Just out of curiosity I checked the Winbond 83C42 datasheet, too. I didn't know it even existed. Regarding your problem this...

Winbond Electronics Corporation has developed a fast keyboard controller and BIOS to improve the performance of IBM PC/ATâ 386äD […]
Show full quote

Winbond Electronics Corporation has developed a fast keyboard controller and BIOS to improve the
performance of IBM PC/ATâ 386äDX/SX and 486äDX/SX machines and their compatibles. Hardwire
methodology is used in this keyboard controller instead of software implementation, as in the
traditional 8042 keyboard BIOS. This enables the keyboard controller to respond instantly to all
commands sent from the keyboard to the CPU BIOS.
The keyboard controller enables popular programs such as AutoCADâ, Microsoftâ Windowsä 3.1,
NOVELLâ, and other programs to run much faster.

...sounds like the source of incompatibilities. The're talking here about the interface to the CPU and BIOS, i.e. the signals /WR,/RD, etc. and not the keyboard TX/RX/CLK lines. Not sure every BIOS and decoder logic on each MB can cope with that without being prepared for it. So I wouldn't destroy the 8042 when desoldering it.

Oh.. Ok, it was not my intention to destroy the one on board, but I'll be more careful regardless. I believe the 83C42 came out of a dead Socket 3 motherboard..

That shouldn't cause any incompatibilities. The changes are all internally, the mainboard should not see any difference between this chip and a 80C42.

It will probably be no issue but it could be. I came across so many things that "shouldn't cause any incompatibilities" but did that I get instantly cautious when I read "enables the keyboard controller to respond instantly to all commands". And many people desolder ICs by just cutting their legs and cleaning the holes with a manual vacuum pump which destroys the IC of course. So I just wanted to utter a word of warning. That's all. 😀