Sometimes conversations can get intense in the moment. We’ve been experimenting with different solutions & today we are rolling out prompts that encourage people to pause and reconsider a potentially harmful or offensive reply before they hit send. blog.nitter.net/en_us/topic…

4:38 PM · May 7, 2021

16
47
26
190
We learned that: If prompted, 34% revised their initial reply or decided to not send their reply at all. After being prompted once, people composed, on average, 11% fewer offensive replies in the future. If prompted, people were less likely to receive harmful replies back.
2
11
5
84
While we’ve made significant improvements to the algorithms we use to detect potentially harmful Tweets, people will sometimes be prompted in error. We hope you understand no system will be able to do this perfectly and we believe the benefits of this prompt are valuable.
5
4
0
58
Replying to @vijaya
Oh wow, that’s super cool and an interesting read! Thanks for sharing!
0
0
0
1
Replying to @vijaya
That's great! Is the detection algorithm under a free license? I could see how beneficial it could be to make thet backbone of the feature open and the learnings shared with other platforms to increase its accuracy :)
1
1
0
9
It's not, but that's an interesting idea for us to explore!
3
0
0
7
Replying to @vijaya
Will it work for quote tweets as well as replies? Seems like QT's are just as much of a problem if not more as they can bring in a much bigger audience to a petty disagreement.
0
0
0
0
Replying to @vijaya
Well done Team Twitter this is an excellent development.👍 You should ask users to participate in the process by commenting on Tweets which they found offensive and why they found it offensive. This will help to tweak your algorithms even further.
0
0
0
0
Replying to @vijaya
I apprecite this, and I wish we had this before the 2020 US election
0
0
0
0
Replying to @vijaya
Harmful/offensive for which class is the question. As a working class person and a communist, I get enraged seeing the capitalist and petty-capitalist class tweeting "my domestic worker/help." Will such phrases receive any prompts?
0
0
0
0
Replying to @vijaya
This is fascinating. What is the process by which Twitter decides what is "offensive" and it sounds like you are tracking how the tweet sender responds. What is offensive to Gen-Z may not offend a Boomer or Millennial - how do we navigate that? Where can we learn more?
0
0
0
0