Twitter Invisibility – A Customer Service Tale

I will start with the dispassionate assessment: for a period of no less than 8 days, my @funranium twitter account was functioning in a very limited visibility status and then it returned to full functionality. I have no idea why it happened or why it stopped.

To roll it back to the larger picture, this has been another one of my adventures in customer service failure. I gave my feelings on how customer service is supposed to work, and how it is usually structured badly, in a previous post seven years ago. Now you may say “Phil, we aren’t Twitter’s customers. We’re the product.” and you’d have a good point. The argument that all of social media’s actual customers are ad purchasers, not the users of the services, ignores that we are the audience they are attempting to sell to. If your service causes enough pain that we aren’t there to sell to, your ad purchasers will wander off to more lucrative outlets. In summation, it’s tricky.

This all first rose to my awareness on Monday, March 5th when my Lovely Assistant complained that my threading seemed to be broken as she went to go look at my stupid Alternative Chowder tweets. I dismissed it as [shrug] “sometimes things on Twitter don’t work so good”.

(ASIDE: I cannot consume any animals that come from the water, fresh or salt, other than whales, so clam chowder has always been dead to me. As a small child, I noticed that whenever my dad ordered clam chowder he got oyster crackers, whereas I got normal saltines for my soups. I became convinced that you could *only* get oyster crackers if you ordered chowder, which made stealing them from my dad all the better. I then developed the theory that oyster crackers were what made my dad’s clam chowder, which sure looked like a soup, a “chowder”. BY SIX YEAR OLD PHIL LOGIC, not yet understanding the transitive property, I came to the conclusion that adding oyster crackers to any other soup automatically makes it a chowder. Now, as a grown ass adult, no one can prevent me from adding oyster crackers to whatever I want to exploreAlternative Chowders“)

On Wednesday, two other friends noticed they weren’t getting notifications from me when I made comments. Again, I dismissed it as Twitter being Twitter.

Well, that’s not good. I seem not to exist.

On Friday the 9th, the journalist Nathan Edwards dropped me a line to let me know that I appeared to be shadowbanned. He had used a tool which did a quick search scrape of Twitter to see if a user who wasn’t logged in would be able to find anything from my account. It came up with zilch. Something was limiting my contact and this presented two possible conclusions:

  1. My name had been handed to a Report Brigade for trolls to hammer such that the automated systems at Twitter would defensively limit my reach as someone that was suspect. OR…
  2. Something at Twitter was broken which had somehow impacted my account.

To be clear, Twitter has emphatically stated that it does not do shadowbanning. I am inclined to believe that, but I am also aware that they don’t *quite* have the firmest grasp on the monster they’ve created.  While they might not have built shadowbanning as a feature, I am certainly willing to believe that effects indistinguishable from it have manifested as a bug.

Searching through my collective friends, I found one other person that seemed to be suffering the same problem, Meredith Yayanos. To be honest, I could fully believe she had pissed off enough people to justify Case #1, which is something she would wear with pride. As I help support her work with the the Blood of the Harpy, I was willing to believe I had inherited some guilt by association and been targeted in kind. But almost as soon as I let Mer know that she was subject to the same thing as me, she was returned to full visibility.

Okay, well, that’s weird…

Twitter Support Says “We Had A Problem But It’s Cool We Fixed It”

As a middle aged white man who works in safety with a firm confidence in Systems™, this meant it was time for me to go delve into the customer service of process of Twitter to get the help which I felt I richly deserved and expected to work.

(You may feel free to start snickering into your drink at this point.)

A conversation between Jamais Cascio and I, except I am invisible with “Tweet unavailable”, as captured by @slartibartfart

My experience of the Twitter customer support process may be described as a running into a brick wall as fast as you feel like. It’s always a wall but the question is how much it hurts. The first thing I did was hit the Twitter support functions. I looked at their feed first and found the message to the far right. It appeared that they had a problem but, that as of the 8th, they’d fixed it and everything was fine. [checks calendar] By Friday everything should have been fixed according to them, but I still seemed to be stuck in a corner where no one got notifications from me and all threads I commented in had broken links, as shown in the conversation with Jamais Cascio.

Phil waves jumps up and down trying to get the elephant to see the invisible mouse.

As it was not fixed, I sent a reply to the @TwitterSupport account to let them know that things were not fixed. Unfortunately, due to the nature of the problem, almost by definition, they wouldn’t see anything I posted to them. Since that wouldn’t work, I checked the Twitter Help functions. They recommended that I use the @TwitterSupport direct messages. Okay, sure thing. Will do.

I try to use @TwitterSupport Direct Messages, LOL

Ah, that was a good laugh. At some point in the recent past, @TwitterSupport decided to turn off their direct messages and set an autoreply but they didn’t decide to change their recommendation to pursue this route on the Twitter Help page. Speaking of Twitter Help, I would describe this page as roughly as functional as looping automated answer phone tree. If your question doesn’t quite conform to any of the pre-formulated answers to frequently asked question paths, well, you aren’t going to find any help here, nor an easy way to ask an open question, nor even really a way to make contact other than the @TwitterSupport twitter account. As of the writing of this post, there has been no response from any channel of communication.

The earliest suggestion on how to fix this, or at least get the attention of people who can fix things, came from friends who went to search what shadowbanning was and found this article. To summarize it’s idea, if you talk to the Twitter Ads team, rather than the Support team, as their purpose is to generate revenue they have an interest in getting your problem fixed so that you can pay them. For the price of $20, I was willing to play this experiment to see how the algorithm coped with two conflicting desires: limiting visibility vs. getting paid for visibility. I regret to inform you that the algorithm is capable of holding these two contrary thoughts. Also, Twitter Ads Support is very clear they they are stovepiped and do not  fix, nor will they even discuss, any issues not related to advertising even if this would impact the effectiveness of their work.

Because the Army of the Caffeinated contains several current and former employees of The Good Blue Website, they noticed that I had vanished from their feeds and were a little miffed about this. When they asked me some questions as to what was going on and then looked into this there was a fairly universal “Oh. That’s not good.” In the last year, Twitter has had a lot of turnover without the best/any knowledge transfer. I am fully willing to believe that recent tweaks to the algorithm broke shit they didn’t expect and they’ve been working hard to band-aid things ever since.

And then, just as mysteriously, on March 13th I returned to full visibility. There was no explanation as to why I had returned, much less why it happened in the first place. It just happened, like a capricious god had granted me the gift of being a real boy again. That doesn’t sit particularly well with me. As a safety person, I demand documentation from my researchers for their potential failure modes so that we expect them and know what to do when they occur. When things like this happen, I want to know what went wrong, how they fixed it, and how they intend to prevent it from happening again, not just that they are “conducting an internal review to ensure that this doesn’t happen again”.

Since this seems to be continuing with other users, it isn’t fixed yet and the process, whatever it is, is ongoing. We don’t know what it is and all avenues for direct communication are decidedly blocked off. That is shitty customer service.