We are talking about
three different things:
1. How the camera determines normal exposure for a give scene.
This depends primarily on metering mode. Whatever and however it meters, it will try to render it as middle gray.
2. Translating that Camera EV number to aperture/shutter pair (and possibly ISO, if not fixed).
3. The photographer's intentions and how things really look (which usually means exposure compensation).
So whether you let the camera figure it out or you figure it out yourself,
normal exposures is just the starting point.If you let the camera determine normal exposure, then you better know how it does it, because no matter what the camera's "rule" is,
some scenes will break it---and cause the camera to guess wrong about normal expoure. Automated methods for determining
normal exposure all assume an "average" scene--though each method has different strengths and weaknesses.
A well-known example is how a backlighting breaks average metering and causes center-weighted metering to under-expose. Matrix
metering (and P mode) does a pretty good job of dealing with most backlighting -- but not other problems, such as a scene that is unusually
dark toned or unusually light toned.
Your camera will render a snow field as middle gray. You probably don't want gray snow.
Also, the camera knows nothing about moving subjects, subjects requiring depth-of-field, confusing backgrounds that would benefit
from bokah, etc. So if you let P mode pick the Aperture & Shutter speed pair, it will just avoid extremes.
Enabling Auto-ISO is a statement that the photographer cares nothing about image quality. That might be reasonable for a photojournalist
or someone doing surveillance work. But in most other photography, it's just a sign of being lazy.
ISO choice depends on how you are going to use the image. For most uses, there is a certain ISO -- maybe 400, maybe 200, maybe 100 --
above which the image will be too degraded for its intended use. (Given that all consumer digital cameras are miniature format or smaller,
you don't have a huge amount of resolution to throw away. You are shooting the same format as a Kodak Instamatic, or even smaller.)
There are really very few uses for an image taken at ISO 1600 --except if it's a photo of Big Foot.
Everything the photographer does -- what camera and lens he selects --where he goes to shoot -- is determined by the final image he is trying
to achieve. There is no other rational approach.
When you go hunting, everything is determined by the type of game you are trying to bag. Same reason.
Talking about an "exposure triangle" is like talking about a food triangle: meat, vegetables and nicotine. If you consume enough nicotine,
it will kill you. Perfectly usable images can be created at the extremes of the aperture range and at the extremes of the shutter speed range.
Buy one always pays a price (in terms of image quality) for turning up the ISO.
The photographer has to take charge because he is the only piece of "gear" with a brain. The tripod doesn't have one, and neither does the
lens or camera. The camera is capable of executing algorthims contained in firmware, but it has limited information. It doesn't know what
it's looking (and if you want proof--or a good laugh--try enabling face recognition). It doesn't know how light- or dark-colored anything
actually is -- only how they appear in a given light.
You know that snow isn't gray. The camera says "what is snow?"