I think we'd also need to figure out some durable and stable way to reach a conclusion on "when should the software be published out of escrow?" that handles a bunch of the various edge cases. "What happens to devices that are one-time programmable? What devices are in-scope/out-of-scope? Does this apply to radio firmware as well as general CPU firmware? Is the software license changed alongside the release of code from escrow? Are signing keys also released? Is code released from escrow just because some individual use case is no longer supported by the mainline firmware? [Is a disagreement with a product decision enough to release the old code?]"
I agree as well, though I don't think we need to figure out all edge cases before the legislation is viable. All we need to do is allow any person who purchased said software a private cause of action in which they can petition a court to release the code. Then a judge could decide based on the merits of the person's need whether the code should be released or not.
I think that situation exists now, which is the essential root of the problem.
It's too expensive and too unlikely to succeed, but I could sue Chamberlain now arguing that they have breached an implied contract and that the remedy I seek is for them to open-source their code.
I disagree; I believe any lawsuit brought against Chamberlain today would be dismissed for lack of standing. Further, even if it wasn't, I think you would have a very hard time convincing the court that open sourcing their code is a reasonable remedy.
Best case, I think you'd get your purchase price back. I'm not sure how you'd argue that remedy is insufficient, either - hence why my preference is to have the cause of action written into the law we're imagining here. It'd be even better if we can write in that the remedy for a degradation of the service is an open mechanism by which the user has sufficient level of control as to recreate their desired functionality.
I think we'd also need to figure out some durable and stable way to reach a conclusion on "when should the software be published out of escrow?" that handles a bunch of the various edge cases. "What happens to devices that are one-time programmable? What devices are in-scope/out-of-scope? Does this apply to radio firmware as well as general CPU firmware? Is the software license changed alongside the release of code from escrow? Are signing keys also released? Is code released from escrow just because some individual use case is no longer supported by the mainline firmware? [Is a disagreement with a product decision enough to release the old code?]"