None of the bans need to be triggered manually via Player Reports nor enforced manually via a in game review by a GM nor do they need to determine/verify the source of the input to make it bannable...
I am pretty sure Blizzard is recording server-side many attributes of your clients connecting and what environment they are connecting from (like what OS name/version, what machine/host name, what internal/external IPs, possibly full netmasks and trace routes, memory/cpu/gpu/vram/etc, things that denote VMs, etc) and also are recording your actions done in game (and likely the inputs and as best they can tell the source of those inputs) with very granular date/time stamps (at least to the millisecond).
It is very doable with the above information to write automated monitoring routines to detect accounts with simultaneous or nearly simultaneous actions and then automatically feed those flagged accounts to automated warn/suspend/ban workflows that require NO human interaction... I know because I do similar things professionally at work for people/clients connecting to our enterprise databases, enterprise applications, and Single Sign On (SSO) infrastructure for compliance & security reasons...
Frankly if I were Blizz I would just stop all the nebulous and halfway statements and just definitively say all interactions that result in 1 human action causing more than 1 action in 1 wow client are prohibited (whether hardware enabled or software enabled) and enable automated scans for that behavior and workflows to stop it and be done with it... Call it the "1-1-1" rule so it is simple/clear and easy to remember. People would then be forced to stagger actions in a reasonable human speed manner per client and that would end all the drama & confusion...
There are ways to try to get around the above but this would stop the vast majority of bots and abuses...
Connect With Us