So, there's a couple standards for tagging photos, IPTC and XMP. The benefits of the standards is that many photo sites and software support them. They work by storing the data in the photo (usually jpeg/exif). This is good, because then the data is stored with the photo, they can't get separated.

But... it violates one of the other precepts I like to have, which is don't modify the photo. Manipulating a photo file may break something, may lose existing data, make the photo not compatible with some software, etc. It also make the synchronization problem harder, by which I mean I have multiple computers, with my photos spread out amoung them. My wife and I routinely both upload the pictures from our phones, cameras, etc to our computers, and I try to maintain a central repository of our photos, backups, etc. Some of this is based on the photos keeping the same names, but names collide with multiple cameras from the same company, reseting camera counters, etc. The other thing that stays the same is the file size/checksum. Changing data in the file makes that more challenging, it means I have to do checksums or fingerprints based on the actual image data, and not the raw file data.

I could do a compromise, I could keep an archive of "original" files, and then have a separate or connected archive of "modified" files, that would allow me to keep the write once data, and the "updated" stuff, but that does double disk space... which is cheap, I guess, but our photo data is over 30GB now, its already getting a bit big for having a full copy on our everyday laptops... though maybe that's just an excuse to upgrade.

I need a better solution to the synchronization problem anyways...