I am personally now drawing a clear delineation between projects for my internal consumption (e.g. ansible scripts) and projects that have potential use for the general populace. For the prior, I now host a private Forgejo instance. For the latter, I'll put it on GitHub but mirror it to my Forgejo instance.
I was pleasantly shocked that Forgejo is literally a single binary with a relatively easy config. All my internal services reference my Forgejo instance so, if I need to bail on GitHub, it's low friction for me.
SmarterEveryDay has a whole series[1] on manufacturing that covers how the US is losing/lost the ability to create tooling. I believe the "Smarter Scrubber" episode[2] is the one that explicitly talks about how not only has the US lost manufacturing capacity but it also lost the ability to spin up manufacturing. The tool and die episode also touches on it[3].
I can't speak to any particular dam closure but there's a lot more to maintaining dams than one might believe: https://www.youtube.com/watch?v=XiUOBdEUqjY (Practical Engineering - All Dams Are Temporary)
I'm not a proponent of facial recognition at all and would not endorse it especially in this use case. But, if you believe the 17,000 hours/year number, that's ~$500,000/year (assuming $30/hour, random guess) in potential savings. Are you saying there can't possibly be system that could be designed to take attendance while not further infringing on a student's privacy that's cheaper than that?
What is probably happening is that this is some sneaky way of introducing facial recognition in schools and something benign like attendance was the excuse. That, however, is a separate conversation from my original post of how would you automate attendance without using facial recognition.
The 17,000 number does seem incomprehensibly high unless it's across the whole district. I'm not a teacher but I'm guessing the attendance gets compiled together and stored? Maybe this starts from hardcopy and includes all the work until its in the final digitalized format?
Speaking purely on the technical side of this problem: if you had to design a system to take attendance without some sort of human in the loop (e.g. a teacher), what other workable solutions are there? ID scanned by itself? A student could scan multiple IDs for their friends. ID + PIN? Due to the bursty nature of entering classrooms, that would cause a jam up at the PIN reader. Short of an ID shackle (that's a joke), nothing really comes to my mind that is foolproof.