Decentralized vs Distributed

If I'm remembering correctly, that wasn't quite possible yet as it would require them either being in bed with or to have hacked most ISPs throughout the world for that to be effective at a large scale. Also, I'm pretty sure the maintainers of Tor were working on a some way of shuffling the packets and introducing a random delay to stop such an attack from being as effective.

I know Holla Forums will never believe me, but I built them a demo of doing this for VPNs in 1999 while working with one of their groups. It used only two taps in the US and NZ at the time but it worked if you were in the areas covered. It would have worked for Tor with no changes.
They only need to have access to the peering points at the source and destination and they have that for almost all western nations, no hacking required. That's already public knowledge. And those countries are where almost everyone hosts. Hosting in most other countries is very difficult and expensive.
The Tox devs are LARPers. They barely understand networking let alone security.

Decentralization is also resistant to censorship. Given it's shitty to be the victim of censorship by part of your main node provider (should have chosen a more trustworthy one), it is always easy to pick another one. Stuff will still be there, and it would be nearly impossible to censor a whole distributed network unless it is so small there are at best a handful of nodes.

If a node is compromised, it could spy on who you send messages to and how many messages you receive, which would produce some metadata. From an outsider's perspective, however, you are just connecting to a node, just like many others, and said node is connecting to many other nodes. It would be pretty hard for the NSA to make correlation attacks over short text messages if the nodes are crowded enough, assuming they have full control over the package's route.

However, more importantly, you are only showing your IP to a node. Say you want to make some shady shit (like drug dealing) with someone you don't really know. Over Tox, if that person has the means to locate you (which doesn't require being the NSA; even the police could get access to that data), you would get v& in a matter of hours, whereas with decentralization they would have to get two warrants, and one of them may not even be under their jurisdiction. Say you are talking to a skid; you would also be protected from his mandatory petty DoS after you insult him.

It's true Tor mostly solves both of these problems, but that's true for both systems. It still would be harder to perform a correlation attack on the decentralized network.

Oh, that must be reason why they are getting mandatory diversity™.

I've heard "talk is cheap" but I didn't think someone could produce such worthless, hollow prose. You've set a new low.

You're trusting they can get extremely difficult and delicate network encryption right despite them being unable to figure out UDP.
I don't know why Holla Forums does this shit - Josh is another example. You expect magic from morons.

You do know there's a huge difference between correlating traffic that goes to and from a VPN that a few people use, and correlating traffic going in and out of a black box (given how you are only talking about watching traffic entering and exiting) that significantly more people use and has fuck tons of different internal routes that will introduce varying latency. The NSA's internal material from the Snowden leaks even says that they can only correlate some Tor traffic some of the time and that they can't target specific people. If you have actual evidence that the NSA's abilities have or could have advanced significantly since the leaks then post it.

It's not hard or novel. Fitst think of a very simple case where there is no encryption - for every flow, take the hash of the first X packets as the flow's id. See if there already is one, if so, record the endpoint pair. It works fine, yes? You don't even need to do this for every flow, sampling 1% of the traffic would be more than enough to catch you within a few page loads due to how noisy modern websites are.
Now let's consider Tor-style encryption. We can't use packet content as an id anymore but a realtime tool like Tor cannot 'encrypt' inter-packet delay without making it non-realtime and breaking the browser, it at best can obfuscate it a little bit. So we'll use that for our id. Easy problem, use whatever your favorite fuzzy search algorithm is. Again, it can be wrong a lot and still catch you as a browsing session involves hundreds of flows - it will, have many chances.
I've not worked with them since 2000, I don't know what they eventually deployed. Maybe since so few were doing it at the time they didn't deploy it. It wasn't a big internal project or anything, I did it over a weekend to have something to show to the visiting NZ guys and justify my existence (was back in grad school).

No, I don't blindly trust them. I've read the code and I'm well aware it's a clusterfuck and unfinished. That's still 100% more _hard truth_ to go on than Signal or Skype or whatever shit you're trying to steer people here towards with your retarded hearsay. Go fuck yourself.

I knew this would happen. The timings of packets can be used to trace it back to the user.

Fun fact: the shorter the ping the easier it is to spy

It doesn't matter anyway, the NSA has long range wireless sensors that can wiretap your 'analog' monitor's cables wirelessly. Cable shielding is just as important as securing your network and local devices.

NSA could also bring thugs to your home who will make tender love to your boypussy untill you tell them everything yourself, but its not practical for them.