Right, but now you’re giving me arguments for using original hardware, not for modifying it.

See, that’s the part that loses me. I use original hardware all the time, I have tons of original hardware and software for a whole bunch of platforms, including ones that are trivial to run in cycle-accurate emulators and FPGA reproductions. All good there. I even have some flashcarts and softmods to allow cross-region usage or to consolidate libraries. No problem with that.

But that is based on using the original hardware, unmodified. Once you start gutting it for mods then you’re working against your argument that complexities and sublteties of original hardware are important. I mean, yeah, I do care to at least have a way to go back to sanity check the sublte ways in which original hardware parses the code in a rom. But for that same reason I want to see how the default composite or RF signal subtly interacts with that output and with a period-accurate CRT display. I want to hear the CD spinning when it’s supposed to spin and the original loading times.

To be clear, I think this is just a case mod, but I’m talking about the modding scene more generally. I don’t see why you would think “total accuracy” is important in the interaction between the CPU, VDUs and RAM but not on the I/O. Wouldn’t the CD drive and the video signal be part of “total accuracy”? Wouldn’t the form factor of the shell and the controllers be a part of that accurate experience as well? If you push me I’d even say I consider a MiSTer FPGA solution with a correct analogue out signal and an original controller feeding into a CRT is far more accurate to the original NES than the original Analogue NT that was made from gutted NES parts, or even than an original console pushed through an HDMI scaler or mod.

I guess there is no accounting for taste, but I do struggle to follow the logic where running the original CPU and video chips on completely different I/O is justified by trying to maximize for accuracy.