A global passive adversary is the most commonly
    assumed threat when analyzing theoretical anonymity
    designs. But like all practical low-latency systems,
    Tor does not protect against such a strong adversary.

To clarify what I think you meant to refer me to, the Tor client actually chooses the three nodes in the path of a circuit, doesn't use two nodes on the same subnet, nor ones the network classifies as belonging to the same "family" (although I'm having trouble determining what this means in practice).

Given that there is a hard limit of three nodes in a route, I'm still have trouble thinking of an adversary that Tor protects you against that a VPN to a jurisdiction of your choosing doesn't, and a VPN is significantly faster...

From what we know the NSA has global coverage with google-style indexing[1] since at least 2012, possibly earlier.

[1] https://firstlook.org/theintercept/2014/08/25/icreach-nsa-ci...

[2] http://www.ip-watch.org/2015/04/24/largest-internet-exchange...

[3] http://www.scmp.com/news/hong-kong/article/1269773/hong-kong...

[4] http://www.newyorker.com/news/news-desk/what-the-n-s-a-wants...

[5] http://www.theguardian.com/uk/2013/jun/21/gchq-cables-secret...

[6] https://docs.google.com/spreadsheets/d/1x6aYnGmbQKzZGLUkWC4m...

It is very much global in the sense that they monitor all communication.

How about actually reading some of the sources that you were just provided with?

Your references seem to talk about major exchanges all over the globe. Practically speaking, because a Tor client can choose the routers for the link it creates, it could choose three routers behind a single major exchange that is monitored (e.g. in Asia or South America), and hence remain anonymous, because the connections between those routers are not monitored.

A correlation attack[1] doesn't care about the intermediate routers. It only requires packet dumps from the entry and the exit node. Both of which, with very high probability, route through networks that are monitored by the NSA.

[1] https://blog.torproject.org/blog/traffic-correlation-using-n...