Whonix Wiki Download Docs News Support Tips Issues Contribute DONATE

obfs4proxy in 2022

apparently the developer for obfs4proxy does not want people to use his protocol because “it’s shelf-life expired years ago. No one should be using
it for anything at this point, and no one should have been using it
for anything for the past however many years since I first started
telling people to stop using it.”

source is directly from source code: internal/README.md · master · Yawning Angel / obfs4 · GitLab

this is worth a wiki page on its own tbh

1 Like

Full text:

Maintainer’s rant

Honestly, it is possible to create a better obfuscation protocol than obfs4, and it’s shelf-life expired years ago. No one should be using it for anything at this point, and no one should have been using it for anything for the past however many years since I first started telling people to stop using it.
People should also have listened when I told them repeatedly that there are massive issues in the protocol.

  • Do not ask me questions about this.
  • Do not use it in other projects.
  • Do not use it in anything new.
  • Use a prime order group instead of this nonsense especially if you are doing something new.
  • All I want is to be left alone.

It is currently the only Pluggable Transport and Obfuscation Protocol listed on https://bridges.torproject.org/options/
this means that Tor is still using it and does not have other alternatives currently, there is Snowflake of course, but it does not try to obfuscate as Obfs4 does, as snowflake uses WebRTC https://snowflake.torproject.org/.

This issue would be better raised upstream to the Tor Project that they use a protocol not even recommended by the maintainer.

So if you write a warning about it on the bridges page, you should quote the maintainer’s rant entirely, but also note that Tor is still using it and that is why Whonix will still have the obfs4 bridge documentation.

No, it is much more probable that people will see if it is on the Bridges page.

Search for warnings through the wiki to get a reference on how to do one.

Also, I am not the one that is gonna decide if it is merged or not, it is not be Patrick, so maybe wait his response?

1 Like

you’re right, i didn’t know there was a bridges pages lol

snowflake is actually better against attacks than obfs4proxy, as anyone can install snowflake and have others traffic route through them, unlike running a dedicated obfs4proxy server.

it increases amount of malicious servers (users in snowflake) it is required to pull network attacks

how would one open issue ? i don’t see any place

First, search if that was already reported to them.
Second, https://gitlab.onionize.space/

There is not that much news here…

I’ve added the maintainer’s rant to the technical reasons here:
Hide Tor from your Internet Service Provider: Difference between revisions - Whonix

It however leaves open why it shouldn’t be used. Because it cannot really hide Tor in a threat model that includes an advanced adversary that uses endless data retention with retroactive policing?

As Hide Tor use from the Internet Service Provider is stating for a few years now already:

It is impossible to Hide Tor use from the internet service provider (ISP). It has been concluded this goal is difficult beyond practicality.

Also when reading Configure (Private) (Obfuscated) Tor Bridges that is abundantly clear for years already? There’s a warning box and it’s being elaborated in detail who bridges are for and who not.

I think there problem here are the user expectations.

  • If there are strong hiding of Tor taking into account endless data retention with retroactive policing, then user expectations will not be met.
  • If however the expectation is a simple circumvention of ISP level censorship then many circumvention methods including obfs4 are still functional for many users.

However, when actually reading the Bridges or Hide_Tor_from_your_Internet_Service_Provider wiki page, that should already be abundantly clear. If not, please suggest why the existing wording doesn’t set the correct user expectations.

The maintainer Yawning Angel seems well connected to The Tor Project. Quote Tor Project | People

Yawning Angel
IRC: yawningangel
Author and maintainer of the Linux Tor Browser sandbox.

Related ticket where the maintainer is participating:

snowflake isn’t obfuscated either? Not sure if better or worse matters here. The hard question is, is snowflake claiming to be a censorship circumvention utility or a hide Tor utility? According to https://snowflake.torproject.org/ it’s a utility for censorship circumvention.

If you want to dig deeper, please ask upstream snowflake is designed be used in a threat model that includes endless data retention with retroactive policing.

The more harsh you ask, the less confident will be the answer. “I am one of the few people in North Korea with open internet access. If I get caught, I will end up in prison or worse. Do you recommend me to use snowflake?” - That’s just an example on how to ask in a very drastic way. Obviously please don’t lie about it and waste the developer’s time. I think you’ll have hard time finding a developer taking on (moral or legal) liability by saying “Sure, it’s absolutely safe.” Highly unlikely to happen.

Therefore I think “strong guarantee to hide Tor” (or any sort of traffic) or strong stenography will remain a pipe dream forever.

2 Likes

I already knew obfs4 was detectable for a big enough network actor, I just never read the maintainer of that project to talk about it like that.
Hide Tor use from the Internet Service Provider
This seems more the correct page, yes, all good.

1 Like

Plus the remarks:

At time of writing obfs4 is the only obfuscation protocol listed on bridges.torproject.org/options and there are no better alternatives available for traffic obfuscation, to hide the fact that a user is using Tor.

There is Snowflake (in Whonix ™ wiki) but snowflake is not obfuscated.

Which I think are necessary to explain the problem and answer why there is obfs4 in the anon-connection-wizard of whonix wiki etc.

1 Like