Sure, in practice it works. I work most of the time at base ISO and I the three parameters will give correctly exposed pictures, if I change speed or aperture.
There's two different things being discussed here.
The exposure triangle is perfectly valid as long as you recognise it's what ends up on the SD card - so it's not "exposure" as such, but the end result uses 3 different things to produce the final output. It's a useful way of thinking about the parameters needed to produce a good image.
Then there's the 'sensor' argument. Mostly a load of rubbish as ISO is set after the sensor, not on it.
And that should be an end to the discussion (but I bet it isn't). 😂
That is my thinking. A beginner mostly only thinks in terms of brightness or lightness. Let them learn the parameters involved in that process and further down the track when they are ready (if ever) they can learn about exposure.
As far as I know, high school science still teaches that electrons exist in shells around the nucleus. It is a valid introductory concept that is enough to then learn about valence, basic chemical bonding etc. Even though it's completely wrong...
We could rename it to "The Brightness Triangle" and "The Exposure Relationship"
exposure triangle. Eg: iso 100 + S+A, iso 800 +S+A, iso 4000 +S+A all voltages changed at photo diode amp level 😎 no different than film. this is 3 exposure triangle examples.
Sigma missed an opportunity when they introduced the SD1 Merrill and their V.5 raw converter. The new converter had a Monochrome (not converted color) Mode - where the three Foveon raw layer values could be combined in any proportions with a Mixer. They already had a three-channel RGB histogram available on the LCD, so a three-channel raw histogram should not have been impossible.