This is probably the best analysis I've heard.
If the Tor protocol was broken in some way, agencies would be sitting on it to vacuum up as much information as possible. If the underlying cryptographic primitives were broken in any way, that information would be restricted to the highest levels of government and used against state actors.
Darknet busts mean JS browser injection attacks, poor development practices on the server side, or bad human factors (probably this one, from what we're hearing).
But more importantly, increasing the frequency of darknet busts gives a hint of what the authorities think about Tor: they don't have fundamental attacks, so it's best to scare people away as much as possible to discourage use.
That is one of the reasons why you're absolutely not supposed to run a relay from the same IP that you run a hidden service from. Because your IP is published if you do that.
If I were to run such a service, I would want to make sure that the IP that the hidden service is running from has as little connection to me as possible.
[edit] If it were my full time job to locate Tor hidden services, I'm pretty sure I could make a decent go of it. Certainly for a lot of them. Given the resources the NSA and GCHQ have, I have to believe that they can do a much better job of it.
QUANTUM + XKEYSCORE + some MapReductions would make mincemeat of this problem.
http://cybermashup.com/2013/09/04/dont-run-a-tor-router-and-...
And something a little more "official" from https://www.torproject.org/docs/tor-hidden-service.html.en -
"It is generally a better idea to host hidden services on a Tor client rather than a Tor relay, since relay uptime and other properties are publicly visible."
It might be more accurate to conclude that information is a currency in an unregulated market: GCHQ shares with NSA who shares with FBI DITU amongst many other public and private sector customers.
In a world of parallel construction, the most reasonable assumption is that anyone can be privy to anything, or at least information derived from it.
TOR has always been traceable to anyone with enough resources as it makes no attempt to guard against timing attacks.
TOR has never been a secure defense against a collaboration between rich states, especially if you are running permanent services with lots of users.
Also, I do not quite see the intelligence benefits of trying to hide something that is mentioned in the TOR faq, this is not a secret weakness, but something that has been a known weakness since the project's inception.
The only secret revealed here is that the security services have been busy tapping lots of stuff, but that cat has been firmly out of the bag for a while and has since had kittens.
The rest should be possible with more standard police work, once you know where to look (there are probably not more nodes in an AS that you could check them all, disregard known relays etc.).
If TOR was broken, they'd be encouraging its use while secretly mining it for parallel construction opportunities across the board. Instead, we get warning shots.
TOR is fine, but now that we know that the FBI has its tendrils everywhere perhaps we should be a lot more cautious about trusting people we meet online. At the very least we shouldn't be granting administrator privileges to people we don't know the identities of, which is a mistake some of these operators seemingly made.