Goldy's comment on Twitter was why can't they automate this? This is way too complex of a system.
Seeing some tweets regarding the silence system, and wondering what are the majority of the situations this is trying to correct?— Goldy in Legionland (@BkfstinMG) July 21, 2016
Ok, a little background. I've been an email administrator for a number of fairly large companies. (Only large companies want a specialty email admin.) As an email administrator, I've been called in on more than one situation to enforce blocking of certain types of communication via email. This ranges from the basics of spam filtering and virus/malware blocking to capturing protected health information (PHI) that is leaving the system. I've come to learn that you can't enforce policy solely with technology.
One year I worked for a CEO who was extremely concerned about corporate image. He asked that we configure an email filtering policy that would prevent receiving and sending email containing swear words. Two members of the team created a 200+ long list of words that could be considered swear. Each word would combine into a total score. If it contained any violators the email would be blocked from delivery and someone would receive an alert about the incident. The problem with email correspondence is that we had customers. Sometimes they live in location that looks like a swear word. Sometimes, they have an unfortunate name that looks like a swear word. Unfortunately, we found out that the system we used to filter, pulled from anywhere in a word. This caused things like "Essex" to be captured. The filter was too op. We had to implement a a weighting system, where 0-5 words would not be captured and more than that was flagged and stopped.
From my about page, you'll know that I am dad of 3. In any adult's life, there comes a point when you are inclined to swear. Someone rear-ends your car while your waiting for the light. Waiter drops a piping hot cup of coffee on your lap! Over the last years, I've trained most swear words out of my vocabulary. Instead of something repeatable by little kids, I now use words like 'cupcake' and 'fudge-nutter'. (which actually makes it easier to swear more often. "Oh cupcake, I forgot my phone in the house!" Point being humans can easily replace spam/abuse words. A computer can be set to scan for specific keywords, but can you can't program meaning.
I am a casual player. I often get into random groups with people as they are about to break up for the night. This means the players are already grumpy, and some swearing is happening due to "wipe 18 on so-and-so boss". More often than not, this chatting is over VoIP (Ventrilo, TeamSpeak, etc.). As people sit venting their frustrations, it's easy to join along with the f-bomb festival going on. Unfortunately, with kids either sitting in the other room, or even sleeping on the recliner next to me. I would be typing my complaints into raid-chat. If a system was setup that simply blocked me based on verbiage and not on context, I'd be in trouble. It requires human intervention. I needs someone with the entire story and not just what was said/typed. Sure, for some interactions, it's obvious, but due to the sheer number of subscribers, Blizzard would have not likely caught this user without their input.