okay, took some time this morning to revisit the process!
here's my setup in my bathroom:
i've obviously got my camera, homemade softbox (on a really wimpy Pentax AF160 i got for free, supported by a rubik's cube on top of the focusing screen), single halogen light source, 1:8(ish?) ilford multigrade developer, a water bath for stopping and washing, and rapid fix.
the whole approach is very imprecise, because i'm dealing with expired paper and didn't want to get my phone to use as a timer while i was in there banging the portraits out. i metered the bathroom light at EV 3.5 for EI 12 (which i estimated a full stop slower for the paper, roughly--so more like a true EV 1.75 for EI 6) with my handy little incident meter, and the softbox i'm eyeballing off the exposure chart on the back as a half-stop under normal ( EI 25 @ f8, 1m = EI 6 @ f2.8, <1m + power reduction from diffuser). the flash exposure came out fairly close to true, though a little under. however the mostly-flash exposure (shortest shutter-open time) came out the nicest.
the procedure for each was to expose, fog the plate a touch with light through the crack at the bottom of the door (~1 s), develop to completion in darkness (~1 min), flash the bathroom light to initiate reversal (0.3 - 0.1s, you get a good look at the negative at this point), develop to completion (~50 s), put in water stop bath for ~10s letting the developer work its way out (and at this point your eyes have well adjusted again and can clearly see your plate has darkened), fix for whatever feels comfortable and wash in the sink for a couple minutes.
these are the fruits of my efforts today, arranged top to bottom, right to left. the first was a gross underexposure (i had accidentally left the stop at f8). the second shot was as lit as i could possibly get it for ~4s exposure and the flash, and reversed with ~0.25s of bathroom light--it came out with badly blown highlights on my nose and forehead that did not reverse. i believe that this unwillingness of fully developed halides to reverse means there are very hard limits on the maximum contrast ratio that can be captured with this process.
the rest of the shots in the right column were experimenting with exposure and flash time. when i do proper scans of these images i'll make a post of my recollected exposure data (whatever that'll mean) in a later post.
the two (i feel) more successful exposures on the left column were the last two, and were a bit different. these exposures were for shorter times, roughly 1.5s, and when flashed with the overhead light showed almost no density at all as a negative image. in fact, for the top image on the left the paper was completely white, i couldn't see any density at all when i flashed it, and agitated it for a good minute and a half in the developer hoping i could get something out of it. the lower image was about the same for exposure and density when i initiated the reversal, and just trusted that i'd see something. obviously they're much lighter in tone, but i'm happier with them as there are very little to no unreversed highlights. i believe this is a factor of both shorter exposure times to avoid blowing highlights on the front end, and more gentle reversing exposures.