The Future of the Internet

Today I attended a lecture: Challenges and Opportunities for Tomorrow’s Networks by Nick Feamster.

Nick began the talk by setting the stage, enumerating trends that are rapidly and dramatically changing the Internet. More people will come online, that those people will represent a greater part of the globe, there are more things on the network and wireless access has increased. Before turning to other aspects of the talk, I want to pause and consider the global piece. I agree with the argument Nick was making, but then saw myself comparing it with some of the rhetoric that has accompanied the Internet for years. For example, calling it the possibility of creating a “global village.” Nick reminded me of how decidedly un-global that the Internet has been. But it may now become so. We’ll know when English ceases to be the dominant language of the Internet. Chinese, which makes sense, is gaining significant ground.

There were some things I expected to hear about in the talk. One I did not was about the struggles that the Networking community have. I had no idea, of course, because I’m in HCI where we’re never suffered from any legitimacy issues concerning whether we were an appropriate discipline within Computer Science. I’m being sarcastic of course.

I’m not sure I’m doing complete justice to Nick’s arguments, but here’s my version anyway. What I heard might be characterised as a success disaster. What does an academic research community do when something, in the case of Networking it’s the Internet, is very successful and much of the innovation comes from industry. If HCI has Apple, Networking has CISCO. Of course HCI doesn’t have the Internet, it just has millions of people repurposing and using all sorts of technologies in all sorts of ways for all sorts of things. They do it whether or not we say it’s useful or even usable (text messaging on a cellphone for example), and we have to understand why.

He shared some examples of the concerns levelled at Networking. Networking, like HCI, and unlike Theory, does’t have open problems (let alone ones you can rank in order). It seems to me that both disciplines are shaped by the relentless evolution of the network and the applications (and threats) atop it. This strikes me as being true of HCI. HCI went home because technology designed for the office wound up there because people wanted to try it out. We can not easily predict the future, whether it be the network or the human-application and increasingly human-network interaction. And to try to structure the discipline as if we could is like banging a square peg into a round hole. Rather than having open problems that remain open over time because they can, we have problems that open up because of the constantly and mutually shaping relationship between technology and society.

So, networking applications and threats co-evolve. Testing new technologies is also challenging for networking. It’s hard to build a representative infrastructure inside academia, although we at Georgia Tech are insanely lucky to have the Office of Information Technology (OIT) who support the campus and all of its networked functions but also have an orientation to and engagement with the research side of networking. Nick highlighted that, and I agree, and just want to provide a shout out to OIT.

The majority of the talk was Nick presenting his approach to moving the field of Networking forward. I’m not going to do justice to software defined networking, so instead I refer the reader to his blog. I’ll just say that he’s built an amazing set of collaborations, with students and colleagues. Even if I hadn’t learnt about all the other things I did, I left with something new to think about and it was a great Imlay lecture.

Revisiting Visions

So who knew that a blog would encourage me to think harder.  Almost certainly some researchers I know, and I apologise to them profusely, but it’s still a new experience for me.

I wrote about vision (and strategy, but we never mention the latter without mentioning tactics). I said that I felt it didn’t come naturally to me, that I was more instinctual.  Beginning with instinct… here are the things that seem important to me.

1. Wellness Informatics. I recently wrote about my version of this idea but its something I’ve been thinking about for about a year. The gist is that  health informatics (or more recently biomedical informatics) is largely (not completely, largely) focused on a medical response to health issues. But health and wellness are important partners. Wellness takes place in a community, and possibly without reference or interaction with the medical establishment. And what really triggered this idea for me was that in some communities, the medical establishment was a complex interaction. It depends from where you start.
2. Human Network Interaction. I have a long standing interest in making the Internet/home network a better user experience. What the popularity of the Internet has successfully proved is that a network architecture/protocols designed for technical specialists is miserable for end-users at home, not to mention technically trained people at home. What Keith Edwards and Nick Feamster will tell you, and I agree, is that this situation is probably resolved when HCI moves “down the stack” i.e., when networking and HCI are co-designed to meet the unique constraints and opportunities presented by unmanaged environments.
3. Narratives of Reach in ICTs. Paul Aoki first explained the importance of cross-cultural flows in computing. He made a compelling case that religious organizations are using ICTs not just to expand out of Westernized countries to emerging nations but vica versa. Research and conversations later, I understand that religion is a fascinating place to examine how emerging nations are using Information and Communications Technologies (ICTs) to expand into the West. One thing that’s important about this for me is that it turns a traditional narrative on its head. I participate in communities where ICTs are helping Western corporations expand their reach into emerging nations (heck I worked on such a project).
4. Distributed Intelligence in Human-Robot Interaction. Robots are a fascinating computing platform (in so many ways). In the last few years a quiet revolution has occurred. Robots have always been a part of our collective narrative (e.g, in science fiction) but in the last few years they’ve been quietly moving out of our imagination, away from our screens, and into our homes. That’s a change… I think that this puts pressure on robotics, and on human-robot interaction to devise modalities by which we may interact. The focus seems, at least to me, on making robots more smart. And that’s a good thing, but I think it scopes the design space in limiting ways. Specifically, I think that if the design space accounts for how/why people want to engage with robots then it opens up the design space to a type of human-robot distribution of intelligence. And, I think that’s the relationship people actually seek with robots.

Each of these is clearly a product of the interactions I have with the students I work with. There are some interesting cross-cutting themes though.

One is instinctual: about looking at problems the other way around. Arguments frequently have a temporal-linear narrative. The last two sort of exhibit this property. Narratives about the expansion of ICTs have them reaching out from the rich, urban, industrialized, to the poor, rural, pre-industrialized. But it can be and is also the other way around. Arguments about intelligence dominate the rhetorics of robotics design, but what about arguments that propose emotional engagement. I’m not saying that dominant arguments/narratives don’t have their place, but I am saying that considering the possibility of what’s not there is well mind-opening.

Another example: decomposition. Software Engineers spend a significant amount of time focused on decomposing a problem into a series of modules that can be worked on individually. But if you think of software as a linear-temporal activity then the process of reassembling them, of creating the whole from the some of the parts, becomes much more visible. And I once argued that it was that process, the process of recomposition, that was why software engineering was so human-centered, it was the need to be able to put things together that drove a significant number of the collaborations required to keep all the individually separated parts in alignment so that they would fit back together again.

Funny I always think that I didn’t pay much/enough attention in John L. King’s class about argument morphology. Perhaps that’s not true. I thought my colleague LP was the one who was paying attention.

Another theme which I hardly know how to express is to do with considering the extremes. I’m not the first to think of that, nor would I claim to be. I have colleagues who have research that takes place in countries like Liberia, or among Atlanta’s urban homeless. We have called this computing at the margins. I feel that a lot of the projects I’m involved in have a feel of take something that works somewhere, in a particular context, and then watch it fail or change in a different context. That certainly describes the Home Networking research. I think it also describes the focus on religion. Religion is ubiquitous and a site for many interesting and diverse uses of ICTs, but it’s not a central topic in HCI. Well it wasn’t, we’re working on its inclusion. But why? Because it’s ubiquitous, because religious values have long shaped the appropriation and rejection of technologies, as well as been some of the major reasons driving adoption. And most importantly because it allows you to look at non-religious use in a different light. It is a new frame, a new perspective from which to re-examine what otherwise gets lost because it is assumed. It also captures my interest in Wellness Informatics. I’m far more interested in the cases where the relationship between the community and the medical establishment and its knowledge is not straightforward, because it reveals the all important hierarchies.

But, what pulls all this together? That’s the question I have going forward.