I don't understand your argument. You seem to agree that the reason the unlocked house is not like the street is shared social conventions. That house across the street is definitely private property whether it's signed that way or not, and I'm expected to know that because, duh, it's a house. At least, that's how I understood this:
> So the right of the owner of a house to control access to his house is fairly well understood and accepted even in the case where a house might be unlocked or a door left open.
Then you discuss the technical and interface features of websites that differentiate them as analogs of houses and streets, respectively, like whether they have access control (locks). But we just agreed that the technical and design features of the door aren't what make a house not like the street. The differentiating feature of a house is not the security of its door, or even whether it has one; it's that it's a house and we're expected to know it's private. I don't get how that difference is analogous to access controls on a website. What's the social convention that's appropriate for determining whether a piece of information on the internet can be fairly accessed or not?
To be clear, I'm not saying there aren't good answers here (e.g. a house has walls which imply privacy, so you need some analog for walls on your site [1]). Or you could argue that the analogy is bogus (e.g. houses and streets just aren't like the internet). Or you could even argue that technical safeguards are the analogous social convention to private homes (I don't get it, but it's noncrazy). Or you could argue those conventions simply haven't been established yet, and that we should consider there to be no such thing as unlocked houses on the web. I'm just saying you have haven't made any of those arguments.
[1] completely off-the-cuff and, like my other suggestions here, in need of some substance.