So why then does the camera give you the option of creating several, quite different, JPEGs from one underlying RAW file? Are they all perfect renditions of the captured data?
The camera (more accurately PP software) also gives the option of saving several differing forms of RAW - as indicated in saved meta-data files - many of which are also 'not perfect renditions of the
captured data' either. You seem to be under the impression that a RAW image contains ALL image possibilities when it is nothing of the kind - just an option. OOC TIFFs would offer similar options.
In these different files, the RAW data are all the same. It's only the meta-data that are different. Before saying more about RAW below, a comment about ooc-TIFF first: ooc-TIFFs are like ooc-JPEGs, in that they are derived from first converting and interpreting the RAW data. TIFFs have a clear advantage over JPEGs by virtue of their not being only 8 bits (provided that the camera maker chooses to output 16-bit TIFFs, but even 8-bit TIFFs are more flexible than JPEGs). As such, TIFFs offer significantly more headroom for post-processing. Camera makers are fully aware that JPEGs are severely limited in that respect and want to give consumers more freedom to manipulate their images without having them go through RAW conversion first. If anything, when a camera maker provides the option of outputting TIFFs, it's an admission that JPEGs are insufficient in many instances.
Like a JPEG from the same machine, it is only an interpretation of the scene based upon limited user selected parameters and the hardware/algorithms of the internal camera processing system.
Nobody is stopping use of RAW - merely objecting to false claims that a RAW file is some form of raw, unprocessed, data. Or that using a JPEG OOC is any less accurate than one created from RAW (by unmatched third party software conversion) by 'elitist' claimants.
The reason why people often refer to RAW and 'unprocessed' is because RAW data are not
an interpretation of the scene based upon limited user selected parameters and the hardware/algorithms of the internal camera processing system (as you write). A RAW file contains data that haven't been demosaiced yet. Try viewing or printing that. Let alone apply the
limited user selected parameters. You'll get utter garbage.
The RAW data first have to be demosaiced to convert them from the Bayer or 'Fuji 6x6' or whatever arrangement to an arrangement of pixels that is fully two-dimensional (doesn't have any gaps) and contains full color information in
every pixel. As you are probably aware, third-party software manufacturers are still struggling with this step. Fuji had a five-year head start on that. Creating a (positive) image, i.e., one that
can be viewed and printed from the (negative) RAW data, i.e., one that
cannot be viewed or printed, is what RAW processing is all about. Here is where the user-selected parameters are applied, and here is where they can best be adjusted if so desired. The result can then be converted into a JPEG or other format suitable for printing and viewing. Thus the distinction between RAW being a 'development' format and JPEG being a 'distribution' format.
Quite why the RAW monkeys think there is something skilled in sliding a few sliders in PP software is baffling.
When running out of arguments it's always good to go
ad hominem...
I occasionally use RAW on my older cameras as their basic processing engines of ten years ago are more limited in creating spot-on JPEGs in all conditions. But the need, a 'compulsory' use of RAW - as outlined by RAW adherents - I find virtually redundant on most cameras of recent years.
RAW PP is - I'm sorry to inform you so bluntly - is not the dark art of creativity you envisage. Just a simple tool ... and one that the dedicated processing systems of certain cameras can easily match or exceed.
Truly, no current in-camera processing system can remotely match what is possible away from the camera. Admittedly, the in-camera processors are very fast for what they are doing, but what they are doing is very limited. Besides, in-camera software rarely, if ever, gets updated to reflect advances in the field. As you said, you yourself are occasionally re-processing ten-year old images, because of precisely such advances. I don't think Fuji has hit on the holy grail of image processing yet, so it is reasonable to expect that RAW processing will have advanced even more in 2020 to make it worthwhile to re-process some Fuji images from today.
The 'compulsory' use of RAW processing that you refer to has many reasons. For example, for practically every camera/lens combination, just correcting geometric and color distortions in lenses makes it already worthwhile to run every image through a RAW converter, even if nothing else is being adjusted. Also, it is generally accepted that every image from Canon and Nikon dSLRs can be improved by simply applying default settings in high-quality software like DxO. Likewise for other camera manufacturers. Thus, many photographers have a RAW conversion step in their workflow by default. These photographers are not 'monkeys' that are eager to push sliders.
The Fuji processing engine might have closed the gap, but it is clear that there still is a gap (and there has to be based on first principles). Accepting the ooc-JPEGs is like using the AUTO mode on the camera: the camera makes all the decisions, and in the majority of cases the result will be completely sufficient and acceptable. But often it will not, and that's where manual controls and RAW processing come in.