Suggestion/Rant: Allow suppressing @mention notifications

api

#1

This is quite a long shot – and I’m not sure this is the best place for this, but I would like to suggest a minor update to the Twitter API.

It would be a great deal of help if developers were able to suppress @mention notifications when posting a tweet.

Here’s a use case. Some time ago I created a bot that crawls a website (belong.io) that collects interesting tweets and then retweets them. The project is open-sourced at https://github.com/fourtonfish/belongbot4000.

Originally, the Twitter bot would just retweet the tweets it would find, which is obviously not an appropriate behavior. I tried to come up with alternatives, but they are all very limiting. For example, the current version of the bot collects all tweets posted in the past hour and posts them as a collection – but those are only limited to 100, so I have to delete the old ones.

Honestly, the rule of not retweeting random people is arbitrary, because to resolve it, I’d just need a way to suppress @ mention notifications (sending disable_notifications: true?) to people who don’t follow my bot. Or maybe bots shouldn’t even be able to send notifications to people who don’t follow them, by default.

And just to quickly address potential alternative solutions: simply tweeting out the content of those tweets feels like stealing, tweeting just the URL the person posted makes the account look a bit spammy, and including the person’s username without an @ creates a sub-par experience for those who’d like to either respond or see more tweets from the same person.


#2

I like the collections idea - but like you say, it is very limiting to only have 100 of those.

I know belong.io is an aggregator, and this might sound like making an aggregator from an aggregator, but I noticed some collections only include 1 or 2 tweets: maybe a good thing to do would be to only create a collection once a minimum number of tweets have been collected, rather than each hour? Like a “latest Top 10” collection that will post at irregular intervals, or every 8 hours?

To avoid spammy behaviour you could try using both collections, and tweets - but only mentioning or quoting the original tweet or user once it reaches a certain threshold of likes / RTs (since you’re collecting tweets already, you could check counts after a few hours and filter the most popular ones? or alternatively limit mentions of a given user to once every week or month at most - and go with just tweeting URL if there’s more links from the same user)


#3

That’s a pretty good idea, thanks! Though I still think using collections is less than ideal, I’d prefer to just see the tweets posted directly.

I do wonder if there is a good reason for my suggestion not to be implemented. I can see someone arguing that people should know when someone else is talking about them, but it’s trivial to remove the @ from someone’s username, and that would be okay.

Or you could just post people’s tweets on a different website, much like belong.io does.

So the problem really isn’t retweeting what other people say, just that there is no way to suppress the notification.


#4

Any chance I could get an official response from the Twitter developer team?


#5

Hey @stefan thank you for the suggestion! Apologies but a bunch of us have been taking vacation… busy year! :slight_smile: oh, and happy 2017!

With suggestions of this kind, I would ask you (and others) to understand that product and platform planning is not as straightforward as “this seems like a great idea, let’s do it!” when Twitter is at current scale and complexity. We want to test, ensure that a change is scalable, works across platforms, and doesn’t impact user engagement in a negative manner (for example).

That said, I think you probably already saw some changes to control over notifications during the past twelve months, and we are always thinking of how best to enable additional controls to improve user experience. A change like the one you’re proposing may take a while to happen for various reasons - rolling up with broader safety and abuse changes, for example.

On the developer platform side, as you know at the moment there’s no current API around notifications specifically, so there’s not so much to comment on from that perspective.


#6

Happy New Year, @andypiper :slight_smile:

Yeah, I expected hearing this, although it’s still better to have an official response. I’ll be on the lookout for any updates in this area :slight_smile:


#7

I’d prefer the current behavior to remain. The benefit to botmakers of the ability to control the notifications that their bots generate is, I think, a relatively small use-case when compared with the vast majority of notifications that your average Twitter user receives (I can’t think of many other benign use cases). As harassment is a (unfortunately) continuing problem on Twitter, being able to stay on top of your notifications is important for the personal safety of a lot of folks.

If, for example, a high-follower user were able to conveniently, and without generating a notification, direct their followers to a specific person’s account with the intent to harass in the middle of the night or some such, this change would allow that behavior. (With the proposed change, it would in fact be of little difficultly to create an entire Twitter client dedicated to discussing folks surreptitiously.)

Even without malice, a retweet or mention from a high-follower account can generate a lot of unwanted attention - it’s often helpful to have some kind of warning of that and be able to take your account private before more people jump in and dissect the rest of your timeline.

As you note, @stefan, re: the person being discussed:

it’s trivial to remove the @ from someone’s username, and that would be okay.

Yes, it’s trivial to do so, but I’d argue that that trivial barrier dissuades a raft of (often ill-meaning) Twitter users who cannot be bothered to search for a user’s profile or the tweet in question. In fact, you acknowledge this, stating:

including the person’s username without an @ creates a sub-par experience for those who’d like to either respond or see more tweets from the same person.

This is especially true if an old tweet (usually taken out of context) is dredged up to, let’s say, ruin someone’s day. The majority of users aren’t going to bother dealing with a hundred “pages” of infinite scroll just to quote tweet that with something offensive.

Just my thoughts on the matter. If this were 2009 Twitter, I’d probably be with you on this, but it’s unfortunately not.


#8

@jcbl You made some really excellent points, thank you for sharing them!