Twitter Censorship - Round 3
Dear diary, today I was let out of my Twitter cage for a while. It was nice, for a brief moment, being able to express myself, and not be instantly silenced.
Of course, it didn’t last, and eventually my first-reply comment with the most replies and ‘likes’ to this popular tweet was relegated to the bottom of the chain.
Following on from yesterday’s experiments, I decided to run a few more. And it was rather illuminating, to say the least. First off, I obviously cannot prove any of this, because I have no access to source code, nor do know anyone in charge thereof. But I did get a number of screenshots which make it pretty obvious that what is taking place is straight up censorship.
So what I’m left with is running experiments to see the results - which is what I did. Again, I wish to repeat that this is not for the sake of circumventing anything, it’s to test how the system works.
The two prior screenshots detail clearly how my reputationsal score was affected later in the day. As I progressively posted wrongthink, my reputation was negatively affected, and I was sent below two ‘more tweets’ buttons once again - one of which falsely claim the tweet is ‘offensive’ in order to protect you.
But here’s my main experiment. First I posted a link to an article on the ‘Online Safety Act’ on WHO’s page, with a full description. No good. Behind two clicks.
I then posted the same, but intentionally misspelled a few key words. Unbelievably - that led to relative success. In other words, it’s not the website, nor article posted which is the issue. It’s the words typed. Keyword search.
I then tried narrowing down the keywords causing censorship. Replacing simply ‘Broadband Commission’ was a no go, but replacing all three - including ‘UN’ and ‘Online Safety’ was a go.
Unfortunately, at this stage I got distracted and accidentally replied on the WHO’s page. From then on, further results were unreliable - though I still was good in catturd’s replies. This, however, was ‘corrected’ later on, as can be seen below.
So from this - and my prior experiments - here are my thoughts.
-
Here’s what I reckon are vectors in regards to censorship.
User (reputation) - this can be seen above later in the day, affecting replies negatively across the board
Interest in user-to-user interaction (is communication between two users ‘problematic’?) - after my wrongthink, i was back to being served tweets by users i rarely communicate with
Is the content ‘problematic’? - so far, no direct evidence here
Does the tweet contain certain ‘keywords’ or a combination thereof? - for sure. from the above this one is fairly clear
Does the tweet receive too much attention? - from prior post, rapid accumulation of impressions/likes/retweets, and then sudden death
Does the tweet refer to ‘problematic’ material? - again, at this stage, no direct evidence
Does it drag in other users guilty of ‘wrongthink’? - insufficient evidence
And here are methods of discouraging ‘problematic’ debate.
User not allowed to see tweets from like-minded individuals (they might get ideas) - this i observe repeatedly, when in the slammer for wrongthink
User shown tweets with those follows with least in common, and/or where there is no mutual interest in the current topic (serve uninteresting tweet) - again, this is par for the course
User shown follows-of-follows with least in common, and/or where there is no mutual interest in the current topic (serve uninteresting tweet) - same goes here. i get this all the time
If user must be shown, but latest tweet is unacceptable, then locate prior post with least possibility of ‘harming’ anyone (serve irrelevant tweet) - plenty of evidence thereof; utterly irrelevant tweet suddenly gains traction when ‘problematic’ material posted
Delay/cancel the notificatons relating to ‘likes/retweets’ if the user is online. This reduces likelihood of debate on ‘problematic’ issues - again, i observe this all the time
To conceal above, if the user has been offline for a given period (10 minutes?), then said likes/retweets are allowed to accumulate, and will be presented when user returns. This can be considered a reward for staying away, but not cancelling account - same goes here.
Place behind either one, or two ‘more tweets’ links, and incorrectly label the second ‘offensive’, to throttle visibility, as most users rarely venture below the 2nd or 3rd reply - clearly documented above
And finally, active punishments
If user remains a nuisance, remove random follower from list, potentially as a repeated, automatic process - i have seen and heard from a fair few that they experience exactly this
Temporary shadowban - posted while nothing of interest served in feed. will see if it improves tomorrow.
I might have missed some. That’s for round four, if so, assuming I’m allowed back without a filter again.
Oh, I do wish to add here, that a while back I compiled a thread on perennial liar, Marianna Spring. Here’s a limited sample of the tweets. See if you can spot the odd one out.
So hard to work out the reason.
