Could we get a censorship tag?
To see entire discussion thread (maybe, unless replies are also getting censored :-P), start at top of thread: https://twitter.com/taoeffect/status/655869593588666368
Academic take on Twitter’s censorship of Turkish citizens (but same stuff is being used in USA):
Collecting over 20 million Turkish tweets from late
2014 to early 2015, we discovered over a quarter million censored
tweets—two orders of magnitude larger than what Twitter itself
We applied standard machine learning/clustering techniques,
and found the vast bulk of censored tweets contained political
content, often critical of the Turkish government. Our work
establishes that Twitter radically under-reports censored tweets in
Turkey, raising the possibility that similar trends hold for censored
tweets from other countries as well. We also discuss the relative
ease of working around Twitter’s censorship mechanisms, although
we can not easily measure how many users take such steps.
“Never attribute to malice that which can be adequately explained by a relaxed consistency model.” – https://twitter.com/cdaylward/status/655959816335069184
The word “malice”, I think, is inappropriate here. Few people censor with malice, and I doubt Twitter censors with malice.
We do know:
So, there’s plenty of reasons to think that some non-malicious (“for the sake of national security”-type fallacious reasoning) censorship is going on.
I agree that we shouldn’t assume we know everything about the situation, but it’s enough to raise serious questions.
Appelbaum is certainly a plausible target, though we need to be aware of the sample bias because of course people are going to notice issues with his tweets that they wouldn’t with somebody who isn’t politically at odds with the US.
I do have to say that I find it unlikely Twitter’s consistency model could delay a tweet by four hours (which is how long it had been when I checked), given that their entire platform is about global communication, and that it manages to get millions of tweets across the Atlantic within a couple minutes at the outside. And given that the rest of Appelbaum’s timeline had propagated fine.
I worked at Twitter, and, knowing what I know about how timelines are stored, it’s totally plausible. The consistency guarantees between the Redis replicas is not strong. Could it be something else? Sure, but the plausibility is in favor of the replica missing a tweet.
Please don’t violate any NDAs or anything, I don’t think the situation warrants that risk - but it’s useful to hear that.
The consistency guarantees between the Redis replicas is not strong. Could it be something else? Sure, but the plausibility is in favor of the replica missing a tweet.
Here’s what I know:
These events fit the model of “list-based censorship” (as opposed to geo-based censorship), where a tweet is visible when it’s initially posted, someone notices and for whatever reason it gets shadow-banned (perhaps a bunch of bots click “Report”). People who are on a list of some sort are not able to see the tweet. The additional censoring of seemingly irrelevant tweets along with important tweets makes the case for “bug” look stronger than it is.
How would those facts fit the Redis replica notion?
I don’t get it. I’m American and I can see the “banned” tweet. Am I on a special NSA whitelist?
The caption in the screenshot is outdated (see replies to original tweet). It’s not clear what group specifically is being censored or how the censorship works, but myself and at least one other person (EDIT: and another) are in the group that can’t see certain tweets from Jacob’s timeline (as well as other tweets from other accounts  ).
Yes, for what it’s worth, I looked a few hours ago and could see the tweet individually but not in his timeline; I’m using a US IP. So you can at least increment the count of people who can’t see it to three. It’s kind of a pain to find “where it would be”, because RTs bear the timestamps of the original but are sorted from when they were RTed, so only the timestamps on Appelbaum’s own tweets are useful for this.
This PDF, linked from the tweet thread, summarizes what’s been determined and how, and theorizes that this is a system Twitter has in place for “abusive tweets”, quoting Twitter’s explanation of its behavior.
Being able to see the tweet, but not from the timeline itself, is consistent with that theory. It’s also expected that it should be visible as if nothing ever happened, about 24 hours after it was made.