4 ms·

Exactly this, and what's so mind boggling is that this is not a new problem by any means. I'm sure that among twitter's worries is how to effectively scale an a

Exactly this, and what's so mind boggling is that this is not a new problem by any means. I'm sure that among twitter's worries is how to effectively scale an abuse prevention operation, but there are some pretty easy automatic filters that would go a long way. A few that come to mind (provided accounts are tied to phone numbers): 1) A block that prevents the harasser from even @mentioning the harassee in any way. 2) Automatic disabling of accounts that are blocked by more than two users whom the harasser explicitly @mentioned 3) A "dog house" mode which automatically disables distribution of a user's tweets who have been reported as harassment without explicitly telling them, leaving them to harass nothingness until they get bored and leave.

josephlord 12y ago
Number 2 could probably be abused. Get several people to respond to somebody and when they reply block them. Number 3 needs to be applied carefully as wrongfully banned people (it does happen) will be harmed.
- newhouseb 12y ago
  (replying to both since they're effectively the same comment) Abuse of such things is definitely a challenge, and there is finesse on choosing the right thresholds. I doubt twitter can fully escape manual moderation and as such one of the goals of a system like this would simply be to reduce the number of situations that require human intervention. Undoubtedly, there will be false positives, but abuse of the abuse system will likely be rarer than abuse as a whole and if a human moderates and finds such meta-abuse it would be grounds for the strictest means of blocking twitter can provide, thereby hopefully incentivizing against it.
pgeorgi 12y ago
Blocking is an issue, since that can be weaponized way too easily: "I don't like your non-harrassing opinion, so I'll get two buddies to engage with you, then block you once you respond." The "dog house" could even work for cliques: If a certain number or percentage of your peers (as determined by follow relationships and metions) block a user (for themselves), the platform could make them disappear for you as well (unless you engage with them first). That should be good enough to keep adverse groups separated while avoiding the issue of "I'm not abusing them, I'm calling them out" that's already mentioned elsewhere.
prawn 12y ago
What if a user could toggle a setting through which users of a certain type couldn't @reply them at all? e.g., accounts created within the last month, or accounts with fewer than 50 tweets, etc.
A_COMPUTER 12y ago
1) kind of already happens, they can mention but it's not linked, which is laughably enough to prevent a significant degree of mobbing by their own followers. 2) Someone in the atheist-skeptic community made a tool called the blockbot that ostensibly was to automatically block people from a list that you never want to talk to in the first place because the blockbot operator already identified them as trolls. Unfortunately the main reason was that the developer figured out that mass blocks caused an automated response mechanism to suspend that users. It was really an offensive, not defensive mechanism. This has been somewhat changed, but the reason I bring it up is that this has been tried by Twitter and unscrupulous people used it as an attack tool. 3) is similar to 2) but I suspect they are already doing this in some cases.

josephlord 12y ago

Number 2 could probably be abused. Get several people to respond to somebody and when they reply block them. Number 3 needs to be applied carefully as wrongfully banned people (it does happen) will be harmed.

newhouseb 12y ago
(replying to both since they're effectively the same comment) Abuse of such things is definitely a challenge, and there is finesse on choosing the right thresholds. I doubt twitter can fully escape manual moderation and as such one of the goals of a system like this would simply be to reduce the number of situations that require human intervention. Undoubtedly, there will be false positives, but abuse of the abuse system will likely be rarer than abuse as a whole and if a human moderates and finds such meta-abuse it would be grounds for the strictest means of blocking twitter can provide, thereby hopefully incentivizing against it.

pgeorgi 12y ago

Blocking is an issue, since that can be weaponized way too easily: "I don't like your non-harrassing opinion, so I'll get two buddies to engage with you, then block you once you respond." The "dog house" could even work for cliques: If a certain number or percentage of your peers (as determined by follow relationships and metions) block a user (for themselves), the platform could make them disappear for you as well (unless you engage with them first). That should be good enough to keep adverse groups separated while avoiding the issue of "I'm not abusing them, I'm calling them out" that's already mentioned elsewhere.

prawn 12y ago

What if a user could toggle a setting through which users of a certain type couldn't @reply them at all? e.g., accounts created within the last month, or accounts with fewer than 50 tweets, etc.

A_COMPUTER 12y ago

1) kind of already happens, they can mention but it's not linked, which is laughably enough to prevent a significant degree of mobbing by their own followers. 2) Someone in the atheist-skeptic community made a tool called the blockbot that ostensibly was to automatically block people from a list that you never want to talk to in the first place because the blockbot operator already identified them as trolls. Unfortunately the main reason was that the developer figured out that mass blocks caused an automated response mechanism to suspend that users. It was really an offensive, not defensive mechanism. This has been somewhat changed, but the reason I bring it up is that this has been tried by Twitter and unscrupulous people used it as an attack tool. 3) is similar to 2) but I suspect they are already doing this in some cases.