I was going to post to the other thread where I discussed this, but it seems on topic for here too.
I finally obtained a workable socket and opened both of my model Ms. The 3151 terminal model had about 5 broken rivets when I opened it and another 5 failed pretty easily from casual finger nail contact. Recall this is the quieter, less pingy board. The 3151 model also used a white plastic plate under the keys, had discrete wiring to the LEDs and "IBM" all over everything. The controller board was attached to the metal plate.
The twangy 1397599 was much cleaner inside (virtually spotless, despite having never been opened). It had NO failed rivets, but 2-3 popped from fingernail contact. They weren't localized in any particular area. The controller was attached to the plastic chassis, BLACK plastic was under the key area (and formed the rivets), a ribbon cable connected the LEDs. Despite being a "blue label" it had IBM everywhere, including the stickers on the controller IC. This may confirm wikipedia's entry that even though it's from 1993 and blue label that IBM still built these 7599s.
So I still don't have a great answer as to why the newer 1397599 is twangy and the older one is not, despite the newer one being in much better shape - rivets included. I may bolt mod the less valuable 3151 model to practice and then do the 7599 just to see if it detwangs it somewhat. This 7599 is in good enough condition that there is definitely something to lose by screwing up a boltmod.
There is a possibility that the key units may be controller compatible with my memory of the connectors (one is at home, one at the office). As I said the controller is mounted in different places though so I don't know if the cabling will reach and be managed properly if I tried a swap.