And speaking of open-source development, would a project/company be liable for every release, including unfinished versions pulled off of their publicly-available versioning system? Or only for “final” versions?

We’d see a lot more perpetual betas, I think.

So you suggest having an OS updater that also pulls updates from an unbounded set of “software vendors”. What’s wrong with that picture?

The risks of having your code base changed at the whim of software vendors is two-fold.

First, by definition, a vendor that has to push fixes is a buggy vendor. If buggy code is pushed out to the whole userbase within 24 hours, the result can be widespread system failures or other DDoS effects.

Second, who is to say what is a “vendor”? How well do vendors protect their update systems against malware hi-jack? If malware enters any one vendor’s “update” channel, the result is instant mass infection that drills right through your defenses.

When there’s a single OS update system that accepts “updates” from anything that calls itself a software vendor, the risk increases, because now there’s a single point of possible exploitation.

Nope, the only real answer is to severely penalize software vendors who screw up. Let them carry the full cost of product recall, rather than get away with popping code on a server and expecting users to pull it down.

That’s the only way you can definitively fix this mess.