How a Linux Command Got My Twitter Account Locked for Abuse and Violation đ¤Żđ
(6 Min Read) A Linux Joke that Twitter AI did not find funny
(May 24, 2022) As I made my morning tea while browsing Twitter, I came across this tweet.
I found this tweet funny, so I replied with the following. Below is a screenshot of the tweet because I had to delete the original tweet to fix the violation and prevent my Twitter account from getting banned permanently.
Let me save you some time and effort before you head to the comment section and tell me my command is wrong, and it should be a âs/all/<pid-id>.â I was not trying to be technically correct. I was trying to be funny. In retrospect, âSudo kill -9 1â would have been a funnier tweet đ
Unfortunately, Twitter doesnât like my Linux joke đ¤Śââď¸
Here is the communication from Twitter as proof
The timeline is:
May 24, 2022 - 6:42 AM PST: Original Tweet posted time
May 24, 2022 - 8:52 AM PST: My unfunny, violation-triggering tweet.
May 24, 2022 - 8:55 AM PST: Twitter bots lock my account
May 24, 2022 - 9:31 AM PST: I submit an appeal to understand WTF happened
May 24, 2022 - 10:11 AM PST: Twitter says FU, you violated, I donât give a sh** and not going to tell you what caused the violation, but your account is locked. Delete the tweet, and we will unlock it after 12 hours of timeout.
May 24, 2022 - 10:30 AM PST: I took screenshots and decided to vent via this blog while waiting for 12 hours timeout to expire.
Below is my experience dealing with every step of the violation. How awful it is, and my recommendation for Twitter on how to fix it.
Additionally, this will be useful when (not if ) Twitter locks your account. And I will also share my views on Elon Muskâs take on open-sourcing twitter algorithm(s).
(Twitter) Your account has been locked đ
Before I mention things Twitter can improve, let me start with the good things about this communication phase.
The Good:
The Twitter violation algorithm took 3 minutes to detect, flag, lock my account, and send me an email. There is no way a human can read tweets so fast, so itâs a violation detection bot army doing at work. Doing this at scale means all tweets are following some fast regex, and likely, an AI is getting trained and making decisions. I plan to deep dive into this topic and share it in a follow-up post.
With 3 minutes from detection to email, I guess detection took less than 1 minute and flagging and other things took the remaining time. That is blazingly fast. Kudos to the Twitter team on that. I guess what Twitter violation detection algorithm SLAs are.
Unified communication across all devices. The violation notification email and pop-up on the Web and Twitter app for iPhone had consistent and clear messaging. Here is the screenshot of the email.
And this brings me to the bad part.
The Bad
The email doesnât tell me which tweet caused the violation. I am not a power tweeter. It was easy for me to figure out which tweet triggered the violation. But this is a serious communication gap for people who tweet a lot. How does one determine what caused the violation and prevent a future violation without any information?
I was hoping that clicking the âGo To Twitterâ would provide more insight. So I clicked it, and that brought me to the appeal process.
(Twitter) The Appeal Process đââď¸
Unfortunately, there isnât much good here to talk about. The white box (in the gif below) is supposed to contain the screenshot of the tweet but is empty. The page tells you the number of tweets that you must delete, but it doesnât say which tweet đĄ
Going back to my above point, the power Twitter users stand no chance of having a successful appeal if they donât know what caused the violation.
As you can see, Clicking âappealâ doesnât do anything. I must have clicked it 50 times to check if the request got submitted.
The other problem I have is accepting violations blindly.
On an important side note, the red âDeleteâ button doesnât delete the tweet. Itâs just an acknowledgment of violation. You still have to delete the tweet manually, or it can get re-flagged.
I had to find the Twitter support page to file the appeal that worked successfully. Once I did that, I saw this.
As you can see, Twitter and I are engaged in an appeal process of an invisible tweet.
Moments later, at 9:31 AM, I got a confirmation email from Twitter stating that my appeal had been successfully submitted. Still, if I would instead blindly acknowledge my violation, I could get full functionality sooner. I decided to wait.
Honestly, I was ready to wait for a week to hear back the decision on the appeal process. But I didnât have to.
At 10:11 AM, I got another email from Twitter with the resolution of my appeal. Damn, Twitter is fast (Yay!!) and vague (Boo!!).
At this point, I have only one option, i.e., acknowledge that I violated the terms and accept it. I canât re-appeal, and there is no way to figure out what caused it. Also, I donât know how many strikes I get. Do I get 1, 2, or 3 violations before my account is locked forever?
Before diving into recommendations for the Twitter team, let me quickly close this up by saying that I did acknowledge the violation and deleted the tweet. While I can access Twitter, I canât engage with it.
So I am currently on a 12-hour timeout which is stupid. If I were to cause a violation intentionally, I wouldnât go through the process of appealing and acknowledging it and then doing it again. This ban seems unnecessary, but hey, what do I know.
Open Questions for the Twitter Team
If anyone at Twitter sees this, please pass these questions to someone within Twitter so they can fix the user experience.
Please highlight which Tweet caused the violation
Please fix the appeal process on mobile devices.
Please explain why we canât re-appeal.
Please share how many violations before an account gets the permanent ban.
Please rename the âDeleteâ button or make it delete the violated tweet.
Please open-source the algorithms to provide more transparency into the process.
And the last point brings us to, Elon Muskâs views on open-sourcing Twitter algorithms.
Open-Sourcing Twitter Algorithm
As we can see, 924,234 (82.7%) people agree we should. As an engineer, Iâm afraid I have to disagree.
Based on the above experience, this may seem contrary. I do want to understand the violation and the appeal process. But I donât want to read Twitterâs code to understand it (I have my own code to read).
Moreover, as anyone who has built similar distributed systems, itâs not like all code sits in one repo. Itâs multiple repos distributed and connected with APIs and ducktape. And even if Twitter were to open-source all the code, 99.9% of people wonât read it. And the 0.1% who will, will summarize it and create a super-abstracted Twitter thread about it.
IMO, the upside of open-sourcing code may be moderate, but the downsides are huge.
Every piece of code has vulnerabilities and bugs. Open-sourcing code will make Twitter vulnerable and easy to be exploited by hackers. So I am against open-sourcing the entire code-base.
But I favor providing more transparency on how these algorithms work. And when I say transparency, I donât mean a document in a lawyerâs language. I am referring to a technical write-up.
In closing
Today was an exciting experience. My novice joke almost resulted in a permanent ban. Without the retro, I would have not realized that my Linux skills were dangerous. TBH, anything with âsudoâ as a prefix is dangerous.
I did learn a lot about the complexity of the Twitter appeal and violation process. I have sympathy for the Twitter team. Keep up the good work. You have a long way to go to make Twitter the platform we all want to hang around.
For me, Iâll be back on Twitter after my 12-hour ban is lifted (9:00 PM PST, May 24, 2022). You can follow me on Twitter by clicking here.
Please share and sign up for this free newsletter if you like this post. I post other useful content on engineering, people management, and personal productivity.
Thanks for reading. Until next time, âsudo rm -rf /*â đ byeeeeee
A free subscription gets you:
đ đ° Every new issue of this newsletter is delivered right to the inbox.
đ đ˛ Free access to previous posts is soon to be moved behind a paywall.
đ đ Top posts that made it to the front page of hacker news with comment thread.
You dont want to open source rhe code because it makes Twitter vulnerable? How lame. Security through obscurity is a futile endeavor, and open sourcing the algo doesn't mean we need the whole codebase. You're being a corporate suck-up. And it's duct tape; not ducktape, you lame duck.