[tor-talk] >600 Tor relays without ContactInfo and similar properties

[joebtfsplk at gmx.com]

On 10/23/20 3:09 PM, Georg Koppen wrote:
> nusenu:
>>> [...]
>>> This is a set of over 600 Tor relays that got added since 2020-01-29  on a limited set of hosters.
>>> They have some similarities in their sign-up pattern and properties.
>>>
>>>
>>> Most of them are middle relays (non exit relays).
>>>
>>> total guard probability: 3.6%
>>> total middle probability: 10.1%
>>>
>>> https://github.com/nusenu/tor-network-observations/blob/master/20200129-20200819_unknown_middle_relaygroup.txt
>>>
>>> +-----------------------+----------+
>>> | as_name               |   relays |
>>> +-----------------------+----------+
>>> | Microsoft Corporation |      254 |
>> all of those that used to run at at Microsoft
>> left (or got kicked?) on 2020-09-19 20:00
> Georg: "They did not get kicked out."
>
> Georg
>
Georg, could you explain / expand on your brief reply, if it's relevant
to anything being discussed (or that SHOULD be discussed)?
Is nusenu saying, 254? relays that Microsoft? used to run FOR the Tor
Network, dropped any MS name & re-appeared as nameless relays?  Or can
that be proved?

Where do the other ~ 350 nameless relays come from?
Is it a good idea for 600 relays to pop up in a short time (how short
was it)?

Wouldn't that possibly indicate a single or a few or a consortium of
operators?
If MS ran 250+ relays (sometime), is that good?  I don't think it's a
secret that MS was & likely still is sharing their ISP traffic data with
3 letter agencies.  That "fact" is long past conspiracy theory.

Tor Network / Devs has better ways to track, monitor or kick out relays,
than having no name - or random alpha-numeric names - yes?

What about users wanting to exclude nodes - including but not limited
to, relays with NO name or random character names or by other criteria
that users choose?
How hard is excluding select relays?  Excluding countries is easy.