How could this be tested?
The only things that I imagine can be tested for is a larger pixel image size
and a larger number of colors used within an image's palette. Both values would have to increase in order for an image to be considered newer.
But even then we're just guessing. A new cover, at a higher pixel size, consisting of random colored noise spread over several layers is going to meet the criteria above and assumed to be newer. Yet it doesn't contain any content - there is no visual or textual data conveyed to the reader what-so-ever.
There are no other quantifiable factors that can safely be used to compare against:
- Date? It's easy enough to use a utility that will change the date and time to something newer/older (as needed for tested).
- File size? Increasing the image's pixel size and/or layers, but not actually changing anything else within the image, would result in erroneous results. But this also assumes that newer covers tend to get more complex and this frequently is not the case; masterwork compilations tend to get simplified covers.
- Filename? Most book covers are named "cover.ext" or similar. I don't think I've ever seen a book cover with version info in the filename.
- Metadata? Theoretically possible, but it is unlikely to have been included. Even when it is, it tends to be generic creator info that is useless for comparison.
Note that I'm not adverse to the idea of being able to compare included cover images. Just that I cannot think of a viable means of
automating the test and consistently getting the best possible result of the available choices.
The best that I could imagine is something like a thumbnail viewer option with a multiple choice, prioritized cover listing. This would allow users to add or remove cover images while also allowing images to be organized into prioritized viewing order. Not sure how ereader software or hardware would respond to all these images, but it might allow for international editions, reprints, etc without changing the metadata.
Of course, technically, this would be flat out wrong because any change to a book results in updated metadata (ie, a new record). But some users don't care about this so much. Wanting the newest, most error-free text, maybe a particular cartography style for the map, and the cover art from when the user first read the book.
Of course, it's quite possible that this is more of a Sigil feature request... but that isn't likely to happen.