According to the theory put forward in the second part of http://www.abajournal.com/magazine/article/the_data_question..., nothing.
Which isn't actually true. They didn't have the resources to follow everyone everywhere all the time. And if they were following you, it was possible for you to observe them. There are methods to detect a tail, so you could employ them and then not go to the secret place until there is no one watching. To maintain "technological neutrality" it would require there to exist equivalent digital methods for people to use to prevent being observed.
Kerr's whole premise of technological neutrality is also questionable unless you apply it both ways. But does it really make sense to say that the police can't use fingerprints or DNA evidence because it wasn't traditionally available? How does that framework make any sense at all?
Honestly Tor seems like a great analogy here. It makes it far more difficult to be observed, but it's fallible, as is trying to spot a tail.
Because the consequence isn't that the police can't get the information, it's only that they need a warrant first.
Or to put it a different way, if Tor doesn't create a reasonable expectation of metadata privacy, what would?