Not the parent or agreeing/disagreeing with them, but to your question: if you get creative, there are a lot of things you could do, some more unorthodox than others.
Tongue-in-cheek example, just to get the point across: instead of calling it Developer Mode, call it "Scam mode (dangerous)". Require pressing a button that says "Someone might be scamming me right now." Then require the user to type (not paste) in a long sentence like "STOP! DO NOT CONTINUE IF SOMEONE IS TELLING YOU TO DO THIS! THIS IS A SCAM!"... you get the idea. Maybe ask them to type in some Linux command with special symbols to find the contents of some file with a random name. Then require a reboot for good measure and maybe require typing in another bit of text like "If a stranger told me to do this, it's a scam." Basically, make it as ridiculous and obnoxious as possible so that the message gets across loud and clear to anybody who doesn't know what they're doing.
The people falling for social engineering now won't be protected by this either. You could gate the functionality behind verification of an anti-scam awareness and education training and certification course, scammers would coach people through the entire course and the verification step, and people would still be victimized.