Facebook, Twitter, and the Middle East
Too few service providers, network latency, and not enough smartphones limit the effectiveness of the Middle East's social networks
Steven Cherry: Hi, this is Steven Cherry for IEEE Spectrum’s “This Week in Technology.” Take a moment and think of all the things you love about Facebook. It introduces you to new friends. It lets you reconnect with old ones. Your news feed keeps you up to date. It’s fast. It’s reliable.
Of course, if you’re a Facebook listener in Egypt or Iran or South Africa, that list is probably a bit shorter. It may not be reliable, and it may not be fast. In fact, the social network’s performance in those countries can actually be pretty horrible. Ditto for Brazil, Indonesia, the Philippines, and Russia. Is the problem in all these countries the quality of the Internet there? My guest today says no. The problem is the nature of centralized networks like Facebook.
Ben Zhao is an associate professor of computer science at the University of California, Santa Barbara, and he’s the author of a number of studies on social networking performance. He’s also just returned from visiting the Qatar Computing Research Center in Doha, where he participated in a roundtable discussion on the challenges of social networking and social media particular to the Middle East. He joins us by phone from Santa Barbara. Ben, welcome to the podcast.
Ben Zhao: Thanks, Steven. How are you doing?
Steven Cherry: Ben, let’s start by asking, How important is social networking in the Middle East? There’s been some disagreement about the role played by Facebook and Twitter in the protests and reform movements there. Some people say they were completely essential, and some other people say just the opposite. You just got back from there. What’s your impression?
Ben Zhao: I think in general it has been very impactful in the sense that when something does happen, information can spread very quickly, especially with Twitter. So talking to folks from Al-Jazeera and other news medias local to the region, you get a sense that social media is one of their main outlets for how to gain access to information, and the real-time aspect of it makes their job much easier. And they can then send out sort of traditional reporters and get local contacts on site where the action is happening at a much faster rate than they can perhaps traditionally.
Steven Cherry: Do you think Twitter is especially important because it’s so easy to use on a cellphone?
Ben Zhao: I think yes, in some ways. I think in the U.S. and in other parts of the world where smartphones are very much sort of nearly being commoditized, I think that is a big factor. I’m not sure about the Middle East, though. Right now, most of the folks that I’ve talked to, they talk about phones in the Middle East as not having nearly the capacity, not the intelligence, the type of advanced circuitry that we have elsewhere. And so the phones are everywhere in fact, but most of the phones in the Middle East as I understand it are typically older-style cellphones that don’t have the full Android or iOS capacities that, you know, standard normal Americans might think of. So in that sense they’re not really hooked on Twitter because of their phones. Twitter, I think, is being used by news media people with computer connections, through those kind of means. Of course when Twitter was blocked, Google and Twitter did get together and they did in fact do a phone-tweet thing which was very interesting, where you could call into a phone line and actually speak and someone would either automatically transcribe it at the other end or do it by hand and have the result be tied into a Twitter stream.
Steven Cherry: So I guess let’s just step back and tell us, you know, what is the experience of the Internet in the Middle East and Facebook in particular? I guess it’s different in Egypt versus Tunisia versus Qatar versus Libya versus wherever, so maybe you could just take us on a little tour of the Middle East Internet.
Ben Zhao: Because of where traditional servers are placed—you know, Twitter and Facebook and some of the other social networks—they’re mainly based in the U.S. And so most of their data centers, most of their servers that process actions from the users do reside in the U.S., and what that means is that many typical user actions—like posting a comment, or sending a “poke,” or adding an entry to a photo—most of these kinds of things do have to go through data centers in the U.S. And that means multiple potential round trips between the local user, who might be in the Middle East or elsewhere and the data center in the U.S. So there are certain things—there’s a lot of things that people can do to speed up the process, but ultimately, signals do have to traverse an extremely long distance. So if you are in Egypt, in the Middle East and contacting Facebook, there’s only so much you can do to optimize the experience—the signals do have to go, you know, back and forth across the Atlantic Ocean several times potentially before the user sees their actions take place. One of the other properties of the Middle East Internet is that it is much less redundant, much less complex, if you will, more sparse than Internet elsewhere. So it is possible for things like censorship to happen by simply shutting down a small number of ISPs. Whereas I think that same type of action, if you were to try to enforce that in the U.S. or Canada or Europe, would be very difficult simply because of so many ISPs having a redundant path around the Internet. Shutting them down and doing a full partition would be much, much more difficult.
Steven Cherry: Yeah. So I guess in Egypt where the Internet was shut off for about four or five key days during the protests there, there were only really five big Internet service providers, and it wasn’t that hard for the government to basically order them all to shut down or ask them to voluntarily shut down. So you think with a lot more ISPs it would be a lot harder to do that. What would it take to bring about more ISPs there?
Ben Zhao: That’s a difficult question. I think there’s a lot of issues local to the region, and some of it’s economic, a lot of it’s political, and so, you know, I don’t understand what some of the challenges might be for an ISP trying to grow into that space. I do know that there are European ISPs that have local access points in the Middle East, so potentially, Egyptians could use a phone line, make a long-distance call, dial up to a European ISP and get access that way. Of course the downside is that the cost of doing such a thing is much higher and probably not practical for a traditional citizen in the region; long-distance phone calls or dial-up is relatively expensive.
Steven Cherry: And also dial-up would have a pretty poor network performance on top of all of that latency that you were talking about traveling across the ocean a couple of times.
Ben Zhao: And locally, as I understand it, the capacity of local network links is not quite the same as, for example, what we might expect in Europe or even in South Korea, especially South Korea actually or in the U.S., and so capacity-wise, you know, a large volume of traffic would cause a lot of congestion, which in turn would cause a lot of delays. So that would also dampen the experience for users in social media.
Steven Cherry: Now I think in a recent paper you’ve argued that that latency issue at least could be addressed by partitioning a network like Facebook. Is that right? And how would that work?
Ben Zhao: Well, so we’re not actually quite proposing partitioning—actually we’ve tried partitioning social graphs like Facebook in the past, and it’s actually proved to be very difficult. The social graph is resilient unlike anything else we’ve seen. I or what we did try to argue was that by placing certain servers—augmented servers, if you will—locally in a particular region, you could speed up some user interactions. What we found was that even though the world is very much a small place, and most social users have lots of international friends, there are still quite a fair bit of local interactions, you know, between users in the same town, between users in the same city or the same country. So it would make sense, for example, to locate some local servers, place some local servers in a particular region and to have them target those type of “local” interactions. And in that sense they wouldn’t necessarily have to change the entirety of how Facebook or Twitter might work, but it would basically target specific local communication and make that faster. And then essentially there would be some potentially small sacrifices in terms of consistency, but ultimately I think as Facebook and other social networks, as their users grow to be more and more international, those kind of sacrifices will probably have to be made sooner or later.
Steven Cherry: So let me see if I understand this correctly; the argument against partitioning is basically that you can’t separate links between friends by geography because people have friends all over the world or at least want to communicate with friends all over the world, and so I guess what you’re suggesting is to add a certain amount of redundancy. Let me just make a comparison and see if this is what you have in mind: It used to be that—we at Spectrum are in New York City and if I were to send a FedEx envelope across town, in the early days of FedEx it would go all the way to Memphis, Tennessee and then back to New York again. And then finally FedEx developed like a New York hub, and they would look at an envelope and say, “Oh, this stays in the New York area; we’ll just keep it here.” And Facebook could do the same thing with packets, I guess.
Ben Zhao: Exactly. I mean, that’s a very appropriate comparison. It’s a little bit different in the sense that in your example, the envelope is targeted only for a single destination, whereas in social networks, typically every single thing that you type or say gets broadcasted out to a number of friends. So there is that respect, but like you said, yes, so if there were local hubs, then that would greatly improve the performance of local users who at least are interacting locally. And obviously, in recent we’ve seen how important local interactions in the region can be. So performance between a friend in Egypt and someone in the U.S. might still take a while, but local interactions would be much faster.
Steven Cherry: Very good. Well then, whatever the role of social networking has been in the Middle East so far, it’s pretty clear that the Internet and communications are becoming synonymous, and the work of researchers like yourself is absolutely essential in promoting democracy and freedom. So thanks for your work, and thanks for joining us.
Ben Zhao: Oh, thanks very much. It’s been my pleasure.
Steven Cherry: We’ve been speaking with computer science professor Ben Zhao about the particular challenges for social networking in the Middle East. For IEEE Spectrum’s “This Week in Technology,” I’m Steven Cherry.
This interview was recorded 19 April 2011.
Segment producer: Ariel Bleicher; audio engineer: Francesco Ferorelli
Follow us on Twitter @spectrumpodcast