So thaaat's what I was doing wrong.  Going to the store right now to pick up some Guinness.

i have a whole routine that involves not shaving for a period of time, abstaining from sex, holding a specific sized wad of chew in my left cheek, and calculating barometric pressure within each cylinder prior to torquing the adjacent studs...

...once it is torqued at initial install, I like to warm the engine up by running it with absolutely no coolant in it for a few "moments" and then cutting the engine off to let the heat normalize and wait a while to attempt to evenly distribute itself. And then I go back and retorque to spec again. I always get more out of it even without needing to back off. My theory/philosophy is that the varnish/coating on the head gasket is softened by the heat and the retorque helps to squish it into Thomas' nooks and crannies and provide a better liquid seal. Then I let it cool completely prior to filling the whole thing with proper coolant. The arp studs can take a lot more torque than the block and head can take, and I figure the studs are asked to perform their job while they are up to temperature anyway, so torquing them while warm seems to make more sense to me and to balance pressure from the shorter front ones vs the longer rear ones than just torquing them up cold...


...And then, for my final round of nonsensical ritualistic behavior, I put on the same pair of socks that I was wearing when the head gasket originally blew/let go (or, if it didn't blow and I was just assembling a new build, I wear NO socks,) and then, after drinking precisely 4/5ths of a pint of Guinness, I urinate over the entire block to seal it up sorta like the christening of a great ship.

Rather than back them off 1/2 turn, I decided to confirm the torque.  Some were snug and didn't move.  Others, maybe 1/8 turn.  Surprising..  These are ARP studs/nuts that have been used before and should have been stretched.  Used a torque wrench of known accuracy and current calibration.  But then I suppose there is +/- calibration accuracy to consider.

I check them again after the first 500 miles. But as Alex implied, sometimes it’s better to let a sleeping dog lie.

If you are running ARP hardware, I would follow their specific instructions.

Torque to 48lb-ft in the proper sequence.  Leave alone.

Worked for me for 25+ years.

When I install the cylinder head, I torque to the lubricated ARP studs/nuts 20 ft/lbs in the proper sequence, then to 45 ft/lbs.  Run to temp, let cool completely, then re-torque to 45 after backing off the nuts 1/2 turn. 
Does anyone else use a different method and if so, what?