And a bit is when it communicates one of two possible values. Noise doesn't count as a bit, because it doesn't communicate any information - you can't tell from its value what the original information was.
Same root. A bit of data carries one of two possible values. When it comes to information (which was the original sense) a bit is when it communicates one of two possible values. If you think about it, if a bit is consumed by noise it doesn't communicate any information, because its value doesn't tell you what the input information was. So typically as exposure goes down and the SNT reduces, the number of bits necessary to communicate the image reduces. If you look at the raw file for very high ISO settings the lower bits are all noise, and not contributing to the image.
Unfortunately, the exposure triangle has been used as a teaching tool to persuade many photographers that their cameras work in ways they actually don't. That has real world consequences.
It shifts attention away from f-stop and shutter speed as camera settings controlling important creative elements and exposure, to ISO as the primary source of noise in a photo. Photographers assume that using a low ISO automatically translates to a clean photo; one absent of noise. Then, they make photos using lower ISOs, see noise and immediately assume there's something wrong with the camera. In fact, the problem resides in the teaching tool used by so many to explain photography.
It also tends to block a photographer from taking advantage of options available to them when processing an image in a photo editing app. Suppose you're photographing a high dynamic range sunrise scene; one in which the brightest highlights and darkest shadows can't adequately be rendered in a single photo. Either the shadows are too dark when settings are used that prevent the highlights from being blown out or the highlights are blown out when settings are used that render the shadows light enough to show detail.
In the "exposure triangle" universe, the two outcomes would be described as overexposed (blown highlights) or underexposed (black shadows). In the real world, you can optimize exposure for the conditions. Start by selecting the widest lens aperture (smallest f-stop number) that will deliver an acceptable depth of field. Next, select the slowest shutter speed that will adequately render movement in the frame. You've ensured two important creative goals will be met and have optimized exposure in the process.
In this hypothetical, we can use ISO as an indicator of exposure. It's not directly affecting exposure but serves a purpose as the canary in the coal mine. The available light, f-stop and shutter speed determine exposure. The photo and histogram displayed on the rear LCD are indicators of our success at maximizing exposure without blowing out highlights. Without getting deep into the weeds, there are limitations and potentially real problems with relying on the in-camera JPEG and JPEG histogram as references. That acknowledged, they still can be useful tools.
The shadows in the resulting photo will look too dark and many photographers might feel they've compromised the image quality by underexposing the shadows. To the contrary, I'd suggest exposure has been optimized...at least, the exposure that can be captured in just one shutter actuation. The f-stop and shutter speed used meet the photographer's creative needs. They also maximize exposure within the constraints of the ambient scene lighting.
Putting any more light on the sensor would result in the highlights being blown and important details being lost. In digital photography, we have much less latitude to recover details in the highlights than we do in the shadows. Pushing exposure by more than 2/3 stop will put highlights at risk of being clipped. By contrast, shadows can pretty easily be lifted in post by a full stop and sometimes by multiple stops before image quality really starts to take a hit.
In this hypothetical, if the shadows can be lifted in post by a stop to make them more pleasing, the final version of the processed image looks better and is no worse from a raw image quality standpoint than if the same scene had been photographed using the same exposure settings (f-stop and shutter speed) but with a higher ISO. Exposure was optimized for the conditions and image lightness was fine-tuned in post.
Freeing ourselves from the mythical notion that ISO determines exposure and is the primary source of noise allows for this kind of thinking, image-making and workflow. It opens creative doors that would otherwise remain closed.
Or, as I see people say sometimes, that the'll get noise if you don't 'nail' exposure. Or, that you get noise if you 'push' exposure. Similar to David's discussion, if no-one told you where the noise comes from then you tend to make what seem to be reasonable assumptions.
Especially at the size of the image on the webpage. Sensors with large-scale correlated post-gain read noise blotches that are still visible after significant downsizing are mostly history these days. Intended display magnification can be a useful consideration when making read_noise/headroom compromises.
correct exposure /brightness for a given subject can only be determined buy looking at the subject with your eyes and comparing the image on your viewing medium. no file information is going to tell you what the correct exposure or brightness is, was or should be period.