[tor-dev] Tor Relay Setup Wizard
karsten.loesing at gmx.net
Thu Jul 14 05:56:32 UTC 2011
On 7/14/11 5:52 AM, Damian Johnson wrote:
>> In that case,
>> setting "Internal Relay" as the default could be problematic. If people
>> use arm on their clients, they shouldn't be tricked into becoming a
>> relay only because that's the default. They should know what they're
>> doing when setting up a relay. But 90% of all users go with the
>> default, so if "Internal Relay" is the default, they'll just pick that
>> and hope for the best. (Ignore this comment if the wizard is explicitly
>> called "Tor Relay Setup Wizard" and if people need to actively start it.)
> I disagree. For the network to scale it needs some portion of its
> userbase to be relays. There's certainly use cases where that isn't
> practical (either when it's a burden or they need to hide the fact
> that they're using tor), but for just about anything else operating a
> middle-hop relay is an easy and complaint-free way of helping. I kinda
> like having people opt-out of being a relay since it makes them aware
> when they're using the network without contributing to it. Also,
> weren't we talking earlier about making people bridges by default in
I didn't follow the whole "make all users relays or bridges" discussion.
But I think there are a few arguments against turning clients into
relays by default:
- If people aren't really aware that they're becoming a relay, we'll end
up with a lot of slow and unstable relays. I imagine there could be
hundreds if not thousands of those relays if we start shipping packages
with arm. This could become problematic for the directory system at
some point. Sure, we could start excluding relays from the consensus
that don't contribute much. But why create a problem (or making it
worse) if we can avoid it?
- Another effect of surprising people by turning their clients into
relays is that they might stop liking Tor. They shouldn't learn that
they're contributing to something they only discovered a few hours ago,
only because it eats all their bandwidth. Even worse, what if people
try out Tor, use it for a bit, let it run, and learn at the end of the
month that they exceeded their bandwidth quota and have to pay for it?
It would be better to make them like Tor first and then ask them to give
- If you make Client at position 4 the default and have three more
options at positions 1 to 3 to contribute to the network, a fair amount
of people will already feel bad that they don't contribute. If we can
give these people an easy way to contribute to the network by re-running
the wizard, we win.
- I'm not sure if making people bridges by default in Vidalia is a good
idea, either. I think you're referring to the Tor proposal where
clients measure if they're stable enough to be a bridge and whether
they're reachable from outside and then turn into bridges automatically.
I'm not a big fan of this idea, but at least it makes sure the new
bridges or relays will be useful for the network. It also means people
had the chance to like Tor enough to accept that they're now giving
> That said, I'll change the default if people would rather it be something else.
Sure, having more opinions here would be useful. This discussion isn't
even arm-specific, but could apply to Vidalia or packaging in general.
I'm curious what other people think, too.
More information about the tor-dev