Saturday Night Data Party: Social Media Detective Work

Hey everyone, it is Saturday night. This is Saturday, April 4th. This is the Saturday Night Data Party. And tonight we have ourselves a detective case to solve. So let's dig into the case that we're looking at.

This is a message from a friend. It says, Hello from the blue. Hope you're doing well. Judging for your posts you are. I need a favor and analysis.

I smell a Cambridge Analytics style rat that's played out in the last few weeks in African social media. The story started out with Bill Gates trending in South Africa earlier this week, and it was dramatically negative stuff about birth control, Africans as guinea pigs for coronavirus vaccine, the alleged vaccine being used for population control, etc. Very strong sentiments out of Johannesburg about Bill Gates. I couldn't piece it together or find a source, so I parked it. Then today my husband shared this in a group chat on WhatsApp.

I think this interview, and there's a YouTube video, uh, created a wildfire anti-vaccination narrative, and I'm interested to know if your fancy tools can figure out if someone's behind it. Very random request. Uh so let's see what we can find. We're gonna start with the crowd tangle uh software, which is Facebook uh bought this company a number of years ago. And what Crowd Tangle does is it allows you to search essentially all of Facebook uh and Instagram.

They used to have Twitter support, but they don't anymore. Uh let's see, so let's start with the BBC news story. Okay, 273 actions. Not a whole lot here. Right?

It's not uh not amazing. So there's not much to start with here. Okay. Let's see. While it's sensitive, the doctor's sentiments are not necessarily racist.

That's a secondary debate. What bothered me was the BBC fishing for clicks with hyperbolic headlines. I'm very concerned about Africans becoming skeptical about treatments during a global pandemic. Here's a video clip that I could find with the actual interview. Alright.

So let's take this video clip that my friend sent along. Okay. Take that URL. Let's take this and stick it into crowd tangle. See what we come up with.

Ah, wow. Okay. Well, there's definitely a something going on here. Uh let's see what we got. We have Tech Talk with Simon.

Two doctors are normalizing racism. Okay. They've got the link. Two French doctors on live two suggestions care experiments for the COVID-19 vaccine in Africa. So we've got some stuff going there.

We've got a lot of activity on Twitter. French doctors about Africa. COVID-19. COVID treatments. Okay.

So what we want to do is we want to try and isolate and figure out where did this chain of evidence start with? So let's go over to TalkWalker. Now what we want to do is we want to identify. Based on what crowd tangle is coming with. Crowd Tangle is going to give us enough of the basics here, but we it you can see it's not really time chained, right?

There's not a good way of seeing how these different things are connected together. So what we need to do is create a codex of words and phrases. So let's do do let's start with COVID 19, COVID or Corona or Coronavirus and vaccine, racism, French doctor, French doctors. Let's see what else we got here. That would be COVID 19.

Two French doctors. Okay, let's get guinea pigs in there. Okay. That's a good starting point. Let's take this query.

Feed it into quick search. Now do we have ah so this is thirteen months. You can definitely see this is just in the last thirty days, so let's restrict down this time frame. Just in the last seven days. Twenty f yeah, there we go.

Twenty-five thousand mentions. All right, let's take a look at the results. Alright, we have Al Jazeera. What's this is the biggest performing story. Let's do our trace back.

What do we got here? Twitter result there. That's somebody who's got the interview. That's where it started. That didn't really take off all that much.

This appears to be the start note. So the Al Jazeera story is where this whole cluster really took off, and then you can see we go into forums, we start getting substantial discussions, coronavirus update, and then you have Twitter, so on and so forth. So at the very least, this information epidemic started with this Al Jazeero story. What fed that appears to be this magazine result? Yep.

Which cites this tweet. So this tweet really is the first one there. The question is, is this account a valid account? I do not speak French particularly well. But the very joined March 2010 looks legitimate enough that that would be a good starter point.

All right, let's take a look now at the next result in the set. Okay, we have tweet here. Didia drogba, which is a football player. That was April 2nd. Fed into online news here.

April 3rd. And then we have ourselves a new article here. Let's feed this one into Crowd Tangle. 47,000 Facebook interactions. African football.

Then you got some Facebook groups. Let's go back to this Al Jazeera article. Hello from Canada. Hi Kat, how you doing? Tonight we're doing detective work trying to try trace down potentially a source of misinformation.

Um the Al Jazeera article, let's open that one up. Feed that in a crowd tangle. What's the timing on this one? The timing on this one was April 2nd as well. Al Jazeera English.

Tens of thousands of reactions, many, many interactions. All April 4th. Let's sort by oldest. So you had that one tweet that got picked up by this magazine, which then fed Al Jazeera. If that's a plant, it doesn't look like a it's a very subtle one if it is a plant.

All right, let's go back into There's Al Jazeera's Facebook post. Africa Fact Zone. Huh. Well that's interesting looking. What do we got here?

So Africa Fact Zone kicks off a bunch of stuff. That is a lot of spread. Let's take a look at this account. This is not exact this is not a human account. I don't know if this is funded by anyone.

It goes to an Instagram account. So if we're looking for something that creates a lot of spread but does not start from a credible source point, this one looks kind of fishy. You know, not a lot of account uh the account is old. Um it's got a lot of followers, but it appears to be a mostly automated account. Umstagram.

Let's take the Africa factsone. That's a tweet. But this time wise is April 4th, whereas the Al Jazeera story was the second. So this may have gotten uh good pickup, but it's later on time wise. All right, let's try this one here.

Sort by most interactions. There's mostly pages, although this is a group here with two hundred and ten thousand members. This goes into online news. Alright. So what we've arrived at here is from the original question, is there anything nefarious at work?

Initially, from a timing perspective, it looks like the the the start of this was the Al Jazeera was that tweet that led to the Alge Zero article, this one here by Rebecca Roseman. This one has the earliest time frame. It also has the highest amount of engagement, um, and certainly has uh a tremendous amount of uh sway when you we look at the network chart and see how fast it is uh s it spread onto mult many many different types of media. It's pretty clear that this one was the launching point. But it began with this tweet here, which again, when we looked at this one, was appeared to be uh not terribly uh nefarious looking.

It was an account that simply looked pretty normal. So let's take a look at the Odge Zero number articles numbers here. You have 30,000 interactions, Reddit, Instagram, Twitter, so that one wouldn't be a big deal. So if we want to put this fire out, um we will have to do a bunch of things. One, um we have to identify a couple of those junk accounts, see if they are in fact garbage if possible, uh, and if they are intentionally spreading fake news, then you can report those.

And that will break some of the chains of transmission, but you can see a lot of the damage has already been done. The bigger thing is reaching out to Al Jazeera and providing uh messaging that while these two doctors doing an interview did something and said a whole bunch of things that were really, really dumb. Um, that you should find messaging that's basically says put faith in the scientists and the doctors you have in your countries and let them validate any vaccine. Um certainly the very colonialist perspective that the French doctors were sharing is is not appropriate. But the science and the the technology exists within Africa, there's a number of outstanding universities, uh, and such that they can validate that a vaccine uh is effectively has will have passed clinical trials, uh especially when we see around the planet that clinical trials are beginning in many, many places.

Um, and as long as they adhere to the general format of clinical trials, stage one, make sure the virus doesn't the vaccine doesn't actively hurt anything. Stage two doesn't work at all. Stage three, does it work better than an existing solution? And then stage four, revalidate stages one through three. I would say that if I was having to message this out to the African communities to say that it is an international effort and that uh in all likelihood, uh the vaccine will probably not come from uh a colonial power.

Uh it will probably come from uh other parts of the world, you know, China, Singapore, uh, South Korea, places where they have extremely advanced healthcare uh and laboratories, uh not the United States, which or or Europe, um, it's possible the Europeans would be ahead of the United States, the United States, not so not so good at healthcare. Um getting messaging to this node to try and effectively reach back out to all these nodes to shut down some of that spread would be the way to tackle that. The two tools that we've been using tonight, Crowd Tangle is from Facebook. Uh it's uh Facebook software, they bought the company a number of years ago. If you are uh a former customer before they were acquired, uh, or uh you are a legitimate journalistic outlet, uh, you can get access to Crowd Tangle and it's free, which is really nice.

And you can see they do cover uh a fair amount of of data. The software that does the virality mapping, this is from TalkWalker. I'm using TalkWalker's Quick Search here. Uh, you can also use the analytics, and the analytics side is I find really helpful for being able to export the data and then you know drill down and do things like language modeling and stuff. In this case, we don't really have to because all of the negative commentary followed a similar enough format that we could construct a very simple query.

But for a more complex, like where you'd want to either get more advanced or use some of the built-in machine learning tools. But you can see the real advantage in this particular situation is looking at the highly engaged stuff and being able to trace back where did uh the problematic information begin? Where did it end? Uh how fast did it spread? It's kind of like in a lot of ways, it's kind of like modeling uh the pandemic itself, except that in these cases you have much better uh chains of evidence because you don't need to go and test anybody, right?

Here's that same exact tweet that started the Al Jazeera story, and then on April 4th, we see it jumps to uh this uh this account here, Tony Rudiger, and then it that it then spreads from there. So identifying you can't this would be uh a node to reach out to, but you really want this node for uh information spread because this person here clearly did not, I don't think intend to make a big deal out of this, otherwise you would see you know more pickup of just this node. But here, uh in fact, let's try that. Let's go to this node. And can I do a virality map on this one?

Don't know if I can. So let's stick the URL into quick search. So this node really was the source node for so many of these things. But again, it doesn't look like that person actively went to do it. It looks like it was picked up by the news stories and then shared out.

And then that creates additional spread here, this Facebook one. One of the um one of the other challenges that we'll face trying to trace this down is that a lot of what's happening is in Facebook groups themselves. You can see there are Facebook groups like this one, uh Ugandans at Heart, four hundred and twenty-two thousand members. So one of the strategies would have to be going into the Facebook groups where this got a lot of interaction and posting, you know, again, whatever the palliative messaging to help overcome anti-vaccination um news. There as well to the to the extent that you can.

But you have to, as you see, you can't use just one tool. You do need to use a bunch of different tools in combination together in order to do this detective work and to find the answers that you're looking for, and then find what are you going to be your solution points? How will you go after those solution points? So it's a really good question. Challenging question for this type this type of information in the world we live in.

Ultimately, was this a plant by you know some nefarious uh entity? It doesn't look like it. It does look like two people on TV said some really stupid shit um that they had no business saying and it has created the the storm that it has. Um so it that in some ways that's going to be harder to fight because they did actually say those dumb things. Um but you have now have a work back plan to go and and close this down.

So that's how you do the detective work. Um if you've got questions, you know, leave them in the comments and stuff. Uh stop by uh next Saturday night. I try and do these sometime around you know 7 p.m. ish Eastern time on Saturdays.

Um stop by, say hi. Let us know how you're all handling uh the pandemic itself. And uh talk to you soon. Take care, everyone. Bye.

Want help solving your company's data analytics and digital marketing problems? Visit TrustInsights.ai today and let us know how we can help you.