, but at least I don't want to unintentionally design a green very few can hit. In hopes of creating variety, I don't just look at each green site and design just to fit it (remember, only a few projects worldwide have sites that just call out for a green, most just have general slope direction and steepness, and perhaps a natural hazard or two to work with.
After calculating approximate green size, I break holes into groups of 4 or 5 - i.e., par 3, par 4 short, par 4 long, and par 5. Among each of those 4 holes, I look at my design preconceptions and the site to try to have a mix of small, medium and large green sizes for those approach shots. It's an attempt to differentiate the challenges on similar length holes. Rinse, cycle, repeat based on a whole lot of other factors. Of course, this is on paper, always subject to future change on paper or ground if I see something different that is possible and better.
But for instance, if I am looking at the group of 4 long par 4 holes, I recall Nicklaus saying he liked a two tier green on downwind, long par 4's, and he would use less spin to "chase a ball up to the top tier" to best avoid hazards. I like to encourage actual run up shots for good players, so I look for the longest downwind par 4 and use it there. I also agree with Pete Dye that the best way to challenge long irons is smaller greens on the longer par 4 and 3 holes, and sometimes the longest of each (and par 5) gets a smaller green that is essentially a single smaller target. It may not be a postage stamp, but it is smaller for the length of the shot that many would expect (also relies on the theory that TD mentions - most average golfers more likely to come up just short, even with multiple tees, so it doesn't matter what size green is to them.) The other two long par 4 are medium size and focus on angle of attack, different "Sunday pin" locations, etc.
One of the reasons for this approach comes from starting my career with Killian and Nugent, who later on collaborated with Jim Colbert. (which I continued when I started my own firm) They weren't golfers, and hearing Jim talk about how good players approached the game was a revelation (which KN didn't always accept, but I found fascinating. Jim came to tour Kemper Lakes one day. The 6th was one of Dick's pet concepts, an hourglass green. Colbert stood on the tee and asked what the ((#* he was thinking? The narrow, long green was fine, but why put it on a hole that normally plays in a strong crosswind? Colbert figured if he would have trouble hitting it (he was top 20 on the money list that year) who else could? So, yeah, I rough in green size (which is, in fact, statistically proportional to approach distance) and I even add a few yards of width in cross winds, and depth on downwind holes (which reduces spin and causes longer roll outs).
I also rough in actual and effective approach shot distance. One reason I am even posting this is I was working on a master plan this morning. One hole, which the pro said is "tough as nails" actually calculates out in an unsurprising manner to his opinion. It is 456 yards, both consistently uphill (including landing zone, which kills roll) and into the prevailing wind. The tee shot is probably 7 yards shorter for elevation difference, 5-10 yards for wind, and another 5-10 yards for no roll, meaning the 290 hitter (I use the average Tour driving distance from the back tees, not the max) would only hit it about 265, leaving another 191 up the hill, again playing effectively longer for wind (another, say 10 yards) and elevation different (only 5 yards this time, for an effective playing distance of 206. The USGA slope guide would call for a green about 77 feet wide, but it has grown in to be about 70% of that. So, I recommend that the green be widened and enlarged, with bunkers moved out a bit to allow a runup for average players, etc.
Again, part of the reason I do some math is to avoid unintentionally building a green no one can hit! I believe it happens more than we think with many architects, although I may have been over traumatized by Colbert's reactions to many KN designs. (I have also found winging it in the field, focused mostly on artwork tends to yield occasional crazy results. And, unlike many here, I do believe that one function of design is to allow golfers a reasonable chance to succeed, although certainly don't feel like I need to guarantee it. When you get right down to it, is it right to design a hole that doesn't allow any kind of shot (within statistical reason) to hold it? In the end, the answer is no, not intentionally.
And again, many here don't seem to think you can pre-think some aspects of green (or other) design and that if you draw a plan, you don't massage it in the field, which just isn't true for most architects. IMHO, some things, as above, and adding desired angle of the green as another big factor, what elevation to set the front (for vision, floodplain, whatever) is best thought out on paper, while the final artwork (and any minor changes to the design as the spirit moves) is subject to continual review and tweaks.
As always, just MHO.