[metrics-team] IPv6 votes and consensus synthetic flags

[teor]

> On 9 Dec 2017, at 02:30, Karsten Loesing <karsten at torproject.org> wrote:
> 
>> On 2017-12-08 00:15, Iain Learmonth wrote:
>> [...]
>> 
>> In the patch above, there is a distinction between "announced" and
>> "confirmed". I think perhaps this distinction should be between
>> "announced" and "reachable". This also gives two good properties from
>> which to generate synthetic flags:
> 
> Would it help if we renamed the "confirmed" column to "reachable" in the
> newly suggested CSV file?
> 
> (In fact, the current plan is to name columns "announced_ipv6",
> "confirmed_ipv6_relay", and "exiting_ipv6_relay". But even then it might
> make sense to choose "reachable_ipv6_relay".)

I think using "reachable" is more intuitive than "confirmed".
It's an existing term that describes what we are measuring here.
And it's consistent with consensus-health.

We don't need to repeat "IPv6", because it's already in the description.

> Other than that it sounds like the definitions in the newly suggested
> CSV file will not collide with existing definitions. Is that correct?

Yes.

If we want to spell out the details of the edge case, we can say:
"reachable on the same IPv6 address by enough voting authorities"

> Thanks for working towards more consistent definitions!

I'm all for explaining things once to users, and doing it well :-)

T