Twitter Is Blocking My Tweets (Updated: Back to Normal)

I had tried tweeting on the Iowa Democrats refusing to announce caucus results and instead calling for an investigation. I tried tweeting this first as text, then as a screenshot. Neither went through:

I then tried Tweeting this:

I do not have any notices from Twitter either via e-mail or in my Twitter backstage about my account. Autotweets do appear to be going through but not new tweets I initiate from my account.

It is also blocking retweets. I tried retweeting a Reuters tweet as an experiment but that was rejected too.

Needless to say, this is appalling. particularly the lack of notice. And I see nothing in my recent tweets to justify this, save promoting a lot of Sanders tweets. Is the basis for this ban?

Update: so far, this Tweet has not gone through as an autotweet, so it appears my account has been blocked.

Update 7:00 PM EST. Tweets now seem to be going through, but they weren’t for at least an hour. Similarly, I’d try to retweet, Twitter would for an instant increase the # of retweets and reverse it, and the retweet would not show on my home page either.

I was using the same browser I always used, tried reloading the page, opening a new page, plus this post when it first ran did not autotweet either. So I have no explanation for this temporary outage, but readers indicate there were a lot of people having trouble without the problems being universal which is even more odd. But things appear to be back on track.

Print Friendly, PDF & Email


  1. anon in so cal

    I am unable to tweet, retweet, or reply as of an hour ago, 12:54 pm PST

    ETA: my last tweet mentioned Dems rigging Iowa

    PollyTickal’s account was deleted on Tuesday.

  2. skippy

    Can it be run around via other accounts posting your vaporized tweets, especially considering international viewership.

    Would seem problematic for the platform, not unlike the latest Iowa debacle.

    1. skippy

      Twitter Support


      You might be experiencing trouble sending new Tweets, but we’re working on fixing this now. Sorry for the interruption and we’ll let you know when things are back to normal.

      Lmmao at – “back to normal [tm]”

      This is some heavy duty “B” grade Sci Fi action …

      1. polecat

        back to kafkaesque .. ‘cray cray’ – is what they negect to state.

        … bububut Twitter be GREAT! .. Right ?? *

        *….. not like those ‘other’ sites … /$arc

  3. FredsGotSlacks

    I was having this problem for a few minutes. Tried to retweet a Jimmy Dore tweet, kept undoing. Tried retweeting a Gaius Publius tweet, same thing. Tried commenting on Jimmy Dore tweet, rejected. Tried retweeting a mainstream news source, no problem. Then, right after getting a comment on the Gaius tweet through, it all started working again. I left a comment on your tweet Yves re: this issue, and that one worked too. Just a random glitch or something nefarious going on behind the scenes?

  4. Situation Normal

    It seems the problem of not being able to tweet was widespread. It may have affected all users (it affected me).

    1. Yves Smith Post author

      No, I saw plenty of fresh tweets when I was trying to tweet. And I howled to some important allies who are more active on Twitter than I am and they weren’t having any problems.

  5. Calvin

    If censorship is the common cause of such removals, then these private Silicon Valley enterprises need to be regulated, just like the the railroads, phone company, Radio and TV were, in the period of America’s greatest growth, which seemed to be abetted by regulation.

    Can someone please contact the Sanders campaign?
    This needs to be an issue addressed by our next president.

    Or, on the other hand, SV could have a council, after which they announce that in the interest of Democracy and capitalism, all censorship is to be dropped with the exception of pornography and obscenities.

  6. richard

    I both don’t tweet much and just started, so I wasn’t sure what to think when it happened, but both my tweets about Iowa cheating, one to the DNC and one to Tom Perez just disappeared right after I made them. Wonderfully filthy they were too.

  7. Fiery Hunt

    “”Damage to our fiber-optic network in two separate locations is affecting Spectrum services for customers in ME and NH. Our engineers are working to restore service as quickly as possible and we apologize for any inconvenience to our customers.”

    Saw on Twitter just now.

  8. lordkoos

    Same thing happened to me today. I sent a message to twitter support, and in an hour it was magically fixed. So maybe it was a systemic problem? Hard to know what is really going on anymore. With me I was able to tweet, but not reply to anything.

    1. drumlin woodchuckles

      Or a legit Lockdown.

      If Twitter were to Lockdown EVERYbody, nobody could accuse Twitter of censoring somebody in particular.

      So my vote is for Lockdown. I betcha this is a Twitter Lockdown. Can people who find that offensive get revenge on Twitter?

      1. Yves Smith Post author

        This is what was distressing as a user. I could see a lot of fresh tweets when mine were not going through, so it was clear Twitter was not having a systemwide issue.

        1. drumlin woodchuckles

          Ahh . . . that makes it look targeted, then. If I were a Twitter user, I would like some evidence that this really was just a technical failure before I generously allowed Twitter to claim that.

  9. Arizona Slim

    True confession: I don’t do Twitter. But I am absolutely, totally, and gladly hooked on Naked Capitalism.

    You rock, Yves, Lambert, and Jerri-Lynn!

    1. Carey

      >as Jimmy Dore says, it is amazing how the mistakes and incidents like this all go one way – helping the establishment.

      ..and that’s why I call bullsh!t on all the recent invocations of Hanlon’s Razor.

  10. jefemt

    At least its only happening in ME and NH, on a debate night, and primary next week.

    No worries, nothing to see here, folks.

    Seriously, my recollection was that Yves is no longer up north-east, but is down south in Alabama?
    Do all tweets get routed through servers in ME/NH?

    Is Trump silent?

  11. ex-PFC Chuck

    Matt Taibbi just tweeted saying that and I retweeted it with a comment. I then went immediately to the top of my page and Matt’s tweet in my retweet was marked “Unavailable.” I doubt if this is accidental.

  12. Daniel Raphael

    Yes, this was just more system hiccoughs. A common thing, as Twitter constantly tries to tweak itself in the endless quest to squeeze more revenue from its sort-of functioning. Back to normal now.

      1. The Rev Kev

        They gotta get the Censorship feature right, including the Total Lockdown feature, before November remember.

  13. Calypso Facto

    Speaking from professional experience working within 2 of the FAANG (altho it’s really just GAMF now), at any given time on any of these massive platforms, something major is broken. It is usually papered over on the front end but don’t be fooled by all the shiny bs about high availability and multiple failovers blah blah. Because of their size it’s much harder to selectively silence a specific account than it is for some major critical piece of infrastructure to get hosed or thousands of devices to be digitally nuked thanks to fat-fingering a config file. This is literally the only argument I can make in defense of the tech platforms not deliberately engaging in journalistic oppression.

    1. Jeremy Grimm

      I do not understand why it would be so hard to “selectively silence a specific account” and what does size have to do with it? I would think someone could create a separate dataset of the accounts that are to be blocked and test each tweet to determine whether it is in the dataset.

      1. Jeremy Grimm

        At times of high traffic selective blocking could be used to throttle down the traffic and also deliberately cut off certain voices. Selective blocking triggered by high traffic would tend to appear as if it were due to the high traffic.

      2. Calypso Facto

        Behind the scenes, every platform is a bunch of different services. One service might be for the timeline/feed queues. Another service might be for private messages. Yet another might be for friend associations. Each service is by design segregated from the other – different code base, different infrastructure running the code. The service applications are distributed, running on clusters and not single instances. Big platforms have thousands of discrete pieces of infrastructure, but when the end user opens the website, it appears to be a smooth monolithic interface. Everything is decoupled so if one piece goes down, the entire thing doesn’t go down. This is what it means, generally speaking, to ‘scale’.

        Some of the few tech regulations with teeth are around Personal Identifying Information (PII). GDPR is one manifestation of it, but in the US the real reason PII is sensitive is because it is used for credit card processing and payment. The bigger of a player you are, the more PII data you have, and the more compliance you have to follow regarding it. Tech companies are unethical but these regulations are pretty strong so one of the very few real software design principles that is followed by serious players is keeping the PII data separate from the rest of the application data. So the PII infrastructure is often segregated from the rest of the application infrastructure and has additional security safeguards around it, and annual compliance testing is done to validate it.

        What this means for the people who manage the actual platform is that the data identifying specific user accounts is not searchable. It is usually blinded by making user accounts identified by a unique userid that can be hashed to some other internal id, but it can never link to the real PII. Because of the sheer size of the platforms, there’s no way to identify which userid is really the account you want to silence, because you can’t validate it against the PII. This is why all the actual censorship has to be done via the front end, with the ‘working groups’ scouring Twitter for trigger words to then report to the censorship AI, which is not human and can link the userid and the PII and censor the account. Is it possible the platforms have an internal tool to report to the censorship AI? Probably, but based on what I know about the working conditions and reality within them, something big broken behind the scenes is more likely.

        1. Calypso Facto

          Should add here that I’m discussing just Twitter’s censorship practices and not for example Google or Facebook’s. My feeling based on the engineers I know at Twitter and their internal practices as that they are the ‘most ethical’ among the Big Tech platforms (so Amazon/AWS, Google, Microsoft, and Facebook) and don’t allow their internal employees to flag accounts for blocks any way to do that except via the same censorship process everyone else uses (which I don’t know for a fact is solely AI but I do not believe relies as heavily on contracted human labor to verify as FB’s). Google has long been rumored to have actual divisions for government agency use and who knows what permissions they have over the system.

          The selective blocking scenario you described isn’t really feasible due to the way highly available distributed applications are designed at this scale. You would have no guarantee that you’re hitting only those accounts among the literal billions on the platform.

          1. Jeremy Grimm

            Sorry — I don’t understand.
            I can believe the implementations of the twitter platform are myriad.
            “different code base, different infrastructure running the code. The service applications are distributed, running on clusters and not single instances. Big platforms have thousands of discrete pieces of infrastructure, but when the end user opens the website, it appears to be a smooth monolithic interface” …
            OK … but Looking at a twitter page on my browser:
            Skinny Pudzianowski@UtahSocialDem Replying to @SteveKornacki
            In a window on the right side of the screen:
            Relevant people: Skinny Pudzianowski@UtahSocialDem
            USU alumnus. Social democrat, secular humanist, & atheist.
            Steve Kornacki@SteveKornacki
            National Political Correspondent for NBC News and MSNBC. My book, The Red and the Blue, is available now. Instagram: steve01450

            … that stuff isn’t secret. Those are addresses. The infrastructure for looking up addresses is not simple but I have trouble believing it is as truly mystical and hidden as you imply. You do not need to know somebody’s name or where they live to couple an address to a sender you want to block. Collect addresses you want blocked — like @SteveKornacki — or Skinny Pudzianowski@UtahSocialDem and make a list of them. Before accessing twitter’s equivalent of a DNS to route a tweet look up the address in your list. If it’s on the list and certain conditions are detected on the net — like heavy traffic — drop the tweet or stuff it in a hole.

            “You would have no guarantee that you’re hitting only those accounts among the literal billions on the platform.”

            If your intention is to block certain accounts in a way that is not obvious — why is it a problem if a some other accounts take a hit too?

            1. smoker

              thank you Jeremy!

              First AI™ and 24/7 surveillance by Quasi Governmental Tech Oligarchies is omniscient, and needs to take precedence over all life – but then, not a thing is known and easily targetable about all those persons whose highly personal profiles have been data banked for both monetization and punishment?

              1. Calypso Facto

                Comments I made are lost by skynet :( I got tired and forgot about this so I”ll paraphrase from something I said that was lost:

                The big 5 (big tech companies) are too big to manage. There are literal billions of user accounts, literal hundreds of thousands of discrete pieces of infrastructure, literally huge swathes of code pushed out every day that is undocumented and untested. It’s easy to assume omniscience, or ability to censor, but from my professional working experience behind the scenes, the problem is that they’re too big to select even a large set of specific accounts to target for silence, except via a moderation tool that is ‘targeted’ via the front end exposing the stuff we users see.

                As I said above, it’s completely possible Twitter has an internal moderation tool to hide specific accounts, but I doubt it, since their entire value offering is to be the single place people who care go to discuss – and if they’re seen ‘taking sides’, since it’s not a service the ad targets/users pay for, if people don’t voluntarily use the service, its captive audience to throw ads at is worthless. This is why ‘Jack ban nazis’ is a thing, and why accounts are bounced for moderation reasons.

                Performance problems under load are almost always lack of resources. There is not enough oversight over any part of these TBTM big tech companies that can perform ‘real’ censorship of the type you are thinking except Go ogle.

            2. Calypso Facto

              that stuff isn’t secret. Those are addresses. The infrastructure for looking up addresses is not simple but I have trouble believing it is as truly mystical and hidden as you imply.

              It’s above my pay grade to fully explain what is happening behind the scenes, but I really want to convey the reality here, so I’m going to attempt a very general and simplified explanation in hopes to really make it clear that the problem is TBTM and not overt censorship.

              As an engineer looking at the infra running the code, the data stores housing the data generated by the active applications, and the code base itself, let’s assume I have standard operator privileges over the environment. I can stop, start, and restart individual services; I can restart or reprovision underlying infrastructure running the service applications, and I can read and contribute to the code if I desire. Let’s say I want to silence noisy leftists with my operator skills. I have a list of public usernames (what you are referring to as ‘addresses’) and all my powers behind the scenes but the public usernames are not visible to me as the public username – it’s a unique userid with 32 characters, hashed to a chain of other unique userid’s that identify that public username in whatever service is calling it. Let’s assume I have an internal tool that will allow me to decrypt that public username to whatever the unique internal userid is for whatever individual service is calling the userid (remember my earlier example of timeline feeds, private messages, and post queue services). Now I, the internal operator with high privileges and malicious intent, can theoretically do some kind of technical badness given this information, IF there is an internal tool that ties all these different services together at a single public username.

              So that’s the application side. Where is the specific file, or code, or ‘thing’ I need to edit, or update, or change, or ‘block’, in order to silence the leftists? Well there isn’t one. There are redundant copies of each different service’s data (the queues, the DM, the posts), and there are redundant copies of the actual data associated with the username itself that we still can’t access directly because it’s PII. Lets say they use a data structure that is sharded – think of RAID striping but it odd numbers instead of evens. There is no file there to be edited – everything is updated by the application code. I, the malicious operator, can potentially find the specific pieces of infrastructure that are hosting the username and I could nuke them (along with the thousands of other accounts on that infra), but I’d have to do it one service at a time. That’s data loss, so then the disaster recovery processes failover and repopulate the nuked infrastructure as soon as it’s gone.

              This is why I’m saying that if there is censorship happening, it’s going through some kind of tool that is autonomous and works from the front end, like a moderation tool for a user to flag an account. Enough people flag for moderation, and everything is hidden by application code and prepared to be gracefully deleted.

        2. fajensen

          You don’t need to censor specific people based on established identities, only people who ‘looks like them’. This method works well for the ‘Signature Strike(tm)’ and also for SoMe.

          That way you can also truthfully say that you don’t censor based on some vague properties like “political viewpoints” and “ideology” because you don’t really know why your MI went and censored something.

          People are biologically designed to be very good at adapting their behaviour to gain from almost-but-not-quite random situations. They will work hard to re-program themselves to follow the ‘hidden rules’, to maximise their ‘potential outcomes’ as it were.

          Therefore, you want to ‘nudge’ people to not discuss specific things, you want ‘the rules’ to be vague and obscure, almost-but-not-quite random, so that the whole ‘user experience’ is just Better, More Rewarding, when nobody mentions what should not be discussed.

          When someone does anyway, ‘the platform’ will sometimes just glitch and lag, causing unexplained frustrations, which acts as the negative feedback. The metal energy spent on rumination about *why?* and *how?* helps with reinforcing the ‘learning experience’.

          Daniel Kahneman and Amos Twersky published a lot of very readable research papers into behavioural psychology and how people actually think and learn. Now “we” are beginning to see the application it, to “us”, via ‘Social Media’ and ‘Information Attacks’.

  14. smoker

    After reading the comments to this point, I vote for 100% Nefarious, which I suspected all along.

    Maybe Nancy invited her fellow resident, billionaire Dorsey, over to her Estate for a chat in District 12 this morning, or something similar.

  15. Edward

    The cyberwarfare methods tested in the Middle East are coming home. This may be part of that. Recently, on Democracy Now, there was an episode discussing a documentary about Cambridge Analytica. It turns out that private companies such as this one are using cyberwarfare methods that were developed in the Middle East. Very dystopian.

    1. polecat

      Call me a CTist all you want, but you can just bet that the Israelis have a big hand in this ‘kind’ internet squirrellyness.

      1. Edward

        The Israelis don’t get talked about enough. One of the absurdities of Russiagate is that there is plenty of foreign interference– from Israel. One example is the 1992 Democratic primary. Initially, Bill Clinton was losing to Jerry Brown. AIPAC saved Clinton by raising a large sum of money for him, which was enough to put him in the lead. So Israel actually determined the U.S. president that year. This interference was revealed in a conversation a Jewish New Yorker secretly recorded with the president of AIPAC. He sent a tape of the conversation to the American Arab Anti-discrimination Committee which published it.

  16. The Rev Kev

    Just remember what James Bond once said-

    Once with Facebook is happenstance. Twice with Snapchat is coincidence. The third time with Twitter it’s enemy action.

  17. RBHoughton

    Looks like censorship to me, although Heaven know why. Keep the noise up NC staff and readers. Does Twitter have no competitors to keep it in line?

  18. CoyoteMoon

    Same for me and many other longtime progressive Twitter regulars.

    Happened several times in same day, especially regarding #PeteBillionaires.

  19. skippy

    @YS remember in the early NC days where some libertarian sorts where chest thumping about using digital media and mega data to control the unwashed – absolutely and completely unabashedly assured in their victory.

  20. Andrew Thomas

    Sounds like a pre-planned use of a selective lockdown in the midst of a technical glitch that did affect a fairly large random group of users. They get to practice targeted lockdowns with the cover of an actual glitch. Then, they get to see if we gaslight the actual targets of the lockdown, and also see if it affects their credibility within their community. Yves was targeted. This was a dress rehearsal. Will there be more before the play opens? Who knows?

    1. CoyoteMoon

      The “glitch” certainly did look and feel like this. Many progressives were online later in night saying they experienced this intermittent outage. DK about other than left leaners.

  21. Jack Parsons

    I had this kind of glitching also and it was because Twitter wanted me to use the latest Chrome. I had to update and then everything was fine.

    To stupidity v.s. malice we must add computer security.

Comments are closed.