Specifications and compatibility, cameras and lenses version

In the beginning, the connections between cameras and lenses were entirely mechanical and relatively simple. I believe Nikon started with a mechanical lever to control the lens aperture, and even managed to do their early autofocus lenses with mechanical screw drives. Those days are long over; modern interchangeable lens cameras conduct sophisticated communication with their lenses about all sorts of things. And famously, how these communications work is proprietary and private information. Camera makers sometimes license third party lens makers and share this information with them, but that's it. Of course this hasn't stopped outside third party lens makers from reverse engineering the (current) protocols and implementing them (although they sometimes get sued over it, I believe mostly nominally for patent violations).

You could look at this situation and say that camera makers are only doing this for crude commercial reasons, namely the licensing revenue they can extract from official third party lens makers. In the broad open source way they'd be better off overall if they made their lens mount communication protocols public and allowed a wide profusion of third party lenses to bloom. As it happens, I think that things are not so straightforward.

When photographers buy an official lens from the camera maker, they expect that lens to work for decades to come, for as long as the camera maker supports the lens mount (and the lens doesn't suffer a mechanical failure). They expect the lens to be forward compatible with future cameras, or to put it the other way, they expect future cameras to be backward compatible with their current lenses, even if those lenses are ten or twenty years old. This is very much to the benefit of camera makers, because the collection of lenses that photographers have built up is a major force in keeping them buying your cameras instead of someone else's. If a photographer moves to another camera system, they don't just have to replace their camera; they also have to replace all their lenses.

(Photographers also generally expect new lenses to work on older cameras in the same system, but I think that's less of a hard constraint. A new official lens that doesn't work on older cameras just costs the camera maker some up-front sales, it doesn't anger people as much as some of their existing lenses not working with a new camera.)

When you buy a lens from an unauthorized third party lens maker, one of the things you're implicitly giving up is this more or less guaranteed future compatibility. The third party lens maker has reverse engineered the current protocol, but it has no special insight into future changes to the protocol, how to interpret them, or what bits its current lenses should ignore as opposed to considering a sign that something is wrong. As long as the camera maker doesn't break official lenses, it's free to change the lens mount protocol to suit its own needs without caring about how this affects unauthorized third party lenses. And the camera maker is in a position to know what's genuinely reserved for future use (or currently ignored by all official lenses in practice and open to reuse), and what's not. Over the years, camera makers have found all sorts of clever ways to evolve their lens mount protocols to do and control new things, and not all of these have been compatible with unauthorized third party lenses. When this happens it's generally your problem, not the camera maker's problem.

(Sometimes the third party lens makers have been able to fix their lenses with firmware updates. Yes, modern camera lenses have firmware; it's computers all the way down.)

If a camera maker officially publishes "the specification" of its lens mount protocol for general use, photographers will expect third party lenses made to this protocol to keep working into the future, just like official lenses. This is already constraining the camera maker's future evolution of its lens mount, but on top of that, "made to this protocol" is an abstract thing and now Hyrum's Law has entered the chat. It's very likely that third party lens makers will take shortcuts, make mistakes, or simply not fully understand the protocol specification (no matter how clear you make it). The resulting lenses will work on your current camera bodies with the current protocol, because that's what the third party lenses get tested and used on, but when you release new camera bodies with a perfectly allowed evolution to the mount protocol, some number of these third party lenses will now malfunction. When this happens, photographers will not blame the third party lenses, they will blame your new camera bodies; those bodies will be slammed for being "not fully compatible with all existing lenses", breaking your implicit or perhaps explicit promise.

(The camera maker could operate a testing and certification program, but it's going to have to charge third party lens makers for this and not all of them will sign up. Plus, there's a limit to how much you can torture test the specifications compliance of something in the face of mistakes.)

If you're a camera maker and you're looking at all of this, you might reasonably conclude that it's more of a headache to make your lens specification open and public than it's worth to increase the number of lenses in your system. This isn't an insurmountable problem and there's at least one example of a relatively open lens mount standard with good long term cross compatibility (the Micro Four Thirds system), but I don't think it's an obvious, open and shut thing that the benefits to a camera maker are worth the potential long term costs.

(Applications to computing standards are left as an exercise to the reader. It's obviously related to how closed protocols can evolve more freely, which was Signal's official reason for dropping federation, but it's not quite the same. Camera makers can't go back to change how existing old lenses work; instead they have to fit new protocol things through practical holes in the old ones.)

(This was triggered by a Fediverse discussion where I realized I had views that didn't fit into a Fediverse post.)