*If we have some software that we know to be flawed - in this case, inaccurate shaders operating in non-linear colour space - and we have the means to eliminate that flaw, why in heaven's name would anyone NOT want to do so?

This would be true if everyone was using a perfectly tuned sRGB monitor, like BB and you probably do. I have a two monitor setup, one capable of sRGB setting and a one that hasn't that feature.  I also have two colour-printers: a laser and an 8 colourinkjet printer. But there is absolutely no way to make a image that looks  the same on one of those mediums, inspite of every calibrating tool I used.
I use the sRGB monitor for rendering, because that is supposed to be a "realistic" setting, but as a matter of fact I like the tonemapping of Poser8 more than the GC-feature of Posepro

But there is more: the human eye can see colours by three receptors, red, green and blue. There are more receptors for green and blue, so the eye is more responsive to greenish colours, but.... the red receptors are more active when there's more light. So in broad daylight we see more reddish colours and in the twilight we see more green-blueish colours. Also the amount of red,green and blue receptors is genetical determined, some people even lack some of the receptors and are colourblind.
In other words: there is absolutely no way to determine if  someone sees a colour the same way as somebody else does, except by statistic inquiries. And just like ergonomics, which is also a statiscal "science", the difference  can be huge.
So there is nothing like a correct way to correct  your shaders, everything is literally in the eye of the beholder.

best regards,

Bopper.

-How can you improve things when you don't make mistakes?