On Thu, Apr 23, 2015 at 9:44 PM, Daniel Tan <danieltan1901@gmail.com> wrote:
Channel mode +T would block non voice or non ops from sending NOTICEs to the channel thus eliminating the necessity to enable the moderation mode.
I believe the focus of flood prevention should be message rate-based controls, not message "form", which is just coincidence. It provides more "useful information" for flood detection if a less-often-used message format such as NOTICE is used, So mitigations/defense should not be used that just attempt to make bad actors "conform" their messages so they look more like good ones; the fact that NOTICE is used provides useful information, which ought to be kept and made part of a detection model, instead of "Attempting to force bad actors to send their floods using PRIVMSG instead of NOTICE". Have we considered that blocking NOTICE doesn't really mitigate the problem, which is flooding, And it's at most a one-line conditional for Bad actors to send PRIVMSG floods instead of NOTICE floods? Therefore... we aren't forcing them to not flood or doing anything that will stop flooding with +T, but instead just forcing an adaptation, And the bad actors are highly effective at adapting and do so rapidly. So the weight of the additional mode in the ircd codebase would seem to exceed the benefits hoped for. Seeing as PRIVMSG and NOTICE are identical, Except for the fact NOTICE is in fact a special PRIVMSG that is to be used for all automatic responses and guaranteed not to send replies, and there's really no strong reason for an evil flooding person to favor one or the other. Meanwhile.... having NOTICE blocked in some areas for specious reasons can very well encourage bad behavior on the part of legitimate bot/script developers, such as using PRIVMSG for automatic responses, which can result in auto responder loops that generate accidental flooding. -- -JH