Most of us assume, when trading messages on a social networking site, that we're interacting with a human. But you might be talking to a bot - software designed to, for example, tweet. @JamesMTitus tweeted a lot and gained a bunch of followers. Yet, he (it) was the winning bot in a contest held by The Web Ecology Project. The Project's Tim Hwang explains.
The WavesArtist: by Tamaryn
BOB GARFIELD: This is On the Media. I'm Bob Garfield. Earlier this year, 500 or so Twitterers received tweets from someone with the handle @JamesMTitus who posed one of several generic questions: How long do you want to live to, for example, or do you have any pets? @JamesMTitus was cheerful and enthusiastic, kind of like those people who comment on the weather and then laugh heartily. Perhaps because of that good nature or perhaps because of his inquiring spirit and interest in others, @JamesMTitus was able to strike up a fair number of continuing conversations. Only thing is, there is no @JamesMTitus. He, or it, is a bot, a software program designed to engage actual humans in social networks. He grew out of a contest to devise a social bot, a contest staged by a group of techies calling themselves The Web Ecology Project. Tim Hwang is the director of The
Web Ecology Project. He joins us from Berkeley, California. Tim, welcome to On the Media.
TIM HWANG: Hey, thank you for having me.
BOB GARFIELD: Tell me how the, the contest was designed. You set up a number of teams and they each had two weeks and, and were given a list of Twitter accounts to send robotic messages to?
TIM HWANG: We thought about it. It was sort of like social battlebots. We put together a list of 500 very active Twitter users, and the way we did that is we, we pulled a hundred people who really loved talking about cats online because, you know, cats run the Internet. And we pulled some of their friends randomly and some of their friends randomly to generate kind of a semi-connected network of people. And then the teams were said, okay, here’s the, the, the boundaries of this competition. Design a bot and we'll drop these bots into the network and see how they do. And the scoring was the bot got a point for every mutual connection it created, you know, it follows someone, someone follows it, and three points for every @reply, retweet, you know, conversation piece they're able to generate.
BOB GARFIELD: Now, this was an experiment about Twitter bots, but what is the bot presence online in things apart from Twitter?
TIM HWANG: Bots are everywhere online. Some might actually argue that there’s more bots online than there are, are humans online. There’s a huge number of bots that generate spam email, and certainly on social platforms bots generate a huge amount of content. They're mostly in the business of trying to get a message in front of as many people as possible. What’s rarer online and what our social bots are trying to do is focus on a much slower and deeper type of connection with people. They're interested in affecting the way people talk to it and, and talk to one another.
BOB GARFIELD: What was it about JamesMTitus that enabled him to beat the other contestants?
TIM HWANG: JamesMTitus was designed, actually, not to be shy at all. He would talk all day, all night to all sorts of different people. And I think most of the teams assumed that that would automatically be detected as an automated identity, right, and people would just ignore it. But it turns out the opposite is the case, actually. When you, when you talk a lot, and it turns out that people want to talk to you more and, and you’re also perceived as more human. One of the really interesting things that happened during the competition was a team launched a, a counteroffensive against JamesMTitus, and the counteroffensive was a bot called Botcops. And Botcops would basically go after James’ followers and say, you may want to be aware, we're detecting that James is a bot. Don't listen to him. And people were asking, oh, James, are you a bot, are you a bot? And, and James actually said, well, oh, I don't know, what do you think, or, you know, oh, that’s so funny. And actually, in responding right off the bat to some of these people, they basically just assumed that there was a human behind the computer on the other side, and he actually had people back off.
BOB GARFIELD: You mean, because his answers were nonresponsive kind of non-sequiturs, people assumed that it was a, a live [LAUGHING] human being? Wow, that doesn't speak too well for, you know -
TIM HWANG: For humans.
BOB GARFIELD: - humanity. [LAUGHING]
TIM HWANG: [LAUGHS] It’s true, it’s true.
[BOB LAUGHING] And these conversations actually end up being quite humanlike, precisely because they are [LAUGHING] sort of so random and non-sequitur in some ways. One really funny incident that we had was he was asking one of his followers, you know, what was a famous character from a book that you really liked? And someone said, Jesus. And his response was, you know, right on, bro.
BOB GARFIELD: All right, so what is the practical application of creating a, a Twitter bot or any kind of social bot? How can it be used for good and how can it be used for e-vil?
TIM HWANG: You know, these bots are cheap to run. They're almost free to run. So what if we were to launch a whole swarm of them to, to change the way people connect on a very large scale, right, 10,000, 100,000 people? Maybe these bots can act as a kind of social scaffolding, right? They bring communities together and then the bots sort of deactivate, right? And, and so what you are left with is an actual community of people. And we could say let's bring together all the people who are civically engaged, or they want to help out and, and promote charities, for instance. And all of these communities are quite balkanized, and what these bots can do online is help to bridge that gap between various communities. Now, in addition to building connections, one of the worries, of course, is that the bots might be used to destroy connections or to disrupt emergent processes happening online, right? You can think of the Iran election or the Twitter interactions around the Syrian protests. And there’s been some indication that bots are actually already in the mix, attempting to disrupt or support various sides of, you know, the—that, that particular protest. And so, I think there’s a number of good uses and bad uses. I think ultimately what these bots do is they're kind of a tool for social architecting, for better or for worse.
BOB GARFIELD: What would happen if a bot tweeted a bot? Armageddon, the world’s most disjointed conversation, what?
TIM HWANG: [LAUGHS] One of the things that definitely interests me is kind of all the things that happen when bots run into bots randomly, online. There’s a really interesting story that popped up last week where they discovered online that there is a 23,830,000-dollar book, a used book being sold on Amazon. And the reason that the book’s price has gotten to that point was that there are two bots battling it out. And it actually turned out that the bots’ strategy was always to outbid the other bot, right? And the two of them were just interacting together until you had this used book that was like 24 million dollars.
BOB GARFIELD: [LAUGHING]
TIM HWANG: And, and I think social bots, right, they in some ways could have this endless conversation with one another when they run into each other. You can even imagine, right, in the future, clusters of bots on social platforms online, which is just bots talking to other bots.
BOB GARFIELD: How do things like Twitter bots force us to change our behavior, change our understanding of the Net? How does it just—change our lives?
TIM HWANG: The fact that social interactions happen on a digital platform now means that software, in some sense, is something that will likely be able to interact with us in the way that we normally associate with just interacting with humans, you know, on a day-to-day non-online basis. And so, I think one of the things may be it, it presages and suggests is, you know, in the future maybe people have to just be a lot more aware that the interactions that they're seeing online, the type of social phenomena, the trends that they're seeing online could be human or, or they could be synthetic or, or an odd hybrid of both. I think the big initial thing that comes to mind, right, is this need to sort of be more skeptical about some of the processes that we see online. You know, the mainstream media, now especially, uses Twitter as such a proxy for what people think in the universe as a whole. In manipulating some of these processes you could actually imagine that you'd actually be able to manipulate the news cycle.
BOB GARFIELD: Tim, thank you so much.
TIM HWANG: Thank you for having me.
BOB GARFIELD: Tim Hwang is the director of The Web Ecology Project in Berkeley, California.