Background
There has been much talk these last few days of what can be done to tackle the problem of the overwhelming deluges of nasty and often threatening abuse that Twitter users can sometimes receive. This abuse can rise to the level of criminality very easily. There have been many suggestions aimed at tackling the social root cause head on, including making it easier to report masses of abuse instead of having to select individual tweets. Some have gone so far as to suggest some sort of annual fee for the use of Twitter in order to cut down on sock puppetry and discourage breaking terms and conditions. Still others have asked for Twitter, like Facebook, to require real names. In my view all of these suggestions would create more problems than what they aim to solve. The problem they aim to solve is not realistic. The world is what it is and human nature is complex and varied. There are nasty people everywhere. It's too much. The problem needs to be redefined.
The problem
The problem is that when a Twitter user is under attack, as happened recently to feminist campaigner Caroline Criado-Perez, she quite understandably fears using the tool that Twitter provides her. She can't have conversations with people because her mentions are flooded with vile abuse. Nobody should have to put up with that.
The solution
My solution to this problem would be for Twitter to introduce something like a panic button that would immediately but temporarily place one's account into a state where all mentions are blocked except for those coming from the followers and following lists. Twitter could also allow the user to open a case and have mentions logged against that case to help Twitter to take action against those who violate the terms. Twitter could also perhaps warn users attempting to put the panicked account into a mention or reply. Perhaps they could ask for confirmation either within the stream or in an email. An alternative would be to actually disallow the creation of the mention or reply. A blocked account is currently not allowed to perform a reply but it can use a mention. The same rules might apply here. The mention would still not be seen.
Benefits
I can see two clear benefits to simply blocking all but the followers and following. This would allow the user to carry on using the tool more or less as normal, while also throwing water on the flames of abuse as the abusers will not be getting the satisfaction of a reaction. Going further, the ability to launch a case with mentions attached would help Twitter to take action against those who violate the terms. This is much more efficient than blocking and reporting users over individual tweets, but it has potential problems so it would need to be implemented very carefully. Warnings, confirmations, and disallowing tweets would cut down on the volume of abusive material.
Conclusion
We're not going to change the world with a social tool. It would be pointless to try. What is important is to give users the freedom to use the tool without having to suffer the indignity of drowning in a sea of hatred. Do this please, Twitter. Thank you.
UPDATE 30 July 2013: I thought it would be obvious so I didn't mention it, but in my proposal you would also have approval over new followers in the same way that protected mode provides this.
UPDATE 31 July 2013: I had another thought this morning. It occurs to me that panic mode combined with a sudden spike in mentions would be a good indicator for twitter to go in and have a look.