Thread by @marcowenjones

[Thread] 1/ Good evening, afternoon, or morning all! Tonight's thread is on #Turkey, and it will be a big one. Many have commented on the massive hashtag "Help Turkey" that rapidly reached 2.5 million tweets today. Read on for an in depth Twitter analysis > #Disinformation

2/ 1st, some brief context. The hashtag "help turkey" involved people calling for international help to combat Turkey's wildfires. Images like the one below were common. The tweet storm prompted reactionary nationalistic hashtags including "Strong Turkey" & "We Dont Need Help"

3/ Some felt the message being generated on the hashtag was designed to make Turkey look weak, incompetent and desperate. This, coupled with the scale of the campaign, suggested a possible influence operation. To be clear though. The hashtag had many real users. See below.

4/ In addition to this, many real people using the hashtag probably do so in good faith, simply because it is quite understandable that one would expect to ask for help during a crisis. Also, the popularity of the hashtag naturally meant others used it to discuss the hashtag

5/ Caveats aside for now. My initial analysis included several stages. Firstly, a network analysis of around 160,000 interactions from around 46000 unique Twitter accounts. The graph below shows how massive the sample was. I will now highlight some likely #manipulation

6/ Pay attention to the circled area. This community in green is interesting for a number of reasons. I resized the nodes (individual accounts) by how much they tweeted on the hashtag. I.e. the more they tweeted, the bigger the node. This is a useful measure because it it shows

7/ those entities that are really passionate about promoting the hashtag. In usual circumstances this could be passionate parties who feel strongly about an issue, or other parties attempting to influence a conversation for whatever purposes. Ordinarily, you'd expect these large

8/ nodes to be more evenly distributed across the network, but this green community is a community with a significant number of nodes using the hashtag a lot. A really interesting finding is that the most active node in the sample, the busiest, if you will, is/was @ege20281770

9/ Now what you'll notice about @ege20281770 is that it has no tweets, even though it was created today (02/08/2021) and tweeted a lot on the 'help turkey' hashtag. Twitter says there is still one tweet, but there are none, meaning the tweets were deleted

10) Now we know from past EPFL research that influence operations in Turkey often use this tactic of deleting tweets after writing on a hashtag. Here the Twitter algorithm reportedly registers the trend, but the account deletes tweets to avoid detection/be repurposed.

11/ With this in mind, if we look deeper into the green community that had a lot of highly active accounts in, we find more unusual things. One of these tactics is handle switching, where users change their Twitter handle. Let's look at Badboy2147. He has a big 15k following...

12/ Watch this video. Badboy2147 actually no longer exists. If you look at the video below, you'll see that his old username is cached. When you click on the link to his account it says 'this account doesn't exist'. Actually it does exist, but it has changed to Badboy353435 🤯

13/ What's more, the old badboy seems to have deleted his tweets about help turkey. This can be seen by trying to look at replies to his account. Also, if you check new badboy's timeline, there are no more 'help turkey' tweets, which presumably have been deleted. #deception

14/ Just to give you another example, here you can see how an account toprakofficial5 changed to jokerqueenn_ . Now within the suspicious green community of around 7400 accounts at least 70 have changed their name or been deleted, and it's still very early on. I'd expect

15/ more to change or be deleted. We've seen this tactic of handle switching elsewhere in the world, including the Gulf. See the linked report for more details. But in the meantime let's move to the next interesting aspect of this. cyber.fsi.stanford.edu/io/news/februa…

16/ A massive aspect of this was the copy pasta. Thousands of accounts tweeting the identical tweet (I won't write it here because it will add to the trend, but you can see it in the video) below.... #Turkey

17/ Now I am not certain of the origins of this tweet, but it was tweeted by what I believe are a lot of Turkish celebrities (see image). As I said before, a lot of the accounts tweeting this message are likely real ppl, inspired by celebrities and others, although what's

18/ not clear is who is telling people to copy and paste the English content as opposed to just retweeting it? Is it possible that people just know to do that? It is also odd that I tracked at least 105,000 instances of these English tweets, including retweets. I managed to

19/ scrape around 35,000 unique instances of this copy and pasted English tweet. In addition to this sheer volume of copy pasta, the peak activity occurred past midnight Turkish time. As you can see from the graph, most activity occurred at around 00.16 - which seems pretty late

20/ Another bizarre aspect of the trend was the copy and pasted multilingual calls translating as "turkey is on fire and needs help". Look at the video below. I managed to scrape thousands of these instances. Again many are replies to other accounts to generate engagement.

21/ Also like so many weird, seemingly manipulated trends, loads of BTS and K Pop fans retweeting this message. I counted around 200 who actually retweeted it. I can't imagine a lot of these are genuine, despite the popularity of Kpop...

22/ Also let's not forget precious nijiko!

23/ Anyway that's enough for today, probably time to go to bed. In sum, there is clearly a lot of manipulation going on 'help turkey'. In addition to being boosted by real people and celebrities, it is being artificially manipulated by what is likely to be thousands of sock

24/ puppet accounts. Exact numbers are hard to determine at the moment but I will try to over the next few days. I will also try and see what's going on on Strong Turkiye as a counter hashtag. Thanks for tuning in everyone. Good night and peace #deception