I am confused about something with Glowforge. We are encouraged to clean our machines after every 40 hours of use or sooner if needed. We can take the printer head off of the base, remove the top and clean the mirror, remove the lens and clean it, remove the base from the carriage and clean the fan, run calibration tests if the printing is off so that you can remotely diagnose problems with the machine. If we are able to do all of that without messing up the calibration or we can run our own calibration test at home and have you remotely check the results, why then can we not just order a replacement print head and install it at home. We could run a calibration test and if we are good to go begin working. If not, then we could send the machine in to have it repaired. It would save significant time and also save money. Could someone please answer this for me.
(My understanding is that) there is a difference between calibrating the laser beam and calibrating the placement view for the camera. Two totally different calibrations. (We never get exposed to the laser beam calibration.)
If you want to see what calibrating the laser beam looks like, go do a quick search on Youtube. It’s apparently pretty fiddly, takes hours to do, and might be dangerous to the customer. (Lasering through the web of your hand or something. Wouldn’t be good.)
Anyway, I suspect Glowforge doesn’t want us to hurt ourselves. It’s not even offered as an option. 
I have had the same question as the OP. The mirror is held in place by a sponge and the head just magnetically in place. Quite a bit of slop there. The carriage plate can be replaced by the user. Doubt the machining of the head has much variability. I haven’t looked at the schematics but am wondering if the output of some head sensors, (not optics,) are tuned to the specific machine’s original alignment? I’m really pulling at straws to figure this out.
This. Vertical calibration for focus.
They’ve never said, but I did look up what’s involved in calibrating a laser beam before I bought the Glowforge, so I imagine it’s similar to these.
Just looks like more than I care to tackle, even with an engineering background. And that doesn’t even touch the power supply which has to be tuned, and requires other skills I don’t have. 
Yeah, all of that is pretty standard. Not sure it’s a beam alignment issue. I could easily see the carriage plate and mirror having a lot to do with beam alignment but both of those are already user replaceable. @eflyguy suggests it might have something to do with the focus. I can think of how that might be but can’t for the life of me think of why all sensors wouldn’t be nulled in S/W vs. H/W. But then again, somebody thought it was a good idea to install the exhaust fan so that it could not be cleaned, an original filter unit that required lifting 70lbs to replace, and a user replaceable tube as difficult to access as possible.
It actually amazes me how some design choices were brilliant and some were, shall we say, not brilliant.
Beam alignment on a production laser machine using a precision rig built for the task takes seconds, perhaps a minute if it needs manual adjustment.
Of course, we don’t have that, but as-stated, it’s likely not (just) a beam alignment limitation.
If you really wanted to, and cost was no object you could make all of the components of the laser path field replaceable, drop in units. But cost an object for most. That is why instead of working to tolerances hundreds of thousands of time higher we cannot just plug in new mirrors. Errors multiply all along the path, so everything needs to be dialed in very carefully and locked per that individual machine.
There are many calibration procedures that we go through at the factory, and only some of them can be done at home. If we did a replacement without going through the factory calibration procedures, the unit wouldn’t work properly.