Twitter might be changing some of its rules

And it could be a step forward. Randi Harper notes that sealioning is now against the rules, although I read it more as a rule against general obsessive behavior.

…well, let me quote the Twitter Rules directly:

If you send large numbers of unsolicited @replies or mentions;

Unsolicited replies or mentions, you say? Sounds a lot like the definition of sea-lioning. YMMV, but I recently had success in submitting a report about a user who had been almost exclusively tweeting about me for weeks.  As far as I know, I’ve never once mentioned her, and I’ve had her blocked since the first time she popped into my notifications. She’s been subverting this block to screencap and discuss many of my tweets, consistently @mentioning me, and her followers would reply and include that @mention. This was a sustained campaign with no response, and I think that might be the the key. I asked in the report for the abuse team to consider all the tweets in her recent timeline, and they asked her to delete a good portion of them. It appears they targeted the tweets she issued where I was @mentioned. This is the first time I’ve heard of Twitter responding to context.

Oh, man, I read that and realized that there are a whole lot of people I never mention, but who have embarked on long term campaigns to yap ceaselessly at me and other people on FtB and Skepchick. They’re going to have to change their behavior now.

Well, that is, if Twitter actually enforces this new rule. I’m not confident that they will.


  1. trollofreason says

    Oh noes, I might be unable to slightly bother you anymore with my generally affable yet cynical demeanor on Twitter!

  2. says

    No, I don’t mind being addressed on Twitter, so I don’t complain about affable tweets in my general direction.

    The problem is that there is a small corps of haters who like to send me messages that are anything but affable, every day. I block them, but as pointed out above, they still leak through because they have mysterious unknown followers that repeat their messages of hatred for them.

  3. s3m3rs says

    @3 – their poor record is, in my opinion, Precisely Why this is happening.

    I think I saw a screencap yesterday of a discussion of how ‘gators can use the ADA to get around the ‘block bot’.

    That crowd has not spent time asking why ‘ethics’ demands unethical behavior to tweet at people who have told them to back off enough times that they are now blocked.

  4. says

    Man, the worst time I was ever sea-lioned on Twitter was by vaping (e-cig) advocates after I posted a link to an article from the AACR they didn’t like. Holy hell. Dozens of oh-so-polite (intermittently) vapers telling me how wrong I was and purposely including a bunch of their friends in the conversation in order to get them Tweeting at me too. Initially I made the mistake of engaging them but I soon discovered that they are relentless and they call in their fellow vaping advocates to swarm. I never blocked so many people in a single day as I did that day. It’s almost to the point that if I see the word “vaping” or any mention of anything having to do with vaping or e-cigs in person’s Twitter profile, I seriously consider preemptively blocking that person.

  5. says

    They’ve not looked at context much before, but if you report people for @’ing you then they get suspended eventually. That has been around for a while. So if you reported a certain Irishman he would get suspended if he kept @’ing you and trying to get attention via replies …

  6. says

    Gizmodo’s take on the new TOS is highly critical.

    Twitter’s New Threat Reporting Tool Is a Useless Punt


    The new tool means that if you’re getting harassed on Twitter, you can get a summary of your complaint emailed to you. That’s it. That’s the whole service offered.

    All the tool does is save people the few seconds it takes to screenshot complaints to send to police. The user is still responsible for forwarding that email to the appropriate authorities. Twitter won’t send it along for you or take any steps to make sure law enforcement prioritize or pay attention to the complaint in any way. And if those authorities do decide to look into the threat, they still have to go through the usual channels to receive additional information about the harasser from Twitter.

    I asked Twitter whether it was planning to work with law enforcement to bulk up this tool, and whether it planned on contacting law enforcement to alert them to particularly aggressive threats directly. “We don’t have anything to share beyond the content in the blog post. Our guidelines for law enforcement (linked in the blog post) explain what private user information Twitter has, and how authorities can request it,” a spokesperson told me.

    It’s easier to claim to have done something than to actually do it. If Gizmodo’s writer is right, then that’s what the “new rules” are.

  7. Usernames! (ᵔᴥᵔ) says

    …sealioning is now against the rules…

    I can see how this is one of those “I know it when I see it” things that are very hard to quantify.

    What’s the threshold for how many times someone mentions you and the duration between mentions? Is it 5 times/day and < 1 minute between? Ugh.

    If I ask someone to stop mentioning me, and they do it again, is that sealioning? What if they wait a week and mention with an actual acceptable tweet?

    Fuck, just kill everyone.

  8. says

    meh, twitter will eventually be wiped out by a competitor who WILL put these rules and and ‘fixes’ in place fundamentally…

  9. s3m3rs says

    one – really, really likely outcome – is that Twitter may end up tiering (ugh – i am too tired to sort out spelling, it seems) the service – free – users can block-bot and use existing reporting mechanisms, including the new one. Users who demand a bit more support (getting sockpuppets suspended quickly, suspending sealions, that sort of thing) ending up paying a small monthly fee. Which is basically extortion – if Twitter would have addressed this years ago, we might be at the point of Twitter users being more responsible with their free speech. In order to make Twitter a less-hostile playing field, manpower and womanpower Will be required.

    I am feeling jaded though, hope I am wrong.

  10. Scott John Harrison says

    I have run up against something like this previously – being marked as “Spam” once for sending out too many @replies too quickly about 80-90% of my tweets are @replies to various people (Usually podcasters to talk about the episode I just listened to or a reply to a recent tweet someone makes.)

    I may be one of the few false positive this turns up but if it helps curb the harassment at all that is a good price to pay.

  11. governmentman says

    Is there any entertainment of the idea that this kind of harassment is an unavoidable phenomena of any large-scale, spammable social media service?

    “Actually enforcing” this kind of policy basically means spending a lot more money paying… someone… to read tweets and decide eventually semi-arbitrarily what crosses guidelines. What kind of people should twitter hire to do those reviews, what kind of training should they get, what kind of pay, and so on. I can imagine office-buildings full of people whose only job is to read reports of tweets and ban accounts which will inevitably just be re-created minutes later. Does this kind of resource-allocation really outweigh the cost of a few people having to deal with spam sometimes?

  12. zenlike says


    We are not talking about spam. We are talking about relentless harassment campaigns waged by a small group of obsessived individuals.

  13. jd142 says

    @13 AOL had something like community leaders. So there was someone(s) in charge of deleting offensive posts on a volunteer basis. I think. I am old enough to have used AOL, but haven’t for almost 2 decades. I am so freakin’ old. :(

    How that would work on twitter, I don’t know. Maybe you get three random moderators to look at your accusation and if 2/3 agree, then the person is banned for X amount of days.

    Slashdot has community moderators so you can read it at the moderation level you want.

    Another solution would be to set a “junk mail” rating on each tweet. That could be automated. And then you could just set your level and never see the tweets with abusive language. And just like a junk filter, you could whitelist some people. I don’t like that solution because it would no actually fix the underlying problem of the harassment. It would keep the victim from seeing the harassment though, so it is a step in the right direction. Maybe they already have something like this; I don’t twitter. But being able to say that I don’t want to see any tweets with the words c*nt or b*tch in them, or their leet speak variants, seems a reasonable thing to ask.

  14. georgelocke says

    I can see how this is one of those “I know it when I see it” things that are very hard to quantify.

    It shouldn’t be that hard. Compile a list of complaints, curate them, use them as your true set. Your sea-lion test involves some kind of rate of @ tweets since the last reply. Now you have a true positive set and, presumably it’s trivial to get a bunch of true negative tweets (non-sea-lion mentions), and you just apply the test and you get false positive and false negative rates. Then you just tune the parameters of your test until your false positive rate is very low.

  15. ethicsgradient says

    Prof Myers, you’re a published author and a public speaker, who often publicly criticises people you don’t personally know. I think ‘large numbers of unsolicited replies or mentions’ for you would have to be in the hundreds by each account before anyone would consider it as ‘spam’.

  16. ethicsgradient says

    The quoted “If you send large numbers of unsolicited @replies or mentions;” comes from Twitter’s definition of spam:

    Spam: You may not use the Twitter service for the purpose of spamming anyone. What constitutes “spamming” will evolve as we respond to new tricks and tactics by spammers. Some of the factors that we take into account when determining what conduct is considered to be spamming are:

    If you send large numbers of unsolicited @replies or mentions;