On the weekend Elon tweeted: Twitter deal temporarily on hold pending details supporting calculation that spam/fake accounts do indeed represent less than 5% of users. Then he hit pause on his takeover deal with Twitter. Then all hell broke loose.
What are bots? On social media, they are automated accounts that can do the same things as real human beings: send out tweets, follow other users, and like and retweet postings by others. These social media bots work through an account like a real human user and are a mix of algorithms and coding to create process a task or a set of functions. So today, Alec and Sascha look at the world of bots and whether it’s as big a problem as Elon says it is.
Tell us what you think of The Dive – email us at thedive@equitymates.com
In the spirit of reconciliation, Equity Mates Media and the hosts of The Dive acknowledge the Traditional Custodians of country throughout Australia and their connections to land, sea and community. We pay our respects to their elders past and present and extend that respect to all Aboriginal and Torres Strait Islander people today.
*****
All information in this podcast is for education and entertainment purposes only. Equity Mates gives listeners access to information and educational content provided by a range of financial services professionals. It is not intended as a substitute for professional finance, legal or tax advice.
The hosts of The Dive are not financial professionals and are not aware of your personal financial circumstances. Equity Mates Media does not operate under an Australian financial services licence and relies on the exemption available under the Corporations Act 2001 (Cth) in respect of any information or advice given.
Before making any financial decisions you should read the Product Disclosure Statement and, if necessary, consult a licensed financial professional.
Do not take financial advice from a podcast.
For more information head to the disclaimer page on the Equity Mates website where you can find ASIC resources and find a registered financial professional near you.
The Dive is part of the Acast Creator Network.
Sascha: [00:00:02] From Equity Mates Media. Welcome to The Dive. I'm your host, Sascha Kelly. Let's play a game. I'm going to read you a series of tweets and you're going to guess the link between them. Olivia Titus I find the FTC case against TurboTax really interesting, particularly because tax accountants are extortion at two headlines. U.S. grocer LeBron James's online delivery deal sends Ocado shares skyrocketing at tiny care. But Love, heart, colon. Please text back to your friends at Big Underscore Ben under school clock. Bang, bang, bang, bang, bang, bang, bang, bang, bang, bang, bang, bang. Got an idea? Don't worry if you're stuck. The answer is they are all bots, automated accounts that can do the same thing as real human beings. Send out tweets, follow other users, and like, and retweet postings by others. So on the weekend, when Elon Musk announced his deal with Twitter was on hold because he wanted to verify the number of bots on the social media platform. We wanted to know more. It's Monday, the 16th of May. And today, I want to know how big of a problem are bots? To do this, I'm joined by my colleague and the co-founder of Equity Mates Alec. Alec, welcome to the dive.
Alec: [00:01:22] Hi, Sascha. One day you'll have an AI bot to do this for you, but until then, you're stuck with me.
Sascha: [00:01:27] That'll be the dream when I can collect the salary for doing pretty much nothing at all. Set the scene a little bit further for me. I know we're talking about bots because of Elon. We always talking about something because of Elon. But what happened on the weekend? All right.
Audio clip: [00:01:40] Well, the plot has certainly thickened when it comes to Twitter. Elon Musk says his planned $44 billion purchase of the social media company is temporarily on hold.
Alec: [00:01:48] So for the past month or so, we've been following Elon Musk's attempt to buy Twitter. And over the weekend he tweeted Twitter deal temporarily on hold, pending details supporting calculation that spam slush fake accounts do indeed represent less than 5% of users. They put the Twitter deal on hold and all hell broke loose.
Sascha: [00:02:10] It sounds like most things Elon does, really.
Alec: [00:02:13] So on Friday, Twitter shares fell initially about 20%, recovered a little bit for the day, fell about 10% as the market realised this deal was less likely to happen. Tesla shares rose about 6% as the likelihood increased that Elon wouldn't need to put his Tesla shares up as collateral. Many market analysts didn't think this was actually about Twitter bots. They saw this as aliens attempt to maybe get out of the deal or maybe renegotiate for a lower price. Whatever Elon's motivation for sending this tweet and I guess throwing this cat amongst the pigeons. Any social media user will tell you there's definitely a bot problem on all the social platforms.
Sascha: [00:02:56] Absolutely. I think everyone's seen it happen on their Instagram or social media accounts, like some weird text jargon thing happening. But give me the definition. What exactly are bots?
Alec: [00:03:08] So bots are artificial intelligence, field chat programmes that can these days do everything from ordering your food to deejaying your parties.
Audio clip: [00:03:17] Sometimes when you're browsing your social media feed, you might find an account or a post that just seems off.
Alec: [00:03:22] But on social media specifically, bots look like normal human accounts. They worked through an account with a profile picture and a name just like the rest of us. But behind that account, they're a mix of algorithms and coding created to process a task or a set of functions.
Audio clip: [00:03:40] Maybe it sounds like a janky product endorsement, or maybe the wording seems kind of funny.
Alec: [00:03:45] So for example, every time Equity Mates posts an Instagram post, there's plenty of bots set up to post links to dodgy foreign exchange trading courses or every time someone tweets the word Trump. There are plenty of bots set up to reply with links to his tax returns. The reason that there are so many is that bots can be made fairly easily. VentureBeat shared a guide to building a Twitter board in less than 30 minutes. And because they're so easy to create, we've seen a proliferation of bots. Now people talk about good bots and bad bots.
Sascha: [00:04:20] I wish I had a dollar for every time you said bought in the last minute because it was a lot of times. But I am wondering then what is the difference between a good bought and a bad bought? Because I only seem to notice the ones that annoy me. The bad.
Alec: [00:04:35] Bots. Yeah. Good bots and bad bots may not be the best description. More accurately, it might be bots that comply with the platform's terms of service and bots that do not comply with the platform's terms of service. But that's a mouthful. So let's just stick with this good and bad distinction that we've got going.
Sascha: [00:04:52] I like that. All right.
Alec: [00:04:54] But bots are automated accounts that are programmed for a specific and often helpful. Or at least fun purpose. And Twitter allows them to exist. There's the Netflix bot at Netflix on the so-called bot that automatically tweets when new content has been added to Netflix. There's the grammar place at Underscore Grammar Underscore that identifies grammatically incorrect tweets and offers suggestions. There's the Washington Post's Power Post, a bot that provides news about decision makers in Washington. You get the picture, Sascha. There's plenty of bots out there that provide useful information to social media users. They're the good bots. But then there are the bad bots as well. These are the bots that we all think about when we talk about social media bots. They're the ones that inflame political conversations, attack public figures, and generally make the Internet a nastier, more divisive and less factual place. There's never something that's trending that's not in some way promoted by bots, and these are the bots that Elon is worried about.
Sascha: [00:05:55] I think you're right when you said that the bad bots or what we think about in social media, so those might be a little bit more obvious to find and seek out. I'm assuming I might be wrong, but the good but some of those I would have had no idea. I would have thought that was someone actually working that. So how can you actually work out what accounts are bots and what accounts are run by people?
Alec: [00:06:16] I think a lot of the good bots identify that their automated accounts in their Twitter bio or their Instagram bio. Twitter also implemented a new feature that identifies good bots with a robot icon with a label automated. So people know when they see that tweet in their timeline that it's from an automated service.
Sascha: [00:06:35] So good bots identify themselves. Bad bots on the most part. Don't back to the EU on story. How is he going to quantify this problem? And surely if it was straightforward, Twitter would have already done it.
Alec: [00:06:47] Yeah, well, I own said in a follow up tweet to find out, my team will do a random sample of 100 followers of at Twitter. I invite others to repeat the same process and see what they discover.
Sascha: [00:06:59] Sir Alec, you're telling me that he's going to look at 100 followers of one account for a $44 billion acquisition of a platform that has a total of over 200 million daily active users. Isn't this kind of a ridiculously small sample size?
Alec: [00:07:19] Yes, that's right. And you're not the only one who thought that when they read that tweet, there was a lot of questioning in the Twitter replies. So Elon replied, and he said, Any sensible random sampling process is fine if many people independently get similar results for percentage of fake spam duplicate accounts. That will be telling. I picked 100 as the sample size number because that is what Twitter uses to calculate less than 5% fake spam or duplicate. Now, the important thing there is that since Twitter went public in 2013, they've said in all their public documents that they calculate less than 5% of their users being fake accounts. And alone, it seems, revealed how they do that with an incredibly small sample size. The reason that we know that he revealed it or we believe he revealed it is again, he quickly followed that tweet with another tweet and said, Twitter legal just called to complain that I violated their NDA by revealing the bot check sample size is 100. This actually happened, so a lot of it is sort of by this company, but he just keeps throwing bombs.
Sascha: [00:08:29] I feel like we have to put a pin in Elon's takeover tactics and keep on the task at hand, which is this question. So while Elon randomly selects 100 accounts, what does the previous research say? You know, this 5% problem. How bad is Twitter's bot problem?
Audio clip: [00:08:47] Pew Research Centre estimates that as much as two thirds of tweeted links to popular websites are.
Alec: [00:08:51] Generated by bots. Yeah, well, while Twitter allegedly only samples 100 accounts and while Elon plans to only sample 100 accounts, the good news is, Sascha, plenty of research have sampled plenty more accounts. And we can look at that data and get an idea of how many bots there are on Twitter. As we said, for context. Twitter, since they IPO'd in 2013, have been saying less than 5% of the accounts on their platform are fake accounts. A lot of research would suggest that that number is low. A 2017 study by researchers from Indiana University and the University of Southern California found that between 9% and 15% of active Twitter accounts are bots. Another 2017 study by PIXELATE suggested that between 3% and 17% of traffic leaving Twitter to outside websites were bots. And Sascha Spot Toro How to look at Elon Musk's Twitter account. Because, you know, he is the man of the moment. Of his 93 million Twitter followers. They estimate that 53% of them are fake.
Sascha: [00:09:56] 53%. Alec is extraordinary because that leads me to think if that's Elon. Problem. It's probably other people's problems as well. So is Twitter alone? Well, I already know that all the social media accounts have this problem because I live on Instagram. But just give me some stats about how bad it is.
Alec: [00:10:14] Yeah. This is not just a Twitter problem, as you've identified everyone who is on social media, whatever platform, Reddit, tik tok, Snapchat, Instagram, Facebook, we've all seen it. But let's start with Instagram. It has about a billion monthly active users. Do you want to take a guess about how many of them would be bots?
Sascha: [00:10:33] Well, Twitter saying 5%. I'm going to say somewhere between five and ten.
Alec: [00:10:38] Yeah. 10% of accounts are believed to be bots. About 95 million researchers estimate, which is a lot. It's four times the population of Australia. And Instagram not alone. Facebook in 2017 admitted they had 270 million fake accounts on the site. In 2019, they removed 2.2 billion fake accounts. And again in 2021, 1.7 billion fake accounts were removed. They only have about 3 billion monthly active users. So this is a huge share of the total. But it also shares how difficult it is to continuously remove fake accounts. It would be an unending game of whack a mole to identify new fake accounts and try and remove them.
Sascha: [00:11:22] It sounds like the worst gardening job that there ever was, just pruning that never ends.
Alec: [00:11:27] Instagram actually has a really instructive story here. There was a big Instagram bot purge in 2015, but the Wall Street Journal had a look after this purge and reported that potentially 8% of remaining accounts were still bots. It feels like a never ending problem that these social media platforms have to deal with. And Sascha, across all the social media platforms, Snapchat, Tik-tok, Reddit, Instagram, Twitter, Facebook. Carnegie Mellon estimates that bots are involved in up to 20% of the conversations across social media, but especially pertaining to elections and other political issues.
Sascha: [00:12:05] I know we're not a political podcast, Alec, but it's really hard not to draw connexions between society and culture and politics when you see how the social media platforms and the way that they work has an influence on the way that we connect with each other. So it does seem to be a massive problem across the Internet. But then my big question is why these social media companies have so much information on each user. They're so clever. You can imagine they could weed out the bots if they really wanted to. So I'm going to ask you that question. But first, let's take a quick break. Welcome back to the Dive. I'm Sascha Kelly and I'm joined by my colleague Alec Bresnahan. Before the break, we were talking about the size of the problem, something I never thought I'd say before. And now I want to understand why these social media platforms have armies of human moderators, and they're supported by artificial intelligence and machine learning algorithms. Surely they could just do more to weed out these bots? [00:13:09][64.4]
Alec: [00:13:10] They definitely could do more. There are simple things they could do. If an account is responding to every post that includes a certain keyword, they're likely to be a bot. Or if an account is tweeting 24 hours a day, at least deserves a closer look. It's either a business with multiple users tweeting around the clock or more likely it's going to be a bot. There are more extreme things these social media platforms could do as well. You could have to fill out a reCAPTCHA where you select all the photos with the traffic lights or all the photos with street signs before you can post a tweet or a Facebook comment or maybe every 90 days, you could have to verify you're a real person with two factor authentication. The most extreme suggestion is that Twitter should request driver's licences or passport verification before you can set up an account.
Sascha: [00:14:01] I kind of want to say I can understand why they don't do those things because, look, the things that I say on Twitter aren't that important. And if I had to do a two factor authentication text message, I'd probably think twice about what I said most of the time.
Alec: [00:14:15] Maybe that would be good if you thought twice before you started trolling the Equity Mates account. But I think I think the point here is not that they can't do more about bots, it's that they don't.
Sascha: [00:14:28] Okay, so explain that a little bit more elaborate on that.
Alec: [00:14:32] So simply the social media business model is built on advertising and because of that, because they need eyeballs. The most important metric for social media businesses, a monthly active users or daily active users? In many ways, it's the metric that rolls the Internet in the attention economy. Your active user count is how much attention you're grabbing.
Sascha: [00:14:54] And the more attention you grab, the more ads you can.
Alec: [00:14:57] Serve. That's right. And with 2022 expected to be the first year when digital ad spending crosses the $500 billion mark, it is a big business. The majority of Internet advertising is sold on a CPM or cost per thousand impressions model. So social media platforms are optimised to deliver those impressions. They make signing up as easy as possible. They make newsfeeds engaging to increase time on platform, and they work with push notifications and emails to get you back on the platform if you ever leave. I remember Facebook's best growth hack in the early 20 tens. When someone tagged you in a photo, Facebook would send you an email and you've got me. You know, Sascha Kelly tied you in a photo, but they wouldn't show you the photo in the email. And that got me back on the platform every time.
Sascha: [00:15:46] So I guess what I'm getting from you is that the attention from a bot in social media land is valued at the same equivalence as a real person.
Alec: [00:15:56] If daily active users or monthly active users is the metric that advertisers look to, that investors look to, that everyone is watching a bot or a real person as an active user counts all the same in that total number. So let's make you Elon Musk buying Twitter or Mark Zuckerberg running Facebook. You decide you're going to take a hard line against bots. Are you going to clear them from the platform? Are you going to set your moderating team to delete those accounts? Well, you might end up losing 20% of your user base, which hurts your monthly active user or your daily active users number, which turns advertisers off and might push them to other platforms with more users. It will turn investors off. They don't want to say monthly or daily active users going the wrong way. So then your share price gets hurt. People sell your stock, and as an executive at one of these companies, a lower share price hurts your paycheque. So that's why these platforms are pretty lax when it comes to cracking down on bots.
Sascha: [00:16:54] Yeah, but basically everyone outside of the platform must hate them. Advertisers don't want to pay for impressions to fake accounts. You said that there's been a study that thinks that Elon Musk's 53% of his followers might be fake accounts. That's a lot of inflation in these numbers.
Alec: [00:17:12] Yeah, Sascha, everyone hates bots. That's not surprising. Well, at least everyone hates bad bots, but advertisers in particular hate bad bots because they pay these platforms and they don't want to be delivering ads to bots that aren't going to take action and buy anything. An estimate from 2019 suggested that fake followers in influencer marketing cost brands $1.3 billion. But the challenge is there's not a lot advertisers can do about. Facebook's family of apps dominate social media advertising. And while advertisers have made a lot of noise over the years and challenged these platforms, they haven't really been able to walk with their wallets and go and spend elsewhere. We saw.
Audio clip: [00:17:55] Unilever threaten to pull.
Alec: [00:17:57] Its Facebook, Instagram, Twitter, maybe Snapchat, Reddit a little bit, and then there's not much else. So if there's bots on all of these platforms and none of them are really taking a lot of action, you don't have a lot of choice as an advertiser.
Sascha: [00:18:11] So what is the future of bots then?
Alec: [00:18:13] So as much as Elon is making noise about it now, and as much as this puts the whole Twitter acquisition in question, whether Elon ends up buying Twitter or not buying Twitter. One thing is for certain the future of social media, the future of the Internet in general is more bots, not less bots. In 2016, Microsoft CEO Satya Nadella said that bots are the new apps. Twitter and Reddit are designed to accommodate bots. All these good bots that we're talking about that add so much colour and a bit of life to these platforms in different ways, they are expressly allowed in these systems. Facebook, but in particular Facebook, Messenger and WhatsApp are investing heavily in chat bots. Don't be surprised if in the coming years you're interacting with a whole bunch of companies via chat bots and there are just so many more use cases coming out around bots across the internet. It's one of the main ways that we as Internet users are engaging with artificial intelligence. A couple of years ago, Forbes wrote an article six Ways Bots Are Positively Changing the World. So I think to end this episode, we should end it on a positive note and talk about some of those positive ways that bots are changing the world.
Sascha: [00:19:28] Mm hmm. So tell me one of your favourite bots.
Alec: [00:19:30] So UNICEF created your report, a free SMS based bot system that more than 2.6 million people in developing countries are using to report key issues in their communities. A number of bots have also been created to help civic engagement and for people to navigate government bureaucracy and in particular to register to vote. But Sascha, you also my favourite. My favourite is a bot called Do Not Pay. Have you heard of this one?
Sascha: [00:19:59] I think I have. It might have been mentioned on Equity Mates before.
Alec: [00:20:03] We interviewed Bill Browder, the author of Red Notice. Do Not Pay is actually his son's app, and it was a bot created to automatically challenge parking tickets. You would submit your parking ticket and it would draught a letter and send it to the relevant authority challenging it for you. I think these numbers are a couple of years old, but they'd helped challenge hundreds of thousands of parking tickets and save people millions of dollars. And now they're building a full robot lawyer that also helps tenants fight negligent landlords and the homeless apply for government support. So, Sascha, we have a bad perception of bots at the moment, there's no doubt about that. But the truth is we're going to have more bots in the future rather than less. But the challenge today for Elon, for Twitter, is finding ways to, first of all, define what a good bot is and what a bad bot is. And then secondly, find ways to identify and separate the good from the bad.
Sascha: [00:20:59] Well, you've already helped me realise that several of the things that I interact with on the internet are actually bots, and I'm quite fond of them. So I want to say thanks to the good bots today, and I think that's a good way for ending today's edition of The Dive. If there's a story you want us to cover, then contact us. Email is the dive at Equitymates.com/contact or just shoot us a message on social media. All those links as normal in the show notes below. Do us a favour. If you've enjoyed this episode, please subscribe. It just means that every time there's a new episode, it's right there in your podcast player feed ready for you to listen. Alex, thanks so much for joining me today.
Alec: [00:21:39] Thanks, Sascha.