 You are listening to the podcast of the Alexander von Humboldt Institute for Internet and Society. We report on the leading role that new technologies play in the context of the Global Information Society, interviewing academics and industry leaders. We do a little podcast with Abrie Doria and Eleven Davis. Abrie Doria is a free consultant with lots of experience in internet architecture design in the wider sense. And so is Eleven Davis. We just had a workshop on the future internet architecture and I thought we speak a bit about what we got out of this workshop. The last point we discussed was the way the architecture might or may not change in the future. Abrie, give your prediction what you think we're heading to. Well, as I said in the paper, I don't know exactly where we're heading to, though I do know that it will change. There are various things that will change it that will make the policy discussions we have now less critical, less what we have to talk about. What I do know or what I do believe is that it will change in small increments, but at the end of a decade it'll look different than it is now. Eleven, do you expect any revolutionary disruptive changes? In a sense, from the point of view of people who use the internet, there will probably be something that looks quite disruptive. There will be changes, people will invent new ideas and it is quite possible, given our experience over the last 150 years, that somebody will come up with a very good idea which will actually significantly change the way that the internet works and how it is used. Give me a concrete example, where would you expect changes? Where would I expect changes? I think the changes will be about the internet will become ever more embedded in our lives. We've seen over the last 15 years the information economy has become critical to our lives and certainly those of us who work with the internet find now that we're addicts. We can't avoid using it, we get very itchy when we don't have access to the internet. That wouldn't require a change in the architecture, that would just mean more and more people use it for more and more things, but how would that impact the architecture? I think the answer to that is who knows. There are certainly going to be many more devices that are going to be accessing the internet. We are already at the state where there are probably more devices accessing the internet than there are people on the earth and each individual person on the earth, well, I doubt that everybody will be in contact, everybody will want to be in contact in 25 years, but certainly the internet will have spread a lot and a lot of our devices are going to be network connected, network enabled, so that there will be a lot more data and we will have considerable difficulties if we don't have good ways of accessing that data and knowing it and categorizing it and looking after it. If you think about the way that books are handled today, the indexing of books and being able to access and organize those books and being able to define the information in those books is critical to them being useful. If we can't index the information and know about the information, hopefully in perhaps a better way than it is at the moment with Google, which basically goes around and tries to find all the information and fails with quite a lot of it, the information will not be useful to us. We need to be able to index it and I think that's possibly somewhere where we may have a revolution that somebody finding a good way to actually keep track of all the information and you being able to use that information in a way that is better than it is today. I know that, for example, that people's ability to find information is very variable and it has been the case that there has been a concentration of the number of sites that are actually used. So Google has used a lot to find things and there are a very small number of sites that people use to find information. It's often found in a very rudimentary fashion. There's a lot more to do. So the Internet of Things, that is sort of a recent catch word that tries to capture some of what you've mentioned. Avri, what do you think how that will impact the Internet architecture? Okay, two things. One is the Internet of Things is another one of those overlay networks. So I think that in terms of that it will change some of the external features of the network. I actually don't know that it will change the architecture of the network but then again I'm not really sure. I know what you mean by the architecture. If you mean, as I was expressing in the paper, will it change the principles on which the Internet is created? I don't think it will change them a lot. Will it change some of the artifacts, some of the addressing methods, some of the search methods, some of the how things are controlled, who has authority, who hasn't, is the authority distributed or not? Yes, it will change those. In fact, one of the aberrations we have in a network architecture at the moment is in terms of a distributed architecture, we have a single focus of a naming architecture. So I think that in fact that will do something to fix that and bring the naming architecture more within the actual architecture of the Internet, making it more distributed in terms of how autonomous is a system and yet still able to cooperate, how the authority is distributed for naming and such, for addressing and such. So I think the architecture will actually deepen, will actually become more mature but I'm not sure that what I consider fundamental to the architecture will actually change its instantiation. It's the way it looks, the way it's deployed. I do believe it will change. If you think of new ways of naming things on the Internet, is that related to what people mean by the Internet of Things, namely attaching sensors to objects and make them interact? Well, they don't necessarily attach sensors to objects. Sometimes they just attach things that can be sensed to objects. But no, that's one of the things that falls out of the Internet of Things. What falls out of the Internet of Things, when you have something that follows an artifact from its creation to its destruction, it's going to pass through many different authorities. It's not going to have a consistent name. It's not going to have a consistent address. It's not going to have all the properties that relate to that are going to be different yet they're going to be related. And that object, that entity will come under many different authorities at different times and they will all have different methods for naming and identifying it yet those things will be linked in some sense. So it's a complexity to that. And so is that the Internet of Things? No, the Internet of Things is for me to follow my stake from when it was a side of beef and they all have their own identifying features and they all have their identifications and those are all owned and controlled by different things. Yet you're going to have people that want to sort of have a coherent picture of what a particular person has at a particular time and then there's the privacy issues with that. And so it makes a much more complex naming and authority control. But no, it's not about that. That's just one of the side effects of it. And certainly it's a challenge when you think about the lifetime of an object which goes through several ownerships. One of the things that people want to do is to make sure that the new owner can't find out much about the past life. The very simple things, you have a crate of paper towels that comes from the manufacturer and it passes in a lorry through various places and goes to a supermarket and is sold. And if that paper towel records its history and you have to also record its ownership, how do you do that? Is the thing that is attached to the paper towels going to have enough capacity to record its ownership but not record its history and how do you control all of that? This is an immense problem which, to actually get this right and actually to do something useful with it is going to be a very big challenge for the engineers and the technicians and the policy makers also to actually understand what is needed in order to control this and yet make it useful. There's this popular phrase called privacy by design. Do you think that's a useful way to look at this problem? And would it affect, I mean could we imagine that as a component of the internet? If you mean privacy by design is design privacy considerations and from the beginning I think it's necessary. I think we've seen that in the discussions in Internet of Things for example and in other places that you have to look at any of the technological things. For example in technology it's always useful to add an identifier. It's always useful to add perhaps a globally unique identifier or at least something that can and yet if you're looking at privacy issues that is always a negative that can be used to identify things too closely, to abrogate privacy and so you have to have those discussions right at the beginning of how can we solve the problem that needs to be solved technologically without building in something that can be used to harm privacy because so often the things that governments and others use to abridge somebody's privacy are artifacts that were put in for technological convenience. It was useful to do that and so therefore it was done but as soon as you put in something that identifies that creates the history and the history not necessarily created in the object but in the sensors and the computers that talk to the sensors outside of it once you've got something that can identify a life history then you've got the ability to do something with it that is an abridgment of privacy. Then again sometimes you want it and this is that whole problem you have between the freedoms and the protections. If what you're following is medication and you want to be able to follow a bad batch of drugs to the people because they've been found to be from a bad batch then you notice it didn't bring up stakes and then you want to be able to track the history so we both want to be able to track the history to protect us and we want to not follow the history because we want to protect the freedom so once again we're in that tussle that says make me free but protect me please. We don't want our medical insurers to know that we're using that drug so there's a two-way process that you want to be able to tell somebody who has that drug that they shouldn't use it anymore but on the other hand the fact that you've got it mustn't go back to somebody who might misuse the information to charge you more money because you've not told them that you're ill in that particular way so yes there are big problems with that sort of thing and the thing that is always said in the technical world is that you should always design these things in from the start because if you don't do it it will be incredibly difficult to retrofit it I wonder what that actually means to design it in from the start if it's so contradictory what we want from it how can that be designed in? Does it just mean that the user controls what to do about certain data? Ultimately you've got to think very carefully about the requirements and that means that you may find that some of these things don't happen as rapidly as you may have expected them to do because the cost is great the amount of effort that's involved in designing it correctly is difficult and the risk on all of these things is that somebody will tight a short circuit this on a few things and after that there will be a backlash against it we've seen that time and time again that people have tried to short circuit these sorts of things and you then get a backlash against it and the good effects that you might have had out of it are lost because people don't take the trouble to do it properly the first time So the final aspect I'd like to bring up is best effort somebody in the workshop missed best effort as a design principle mentioned in your paper and I think to the surprise of the audience your take on best effort was all that was never a design principle and I think you confirmed that perhaps you might explain that again how you understand best effort because lots of people think that it's more than a sort of that it's actually a good principle, best effort that this is what distinguished the internet from the traditional telecom networks in many ways best effort was the best that could be achieved there certainly was intention to try and deliver the same sorts of stories that the telephone companies were keen on and if you think about the original purposes of the internet which were was the military yes the intention was that the data had to get through and it was the best you could possibly attempt to but on the other hand the military is very keen on generals words getting through to the other end to be told to the soldiers because generals are important and if the grunt on the ground if his package is dropped so what? it turned out that the technology that we had available to us and that we have used subsequently actually was not very good at doing that prioritization as a result the generals words actually were treated much the same as the grunts words and in many cases they got dropped because there was a hole in the network somewhere and so best effort has become sort of enshrined as what we can do but that is really something to do with the technology that sits at the bottom of all of this and queuing theory and the difficulty of actually knowing the value of the data in the core of the network because how do you value one piece of data against another when they come from totally different sources and the other problem with this tends to be that we tend to turn value into money so people will pay in theory for better quality but unfortunately you then get to the point where everybody pays for the same quality and it just comes with the best effort again because they are all at the same level of quality ultimately but best effort as a principle is something that could disappear in the future I think it was ever a principle I think the types of service or quality of service are now quality of experience that people talk about have been the holy grails of internet technology just things that we've always failed at creating I think that they have become confounded with the network neutrality issues where all of a sudden people want to use these kinds of techniques to give commercial advantage to one content supplier's information over another where they were really supposed to be things that I could move into a higher level of quality when I wanted to watch something on TV and was willing to pay a little bit extra to have my picture not be grainy or it would work anyhow the problem is we've never succeeded in creating it the technologists have been working on quality of service forever now it should also be pointed out that there's always been somebody getting better than best effort and that's for example the routers inside the network have always had their data marked for priority so the fact that the network works is based upon some traffic being more equal than others so another miss destructed just easily thank you very much