Cooperation Working Group
25th October 2017
At 9 a.m.:

CHAIR: Good morning everybody. Welcome to the Cooperation Working Group session. If you are interested in routing or routing, or however you pronounce it, it's the other room.

I really want to thank you, everybody for coming here this early in the morning. Somehow we always seem to get the morning slot but we just have to live with that.

This is a Working Group, this is not just a presentation session so I, as a formality at least, we need to approve the agenda, which is being shown here. I assume everybody is okay with the agenda or is there somebody who very strongly feels it has to be changed? I thought so. So, thank you.

I guess we can move on to the actual programme. And if there is somebody here who doesn't know what the Cooperation Working Group is supposed to do, we are supposed to liaise with external entities such as, let's say, the European Commission, governments, IEEE, ITU, IETF and so on. That tends to mean very different things at different times and it depends on what is going on but at least the idea is that we, one, keep an eye on what is going on in our environment that is not directly in our focus, and also try to help other organisations and work with them on things where we have common goals.

So, one of the organisations we try to work with fairly closely is the IETF, and I am glad to welcome Salam Yamout talking about what is going on there right now.

SALAM YAMOUT: Good morning everybody. And I am going to be talking about the IETF. So, first, disclaimer, no one speaks for the IETF. So I am not talking on behalf of the IETF, but I am talking about the IETF on my own behalf. It's very important because it's a wild beast. So, the IETF is the body, the real glue, of all the technical community, is the body that makes all the protocols, I like to think of them as algorithm, that then are translated into code and then they live on your machines, on your telephones and your computers and routers. So this is really the stuff, the Internet is made out of. There is 5,000 protocols so far. They are called RFC. And they are divided into around more than 100, between 100 and 150 Working Groups, currently being open in these categories that resemble the layers of the ISO but not exactly.

So, in these Working Groups the IETF is discussing matters of current interest for the Internet. You know, the current problems that they are trying to solve. But however, there is another area it's called the Internet research task force, and it's related, it's inside the IETF but here they are thinking about future problem that could come upon the Internet. So, let's say it takes about three years to implement the protocol if it is in the IETF today, it's going to take seven or eight years to see something coming out of the IRTF so it's basically a future‑looking exercise.

So, what is the IRTF thinking, what does the IRTF think that the Internet needs help with in seven years? Crypto Forum, right, is where all the crypto currencies and all the crypto things think about. Thing to thing, not Internet of things: My refrigerator going to talk to my washing machine. Internet congestion control, measurement and analysis, network function virtualisation, don't ask me what it is.

Network management, information centering networking, network coding, global access to the Internet for all, and human rights protocol and that is what I will be talking about today.

This is ‑‑ so before I go into the human rights thing, anybody can participate in the IETF or follow their mailing lists. Anybody can submit a draft document, nobody will ever be invited and if you have something to say, please go ahead and participate in the mailing list and start interacting with them. But why is the Internet, why is the topic of human rights important in the context it of the IETF, because forget, don't forget these are 1,500 engineers, they are pretty much nerds, and here they are thinking about human rights. They wouldn't have done it if they didn't think that it was a real challenge, a real problem coming upon the Internet. This finding has been also duplicated by the Internet Society futures report that identifies human rights as one of the impact area of the futures report. We are going to see a lot of bad things happening to human rights because of the Internet. So I think this is why the engineers, the nerds, decided to think about a topic that is usually not of interest to them. And they are really focusing on two areas, they are focusing on freedom of expression and on the rights to assembly.

So, they have been thinking about that for two years, the IRTF human rights protocol considerations research group was established two years ago in 2015. The researching whether standards and protocols can nail or threat enhuman rights. (They are) can the human rights characteristics of the Internet be degraded if not properly defined, described as sufficiently taken into account outright from the protocol development process. So they are really thinking whether they should, they are still thinking whether they should be looking at like a master RFC that rules all other RFC that has human rights consideration into it or whether every protocol is going to have to look into human rights consideration as like say a mandatory section in it. And they believe that an open, secure and reliable connectivity is essential; for example, encryption and this is not the opinion of a lot of people, not everybody agrees that encryption is a right to every person using the Internet, but the IETF does.

So there are four documents that being worked on, the most important one is the RFC 6973, and the other three will just pass by, they are very ‑‑ they are at the very early stage and this is the main one, it has a privacy considerations for Internet protocol and here they are thinking about, thinking about preference considerations when ‑‑ ‑‑ including privacy consideration in the protocol specification itself. So the idea is to be aware of the privacy consequences of the code that we are writing, and that suggests whether any individual RFC warrants a specific privacy consideration section will depend on the document's content. So this is the idea of every RFC having a section about privacy.

Also, there is some farfetched off questions that this group is asking, for example how does http impact privacy, and human rights. You know, I was thinking about this this morning, I couldn't ‑‑ I could find implication of NAT about privacy and human rights because if your network is using a NAT, then you are hiding behind and nobody really ‑‑ it's going to take them longer to find out who you are, right? But if, say, you don't use NAT, then chances are you are going to be identified more easily if you are a human rights activist or democratic advocate in in a society that doesn't accept such advocates, then you are in trouble.
So these questions, I believe, are not easy questions because, I mean, we are not used to think about the implications of code on our human rights and our freedom of expressions.

So, also, they are considering the implications of traffic interception, manipulation, threatening, logging and DDOS on human rights. So everything the Internet does, every technical protocol they have to think about the human rights consideration of it.

I will leave you the slides, you can read them later.

So, the other three drafts are, one, on the freedom of association on the Internet, the other one is on the politics of standards and the unrequested communications.

So the next meeting will be in IETF 100 is next month in Singapore. They will be talking about the human rights group on Friday, and remote participation is available if you want to follow the discussion remotely. After that, the IETF 101 will take place in March in London and because we are here in Dubai, and all middle eastern people then would be encouraged doing to the IETF in London, there will be plenty of fellowships available for people interested, from both the government and regular fellows to attend the IETF in London.

For example, these sample of IETF fellowship programme, I recognise two people from the Middle East in this picture. And we also have something called the IETF policy programme, one of my favourites, and where they take government people and they put them in a one week‑long emerging programme into the IETF and explain to them everything from A to Z, so they explain to routing, how the IETF works, and have a funny story to tell.

So I took my first Middle East delegation to the IETF programme last IETF when we were in Prague and the delegate from one of the countries, middle eastern country arrived, and says okay, Salam, Hi, where is my session? Which panel am I on? And I told him, well, there is no panel, and he is like, is it possible? A conference without panel? I said, yes this is a conference without panel. And he raised his land like half an hour later and says, I am really interested in routing, can you please introduce me to the routing people because I really would like to participate in the routing, you know, Working Group. And again, the answer here is, nobody will invite you, you can just walk in the room, you just look in the agenda, you find out the room and you go to. So these things are not culturally acceptable to some regions of the world but all we can do is really explain to people how it works and just like we had to learn another culture or language we can also learn the language of the IETF and all participate in making the Internet of tomorrow.

I think that's the summary. The IETF and IRTF make the Internet work better we believe. It has fundamental role in Internet administration. Of course it's the glue of the Internet, the substance, all the protocols that you are using. Your participation is critical to its success so please participate and for more information go to this website. Thank you.


CHAIR: Questions? We have five minutes for questions.

AUDIENCE SPEAKER: I can't see very well so at close up ‑‑ Lee Howard and my company is Retevia, but I am also a Working Group co‑chair on the IRTF and on the architecture board and what I really wanted to amplify and repeat suggest said, and amplify which is participation is extremely open, you don't have doing to meetings, you can just join a mailing list and for every Working Group there is a mailing list and also the IETF is in desperate need, in my humble opinion, my personal opinion, not anybody else's, needs more operators because there are a lot of vendors at the IETF who think they know what operators want from them but what they think operators want is to buy more equipment. We need operators to say here is how things will work well for us. Whatever your interest and expertise is, it's a wonderful place for participants to pick a Working Group drop in and participate with what is going on.

SALAM YAMOUT: That is a very good point and, you know, my presentation was made expecting more people from the Middle East in the room. I am a bit, you know, I felt a little bit humbled by talking to you about the IETF and knowing looking at the people, I know that you probably know a lot more than me about the IETF. And this absence of ‑‑ soly is asking for operators, I am asking for people from the Middle East region and from other regions but, you know, I mean, maybe we can think together about how to make, how to except for the IETF policy programme and IETF fellowship programme offered by Internet Society really have doing beyond this, we are open inclusive and really find ways to bring these people over. I mean, I think you know we are missing a point of view and we are missing friends that later on are going to be our allies as to later on becoming our enemies.

LEE HOWARD: I did discover recently the Internet Society has a programme for academics who can, get a free day pass or discounted pass or something like that too.

SALAM YAMOUT: But it's not enough. Let's think together about ‑‑ not enough saying we are open inclusive, we have to think harder to bring the people of the Middle East into this fora.

LEE HOWARD: I completely agree, the middle east is under‑represented.

CHAIR: So our next speaker is Karen McCabe speaking about IEEE and ethics.

KAREN MCCABE: I am with the IEEE I am a senior director of our international policy and affairs group. I do reside primarily in the standards department I do work cross functionally in the technology space and I am pleased to be here today and to share some information about what we are doing and on a couple of fronts may not be obvious to some, but it's definitely a priority for us moving forth. And I just realised, I think I changed the name of your Working Group, so I will fix the slides and reupload them so my apologies for that. So today, I am going to share a little bit of introduction about the IEEE and its standard association. I promise it will be brief because it's a large organisation and talking about every aspect of it would take quite some time but I want to do that to set the context so it makes sense when we talk about the specifics I want to get into on three major initiatives that we are working on that you see listed. One is on ethics and one on digital trust and identity and the Internet initial that I have has been going on for the last five years. So I will start with that.

So briefly, for those of you who may or ‑‑ or may not know about the IEEE it's building one of the largest professional associations in the world. We have an impressive number of members, about 420,000 in about 160 countries. It's a large organisation made up of many technical societies, 42 technical societies, so with that we are covering a wide range of aspects from power and energy, aerospace, biomedical, computer technologies, computer vehicular technologies and within each of those societies, technical societies, there is many various committees, they go pretty deep so the organisation is pretty broad and deep. So we have a very large army of technical professionals working on many aspects in electrical engineering but in many domains as well. We also, because of the size of the organisation and the structure of it, one of the main focus areas is putting on a lot of conferences, as you can see quite a lot of conferences, 1,600 over any given year. And here is where technical professionals, academics, scientists, enter presenters, come in, present their work they are doing in any technical spaces that they have aligned themselves with, and then with that we do a lot of publishing with that, we make those conference proceedings available as well as many technical journals that they have in the periodicals as well.

But we are lass standards developer, which is interesting. So sometimes bag standards developer in a large organisation like that is a little different, it could be a little challenging because our community is a little bit different, sometimes people think of IEEE and they think primarily of research and academics and things of that nature. But the standards association is probably the piece of the organisation because of being a standards developer, that is probably more industry‑focused. And our mission is to provide high quality market relevant standardisation environment so that it would make it easy for standards developers to do their work. And then it's respected worldwide. So in a sense we are kind of independent global community and we abide by what we call our ‑‑ in 2012 we worked with a few other organisations including IETF on open stand and that was really to highlight and to bring to the forefront a different type of standards development process that was more market driven, bottom up, open, inclusive, transparent and if you want to look at principles or values perspective. And we abide by those processes. Many of the organisations that work in market driven paradigm standards development have different types of processes and we do but we are all connected by those common principles of being open, transparent and inclusive. And our standards span a lot of technologies, as you can see here. As I mentioned, Tripoli has about 42 technical societies, not all of those produce standards, probably about 20 to 25 of them do and some of them listed here on the screen for you, reading pleasure.

So, what we ‑‑ this sort of a bit of a transition because I know you think of technical standards body and why I am talking about ethics and Internet and things of that nature. You know, we do offer more than what we call globally open or rules based organisation. We have, we are pretty old organisation in a sense we have 130 years of experience behind us, and we really are dedicated to building consensus on collaborative platforms. And we have this, as I mentioned a pretty large community of technical experts that have a tremendous amount of experience from various domains, industry sectors, multi‑disciplines, if you will. And we set standards but there is so much more that we should be addressing especially in these times when we are addressing the challenges of privacy, security, you know, technologies, at this rapid rate of development and society sometimes struggles with that from a policy maker perspective and even as from individual end users. And we are trying to take this very seriously in what we can do as an organisation to help address some of those challenges.

So hence why the IEEE in general but also the standards association like, we use term technology policy, intersection of policy and technology and see how we can leverage and we can deploy those technical experts that we have to help address some of these challenges, primarily from a technical perspective but not necessarily always from that perspective. It's equally important for policy makers to understand the tellical aspects of what they are addressing and when they are making decisions addressing some of these serious challenges but likewise it's very important that the technologyists really understand the other environments, the impact of the technology that they are developing, the impact of the technology that is going to be standardised and understanding a‑ha what has to be addressed and faced from regulation and policy and user perspectives as well. So with that, we are really dedicated to see how we can connect with technologists, with policy makers so we can bring that voice into those discussions and that experience and that talent into those, addressing those issues.

So, with that, we are really working to facilitate and promote technology environments, also that are open and inclusive and transparent, because we really do feel that it's a way, if everyone sort of at the table talking openly, we can maybe hasten some of the solutions to the challenges that we are facing today.

So with that, we take an integrated approach. One of the core themes for us is about trust. So we have several programmes that I will be briefing walking you through here shortly that are addressing sort of the issues of digital and Internet inclusion, digital identity as well as the impact of autonomous or systems and artificial intelligence, from a broader perspective intelligence systems. And we are seeing how we can work at an intersection of all of these so that we can enhance and inform the developers and the end users that we can advocate for collective action, especially when it comes to trust an agency over data.

So, the first one I would like to talk to you about is what we call our IEEE global initiative for ethical considerations at AI and AS, artificial intelligence and autonomous systems ‑‑ side note, we have no small titles for anything. And actually this even has an acronym but makes no sense. But basically in ‑‑ about a year‑and‑a‑half, two years ago we launched this initiative under what we call industry connections programme. That is another misnomer of a name, if you will, because probably about so or 12 years ago we controlled out this industry connections programme which was meant to be a incubator programme, where people and companies can come together and figure out what are the challenges, what is the best way to challenge this, do I really need a standard, do I need a registry, a database, a conference so it's a a broad‑based programme that many things fit under, but it has this unfortunate name of industry connections, so we are working on that because it doesn't it doesn't explain, if you will, the intent of that programme.

So, that is where this initiative formally sits under.

So, the work began in late 2015, as I mentioned it's industrial Czechs programme. Right now, we have over 200, I think we are close to 300 at this point, of experts involved working on this initiative, and they come from all around the world, as you can see, the EU, China, Japan, Korea, and I will get to this, but one of the primary outputs is what they call the ethically aligned design and they have the first version was issued last year and they are working on version 2. So we have been really working hard on this, and it's addressing many of these issues that are represented by these 13 Working Groups here.

So, this is an example, this is sort of the shot of the document, I can give you the URL and you can Google this if you want to see this but the second version will be coming out in December.

From this initiative, interesting enough, a series of standards have emerged as well, and these are hybrid standards, they are not necessarily to all highly technical stuff, they are really a working on the intersection of technical and non‑technical addressing some of these issues about personal data, well‑being of people and things of that sort.

Another programme briefly is what we call our digital inclusion through trust and agency and here we are trying to address that challenge about the threat to security, about everyone's data, if you will, and looking at it from a digital inclusion perspective. And here we are looking at things such as blot chain, technologies, among other things, frugal 5G, technologies that can help bring digital inclusion to people but doing so in a way it doesn't create a bigger divide and works to close it between the haves and have notes when it comes to technology.

And then we have lastly our Internet initiative was put in place about five years ago (notes) and this is really kind of a ‑‑ (notes) this ‑‑ how we can engage our technical communities with policy community, and it was really done through the Internet initiative, if you look at IEEE touches upon all of those societies and it was really a rallying point for the organisation. And here we are looking to how we can build trust and confidence in the Internet, in the underpinnings of its technology through access, etc..

So, it's a global community and we, under this programme, have seven Working Groups that are working in these particular areas, from community networks to digital litracy, public access, new models and business models for how we finance connectivity so we don't, we can connect the unconnected. We have been ‑‑ strong partnership organisatgion. We work with bodies like ISOC is highly involved as well as world economic forum, people centred Internet and many others so this is really of a distributed community of a network of networks on how we are collectively working on these issues and we are providing a forum where these people can met and these solutions can be worked on and advanced in a more collaborative consensus‑driven platform. So that is what I have for you today. I would be pleased to talk to us, if you have any questions that you have for me, I know that was a lot in about ten minutes, but. That was just a high level view. Thank you.


CHAIR: Questions?

JIM REID: DNS guy from Scotland. Karen, IEEE a lot of its stuff is done behind closed doors because it's a membership based organisation. How do you deal with that from the point of view of multi‑stakeholder engagement? Does this create any tensions inside your organisation or does it potentially affect your ability to attract people from these other groups to come in and participate in the work of IEEE because of the fact it's perceived as being a member only closed club? .

Karen McCabe: That is a little bit of a misnomer because it's open, and we don't ‑‑ on a Member State, we don't have Member State membership. It's anyone from anywhere can join and participate.

JIM REID: How can I get access to IEEE documents? My understanding I can only get access if I have paid to back member.

Karen McCabe: I hear what you are saying. If you participate you can have access to those documents. It is sort of a interesting business rule that we have that we are working to change, in that really it has a lot of legacy associated with it from the sense of a lot of funding from the, to sustain the standards development programme was coming from the sale of those standards, so sort of a midway point is that if you are involved in any of the Working Groups you have open access to those documents that you are working on. But it is a challenge and I do thank you for bringing it up here because it is something that we are working on and that is why we have some other programmes from a business perspective, to see how we can do other funding mechanisms so that our goal is really to make all our standards available free to anybody.

JIM REID: Thanks. I wondered too actually if your experiences might actually complement the experiences we have here at RIPE and other Internet fora because in places like RIPE and ICANN and IETF anybody can pretty much sign up to the mailing list and they are done and that is it. But other organisations and institutions are uncomfortable in that kind of setting and framewok and they are much happier engaging in some kind of membershipbased type organisation. So it's quite possible that some stakeholder groups would be more comfortable or more, would feel more comfortable in IEEE setting than here and I wonder if you had any insights into that

KAREN McCABE: I mentioned the business model, but also from our process, so we have what we call super majority process in how we develop standards, what that means when you get to the point of the Working Groups where the standard needs to be balloted ‑‑ open to larger community; we have to have 75% of that community participate in that ballot and 75% of those participating have to say yes, so I think a lot of our process rules also kind of are playing into the issue that you are raising. But I can tell you we are taking a very hard look at all of these items and these issues and see how we might need to evolve our standardisation process and quite frankly, if you look at the standardisation environment as a whole, it is changing, there are many other issues that we need to address when we talk about standardisation and how that works with open source, you know, what is the relationship between those two things, how do we put mechanisms in place that make it easier for people to come and to engage in the work as a global community, so I think if you look at it from even standardisation arena perspective, there is some work that we have to do there but we are starting to address it.

JIM REID: Thanks. Just one last comment is that I noticed there seems to have been an element of ‑‑ among standard developments organisations and appeared to be some people doing forum shopping, but I don't want to talk about that just now but hopefully we can have a chat with that over some coffee outside this room.

AUDIENCE SPEAKER: Thank you for that. Chris Buckridge, external relations manager with RIPE NCC. IEEE is an organisation that RIPE NCC has been starting to work with a little bit more in the last few years and you and I obviously have done a lot of work in that OECD base where they are both part of the technical advisory group there. I have colleagues who have been involved in the IoT related events, Alex signature next to me here, that IEEE has put on and I think that has been useful so thank you for coming here today and it's good to see this coming more back into RIPE. I had really just a sort of quite high level observation and it's about your presentation but also a bit more about the conference this week. In your slides one word that I saw really come through repeatedly was trust, and that's also I think been a work that has come through in the discussion of interconnection that we had earlier in the week and in some of the ‑‑ a lot of IoT discussions we hear about trust and trusted devices, ability to trust, trusts of protocols and security. It's interesting to see that in a standards development situation, I mean, how does the IEEE view that idea of trust? Because I know RIPE NCC also has its own discussion internally and with the community about what level of trust we place in others or others place in us. I would be interested in your perspective.

Karen McCabe: Sure. You bring up a good point and that issue of rust is really sort of a driving force for the work that we are trying to get engaged here in what we call technology policy space. We are primarily known as technical organisation, we have thousands of technologists developing technology working to standardise that technology, but if the technology is not going to be trusted and the organisation is using the ‑‑ is not going to be trusted, the value and benefit that that can bring to humanity, our mission is about advancing technology for the benefit of humanity, will be lost, and so that issue of trust is really important to us. The issue of ethical considerations of technology is really important to us. So, at this point, you know, if I look at the IEEE's submission for the benefit of humanity, it really is a little overdue that we have taken a deeper dive into these issues. And we are looking at it from all aspects including the standardisation perspective which sometimes people look at, that is a bit of a disconnect, mostly markets, industry driven and we are very dedicated to raising the issues among other own constitute see, including those from industry, looking at how they need to be paying attention to those issues because I think that they are developing and they want to deploy from a business model perspective or business perspective is going to be a challenge if people don't trust it and people feel like their data is not being utilised appropriately, they don't know what is going on with their data so from an enterprise level to end user perspective it's something that is very important to us in taking it very seriously.


CHAIR: Quick last question.

AUDIENCE SPEAKER: This is NAT at that from RIPE NCC. I would like to see discussion in Middle East in this region. IEEE is working very hard in institutions here in the Middle East, we have like study groups for students but what about the industry? What about the technical professional communities? I mean, wave lot of members working from operators, telecom operators, mobile digital transformation, they are looking something to be working with third party which is something, does it mean they have to rely on a business aspect from the suppliers, from the vendors? So I want to know what is your, what is IEEE roadmap in the Middle East touching or tackling the technical professional industry? Thank you.

KAREN McCABE: Thank you. The IEEE is large and one of the foundations of the organisation is it's distributed community so we have hundreds and hundreds of sections and chapters around the world, including in the middle east to your point, many of those are sometimes more university academic focused that we are working side by side with our sections in ‑‑ around the world to address that issue. Every three years the IEEE has called sections congress which just happened a few months ago and I have been at the IEEE for many, many years and it's interesting to see that we now are seeing the section leadership kind of perk up and pay attention to the technology aspects of the organisation and the business as well as the industry aspects. So, I think that will be shifting, we to have dedicated strategic efforts and outreach efforts in various parts of the world. I have to say I think the Middle East is ‑‑ needs to be more robust, we have offices now in India and China, Singapore, etc., just opened an office in Europe, but I think the Middle East as well as south and Central America is definitely a little under‑represented and we need to address that and I know working with the board of directors of the IEEE which is the board that oversees the organisation they are building out some new regional strategies and I definitely will bring back your comments to them. Thank you.

CHAIR: Thank you, Karen.


In our ‑‑ it's now ICANN and Alain Durand.

ALAIN DURAND: Good morning. I work for the office of the CTO at ICANN so I am part of a research organisation, so I used to say we are part of good people the ICANN that you can actually talk to.

So one day my boss came to my office and asked me a simple question: How do we know that the system of unique identifiers that ICANN helps to coordinate is doing better this year or worse than it was last year? Okay. That is an interesting question. And she gave me a project, go find out. And it's going to take a while.

So, when you want to know if something is doing better this year than the year before, you need to have some kind of a baseline. So that is what we are trying to define here. Find this baseline and then see if we differ from the baseline. So, we went on this quest to figure out what does doing better mean. So that is ‑‑ this health came in and this was attached to this project as a stickie label and some controversy about the analogy with human health or not but at the end of the day, after a number of iteration of the project what we did was to look at problem areas, places where we know that we have issues. And tried to understand those issues, what do they mean and what does it mean that it's not going well. What are the characteristics of the situation when things are not doing well so we can understand this better. From this, we try to derive metrics, try to say okay, can we measure this. And when you have metrics then you can get some measurements, meaning that you can go and find some data and then apply this data to metrics and you can get numbers. So, two observations, the first one is, when we have those numbers, if you look at them and you have a number the 740, well, 740, right? It's a number. But if last year it was 800 and you know if the number goes down you are doing better than 740, certainly a better number, how much better? I don't know, that is for future generation. My kids maybe can look at this or somebody else can look at this but I know it's better. If it were doing to 850 yeah, that will be worse. So that is the first observation.

The second observation is, because of the this approach that we took, this is a project that is going to run for a long time. So this is not we are going to measure something and then oops, we are done. First it's going to take us a while to define metrics because it takes us a while to define problem areas and we have to find the data, that may or may not be there or that we may have to acquire or we have to find some cooperation with other people to get. The other approach that some people have taken is, to say we have data, let's go and analyse this data to see if we can find some trends. It was our approach to not go there, and to do the opposite.

So, the other thing I wanted to emphasise is, the scope of our work is to present the data to the community, it is not to take action on it, so ICANN the organisation is not going to look at this data and impose things on the community. The data will be there for ICANN, the community, to look at this, and part of the community that has a role in policy making, hopefully will look at this data, if they find it interesting can be using this data in order to make better educated policy. So that is a distinction that is really, really important. We are not going as an organisation to use this data to take any action.

So ICANN three domains, names, numbers and protocol parameters. The first one I say to a word about is the one on the numbers side because we are here in RIPE meeting.

So, this is a path that is completely driven by the RIRs and the registry services have been working together for quite a while now and they have defined some metrics, and they are starting a global consultation process to get input from the community on those metrics and there will be more information about this that will be provided tomorrow. So I am not going to say much about this except that if you are interested go talk to them, go participate to this global consultation, they have done great work here and it's really important to get this feedback from the community.

So, on the names side which is more what we do, I mentioned that we define some problem areas so we started with 5, why 5? Because it's relatively small number that is somehow manageable. We may add some later, some may be removed, but essentially this is in the domain of data accuracy or inaccuracy, abuse, we had some issue in the overheads of DNS with traffic, leakage of names in the DNS, name resolver misbehaviour and things like that. So we had a present a couple of years ago from Daniel Karrenberg who was talking about names going to the root a little bit too often so this is the type of things we would like to measure, essentially if we can have some feedback to say, yeah, this is doing better because you are doing X, Y and Z, that will be interesting use of the data. So the process that we are using is once we will have defined over metrics and find some data for measurements we will have some sources of data. That may or may not be public. If we used this source we will have to extract data, abstract data from it to digest it and present it in a way that is being offered to the community. So we will have to remove any kind of private information in there, if it is a commercial data we may have to abstract it in a way that is essentially free of royalties or free ‑‑ that is available to the public. We will do a minimum analysis on this in terms of, okay, we may do some graphs to say there is a trend here. But that is essentially where we are going to stop. And when we will publish all this, being this pre process data and those graphs through the open data initiative at ICANN which is another project aimed at publishing everything that we have to the community.

So we have a number of metrics, I am not going doing into all the details but they are in the slides and if you are interested please come talk to me. First one is about data accuracy or inaccuracy, that is really the thing that we care about. There is a difference in ICANN on what we do and what IRs are doing, the IRs are owner of the database so you guys know who an address belongs to, that is your job. In ICANN we don't, the registrar has this information. Sometimes the registry. We don't have it. So the only way we could actually do something is indirect measurement by looking at the number of complaints that comes to the ICANN compliance department that are validated meaning someone is making actionable and that would give us an idea of how much inaccuracy there is in the system. That is not perfect, we know that, but it's better than nothing.

There is another project, third one in ICANN called D AA R which is domain abuse activity reporting. We gathered a bunch of anti‑abuse mailing list things like Spamhaus and others and tried to look at every top level domain or generic, because we have contracts with them and we know some of the database. And tried to see what concentration of spam, phishing, malware and Botnet, and a analysis we have done is showing that for different types of abuse are going in different places. So, for example, spam tend doing into the classic top level domain, phishing tends doing there too but malware and Botnet go in different places. It's actually quite interesting.

The third one is about overhead of traffic in the root so Daniel Karrenberg talked a bit about this. We realised essentially there is only a couple of percent of traffic to the root that is useful. Everything else is either duplicate or no such domain. That is okay, the root servers are designed it handle this traffic but it would be interesting to see if these numbers are going up or down because it will an indication to see if system has to be resized. Maybe shrunk or maybe grown. Leakage, well you all know about things like cop hop and mail, was not delegated by ICANN, also names that are in the IETF registry for special use names and that sometimes also happens in queries into the root and having an idea of how much leakage is certainly something we would like to track. We saw, for example, some randomly generated name, things like 8 character long but always different. And obviously generated by computers and we have been able to trace this back to some software. We are trying to figure out if they are behind a firewall or they are connected to the Internet or not. So tried to see can we talk to the root and have authoritative answer from the root and if yes, we are connected to the Internet; if not, we are not. That is some interesting behaviour, to say the least.

Sometimes people don't always tell the truth, we are talking about trust earlier. And that is what DNSSEC was supposed to address, making sure that when you ask a question the answer you get is the true answer, it does ‑‑ it has not been mangled with by somebody. So, we have a number of anecdotal evidence to show that sometimes DNS resolver did not tell the truth, they changed the answer. So you ask for this website and you are redirected to that website because it's a competitor. We have heard a number of cases about that. We also have heard about cases where you want to send your DNS query to some serve but a service provider is redirecting that query to their own server where it will be their own interpretation of the truth. So, anecdotal evidence is an interesting thing but having numbers that actually measure this is more interesting. So we can say yeah, this is a problem or no, this is not a problem or it was a problem two years ago, but now it's not any more.

Protocol parameters and we scoped this to what ICANN does meaning DNS‑related registries, and this is about trying to figure out what DNS protocol parameters are being used. Are people using parameters that are not in registries or are some parameters not being used and essentially a proxy for us to see if a specific technology is being adopted or not. For example, TLSA, for the DANE Working Group at IETF, in some of the early experiments we have done we don't see much use, if at all, of these parameters. So maybe it shows it's not yet adopted and it will be an early signal for us to see if it will be adopted or not. But also maybe and that is why I have a call of action for people participating to RIPE community, is we are looking in the wrong place because we have been looking at data at the root, that is what we have access to, but maybe those queries don't go to the root, they only stay to recursive resolvers so we have now tried to reach out to DNS recursive resolver operators to ask them to work with us and help us try to run some of the programme that we have that extracts some of this data so we can have a better idea if those protocol parameters are being used or not. But programme will remove all the personal identification data and essentially just aggregate all of this and give us a table for those parameters, this is how much being used. So if you run ‑‑ run one of those DNS servers and would like to participate, please come and talk to me.

And that is it. So if there is one thing that I wanted you to remember, is, we have a good people from the ICANN (we are), but more seriously, this is a project to try to figure out if we are doing overall a better job year after year. If the policy that we are doing are having an effect or not. We are not going to use this result to have any direct action, we want to give this to the people who are developing policies to, as an input to them in their processes. If there is any question now?

CHAIR: Thank you. Questions?

MARCO HOGEWONING: RIPE NCC. You need a baseline at some point. How much historic data do you have or are you collecting data to establish that baseline on a long‑term effort? Or are you basically just going to say today is the day, today is 100 and we will go from there?

ALAIN DURAND: We are going to do the latter. We are going to say this is 2017, and we would like to do this maybe month by month or week by week, it depends on the data if we have enough access to the data or not and this is day one of the project, and day two it will be at a different point.

AUDIENCE SPEAKER: Alex, RIPE NCC. And concerning the protocol parameters, did you consider the opportunity to cooperate somehow with passive DNS database operators because probably they have some bit of information you are looking for?

ALAIN DURAND: That is a great idea, if you can connect me with them I will really appreciate it.

PETER KOCH: DENIC. I have two questions, one to actually what ‑‑ the slide that is on the screen at the moment. So when you said restricting this to DNS, which is what ICANN is concerned with, and then comparing that with the goals in how well the community, the organisation, it, working and you mentioned TLSA but as far as I understand, ICANN has no role in deployment of it. LSA or promoting it or anything like that, so I wonder what that means? However, on the other hand, you left out the provisioning side, the whole EPP suit of protocols, including the interesting diversity in special features that we know is there, can you elaborate on that?

ALAIN DURAND: It's very simple.
Answer: We are going for the low hanging fruits and this is something that we think is relatively easy to do. Now, this is part of DNS operation and as such it falls into the, it's easier to argue it falls into the umbrella of what ICANN is looking at than, for example, parameters foretell net or FTP. Now, to your second point about EPP I mentioned earlier we have five problem areas that we have decided to focus on. If you would like to create another one on EPP, for example, registration, we would be more than welcome to work together and develop this and it may come in the second phase of the project that we will have this too. It's absolutely open to committee input to define some newer area of focus.

PETER KOCH: I have my hands deep in my pockets like every good lab engineering, not to touch things I shouldn't touch. Regarding this DNS abuse and after all these years I haven't understood what that is. I have a bit of a concern that this is mentioned under technology health indicators. It is a bit like the car industry running around and finding out, since we are going to be in Marseilles next time, I remember these black and white movies of whatever the '50s or 60s where all these nice Citroen cars, where the criminals run through the streets and do some research that the criminals are the one driving these kind of cars, reading tea leaves or getting into the area of chem trails and so on, it might be interesting from an academic viewpoint but this isn't health indicators, so a bit more truth in advertising would be appreciated.

ALAIN DURAND: That is where the ‑‑ is a bit problematic at the beginning of my talk. It depends on who you are talking to. We have seen in the early measurements of this abuse things that there seems to be surprisingly high concentration in some areas, and some of those areas could be either some TLDs or some registrars and this is of concern, at least to number of people. So, that is why tracking seems important.

PETER KOCH: True. But again, I mean, it's the car industry, are they responsible for some community's preference for certain type of car to escape the police or something like that? Or to put it in a more serious abstraction level here: In the governance field, several scholars differentiate between micro level and macro level, so the concerns are actions that are happening several direction layers away from the actual identifiers as in content in a broader sense, whereas what we are concerned with, both in the names community but also like here in the numbers community is the identifier itself. And mixing or not and not appreciating the boundary between these two systems is very dangerous. And that is why I said.

ALAIN DURAND: If you look, for example, in the case of Botnets, the identifier is the one thing that triggers a Botnet and DGA ‑‑ digital generic names, so ‑‑ between the identifier and the type of abuse.

PETER KOCH: There is always a link because the identifier ‑‑

ALAIN DURAND: A direct link.

PETER KOCH: Yeah, I guess we need to agree to disagree there.

PATRIK FALSTROM: So Netnod. So I need to bring up GDPR. Question: How much do ‑‑ have you investigated yet how much GDPR regulation in Europe, in EU, is affecting you? And more specifically, do you have a ready data processor agreement or are you working on one that you can share with us to help the industry?

ALAIN DURAND: So, I need to go back to this diagram here. The whole idea is to remove any type of personal identification data from the raw source that we have.

PATRIK FALSTROM: Are you aware of what data processor agreement means? Otherwise I am happy to talk to someone else, that is not a problem.

ALAIN DURAND: My understanding of this is that we have to be able to state what we are doing with the data.

PATRIK FALSTROM: How you handle it.

ALAIN DURAND: Store it, handle it, etc., etc., yes, I am aware of that.


ALAIN DURAND: What I am saying, for example, in the, let me take a very specific example. There is script that we are talking about ‑‑ working with DNS recursive operators. Going to make sure that there is no such private data that would be exported to ICANN, that will stay at the data source, this abstraction will be done there. So in anything that we would get at ICANN we will not have any of those PI I, first we don't have to worry about keeping PI I or having a policy to retain PI I because we would not get it in the first place because the script will run in the service provider network and will only send to us the abstracted data.

PATRIK FALSTROM: So what you are saying is that even though you provide that script and you have an agreement with whoever is going to collect the data on your behalf, your interpretation is that you are not a data processor?

ALAIN DURAND: That will be my interpretation at this point in time.

PATRIK FALSTROM: Thank you. Because these kind of things actually helps the rest of us that also are sort of dealing with the same kind of headaches. Thank you very much.

CHAIR: I was going to make an exception for Hans Petter but I can do it for you too Jim.

JIM REID: DNS guy and exception to the rule. Alain, just to follow on from what Patrik was say, I think someone in ICANN used to do this but more homework on this, you are saying you are hoping to strip out personal identifying information, that can means things like DNS query names, almost certainly IP source addresses. So I think you need to be very careful about that and remember, too, that what is considered PI I in one EU jurisdiction might not be considered PI I in another. And probably the long‑term solution for that is going to be for ICANN to enter into safe harbour arrangement so it's able to process personal identifying information and store it, albeit under some kind of terms and conditions which effectively make you appear to be an EU organisation. All I am saying, more homework is needed here. I don't think it's going to be quite as simple as you seem to be suggesting it.

ALAIN DURAND: I agree with you, it will depend on which source of data we are talking about. In the example that is most advanced, this is work in progress, let me remind you, we have not defined all the data sources, but number one of the example that is most advanced in looking at protocol parameters and DNS collected from recursive resolvers, the only thing that is extracted will be a frequency of a protocol parameters, there will be no DNS name, no IP addresses, nothing like that. So I think in this particular case we are safe. We may or may not be that safe with all the other collections and I think I take your point that every time we have a data collection we think to think about the PI I and GDPR.

CHAIR: I do note the number of ICANN board members in the room waking up when we had this question here. So thank you.

Then the last presentation is actually not going to be about I‑STAR ‑‑


One thing that governments worry a lot about these days is being green and they do look at our industry, so that is definitely something we do need to worry about. So Mike.

MIKE HAZAS: Thank you for coming. I appreciate the sizable audience in the first session. I just wanted to say that this is all kind of, this is a necessarily partial view out of the evidence we have been able to cobble together over the past few years and I welcome any insights. We are bound to not have the view that a lot of the people do on some aspects of the system. So definitely welcome comments and the reason I am secure here to discuss and hopefully collaborate.

So I should say that this work that I am about to present is done in conjunction with a set of colleagues and in particular Dr. January even morally who is at the demand centre, which is a centre based in Lancaster which looks into all manner of aspects of energy demand and there is a whole other research team which has a lot of input into the thoughts and analysis and so definite credit to them. I tried to credit them where possible on the slides but just a general credit to the group.

My own background, going back to undergraduate days is electrical engineering, I did PhD in mobile sensing, embedded sensing and then I went on and did sensor networks and more sensing in my post doc. And then more recently have done under‑graduate degree in sociology, which pulled me over to sustainability and digital systems.

I am going to talk about the growth of the Internet, not something I will spend a lot of time on because I think a a lot of people here have a haired appreciation for that and what links there are to electricity demand in the global sense.

And then I will spend a quite a lot of time talking about what people do with the Internet which is what it comes down to really, the investment of resources and what is that for and a lit about the policy implications. SPO in the UK anyway, if you look at the monthly kind of per household data volumes they have been steadily increasing, so we are approaching 150 gigabytes a month per house per month, which won't surprise many of you and likewise many of you will be quite familiar with the growth across the Internet Exchanges, the Amsterdam exchange being one but we have seen this kind of trend, upward trend. Obviously that is a thing, right?

And certainly mobile data because of the, I mean it's difficult to deploy maths but the number of subscribers is going up, the amount each use ‑‑ draws on the system is going up so we are looking at around 50% annual growth and that is forecast for the next four to five years and likely more, I would expect.

So again, this should be familiar to the audience, so I am kind of skipping through these pretty quickly.

You do see things in the media from time to time, so like sort of the relationship between data centres and emissions and how if we keep going like this we are going to lose all our power and all the lights will go off, it's a favourite headline. But really what we want to know is how does it ‑‑ what is the relationship. And there is a whole body of literature known as life sick he will assessment or analysis and these are people, scholars, actually, who spend a lot of time worrying about the total emissions of societal systems, something like agriculture or distribution or whatever sector. And some of the things they have looked at have to do with IT and its systems and they say that IT, the Internet, end user devices all together probably consume about 10% of global electricity currently. It varies a little bit but that is around where we are. And forecasts out to 2030 say it might be around 20% if we keep going like this. And of course the assumptions vary, right so, depending on your assumptions, you might get half of global energy by 2030 or actually it might decrease going on from here. So of course forecasts are forecasts but I guess the indicators are this could be a problem and we should at least watch the trends rather than just let things develop organically. Is the argument.

So I suppose that connects to larger debates in society about resource allocation and energy demand and cash on emissions, we have discussions about transport, healthcare and about heating and cooling in the form of building regulations and low cash on initiatives so why not about online services rather than just sort of the odd newspaper article and the discussion about well, how do we deploy faster broadband and that kind of thing, all very important discussions but maybe there needs to be a broader debate so that is motivation for this work.

So that brings it to what the Internet is actually for, because that relates to the debate about how we deploy our resources and how we justify the use of energy and the carbon emissions associated with that deployment.

So we can look at sort of industry reports, for example the Sandine reports which some of you may be familiar with, which show that streaming, I guess that would be audio and video‑streaming, realtime entertainment, that sort of thing, comprise nearly 50%, fixed traffic and an increasing share of mobile traffic as well with social alongside web browsing for mobile and still significant onyx iffed. So obviously, there that is kind of interesting, but that doesn't really tell us what the information is for, it gives us an idea about how the traffic ‑‑ again these are quite broad categories and we are limited to the top 5 and so as things drop in and out we kind of lose track of them but they are quite useful as a starting point.

So the, you will see the little icons in the following slides, that refers to the fact we have done three studies over the past years and that tells us, which, iOS devices and a another study of Android devices and these are combined both quantitative measurements of the traffic and car gorisation as well as qualitative interviews to understand the meanings and why these things are important or perhaps not important to participants and there was a larger slide of Android data set, nearly 400 tablets and phones looked at. I won't talk about these any more but that is what the little symbols are in the slides in case you are interested.

So of course you get all kinds of stuff, the Internet is actually used for everything and everyone I'm sure will agree with that, obviously. So it's used for work and variety of tasks, online banking and the list just goes on and on and this is from our own categorisation just from the small study and the larger set of phones.

But actually what holds the largest share? Well, similar to the Sandine reports where they had a big realtime entertainment we can break that down, 21% of the mobile and tablet traffic and again this includes wi‑fi and mobile data, was actually used for streaming video and around 11% was for audio. So that is kind of vaguely corresponds to the practices of watching and listening to things which we do. Social networking is quite important, particularly on mobiles. There was kind of a 10%, thank no matter how we dug, Samsung dot whatever we couldn't attribute to any particular application or purpose or anything like that so that is around 10% of the traffic on those 400 phones.

And I would like to, you can look at those four categories over average over the day and obviously one thing I'd like to observe here is that and again this should be a fairly common sense I think that most people can relate to, the video streaming has changed the nature of the practice of watching. So of course we still watch things probably on large television, maybe that is broadcast and streamed but actually what has happened is, watching continues on mobile devices and is perhaps done at the same time, we can do two things at once, oddly and take those devices into the bedroom or another room and the watching continues past midnight quite significantly.

The second thing about this graph that is striking is that the peak roughly corresponds, you can see the shaded area, is the national, the grid peak of the UK. And so the height of activity particularly with watching but to an extent with social networking actually corresponds to national grid peaks. And the national grid peaks are when we are burning the most coal to keep the grid running, so those carbon intensive times correspond to high times on the Internet.

The Internet is used for just filling time, waiting for the bus, out and about, got a few spare minutes I will check it because I am bothered, that is actually what it is. And notifications seem to play an increasing role, this happened across both studies of checking apps and how often we check and when we decide we are bored or not. And sometimes far the sake of it and that is relying on Internet as a service in various ways. A notification might lead to a tweet to another, might lead to a web page to looking at a video and all that can happen because of a notification.

There is also an importance ascribed to background noise so that was kind of letting a video or some audio play in the background for the sake of it, maybe living alone, maybe just kind of wanting to listen casually while doing something else. So again, fairly common things. That happened especially with the phones, we found that people were kind of using them a little like radios on the commute and which is why it comprises 11% of the overall traffic.

We also find, perhaps rather encouraging from an energy standpoint, people willingly took breaks from technology, wanted to shutdown before it got too late and wanted to take break at weekends or when they go on holiday, don't connect, that kind of thing. There is some possibility for working there. And likewise, people would adapt, so they didn't, even though the apps tend to expect stuff to be on all the time and moving quickly and the connection works because otherwise stuff breaks down people were quite adaptable and think ahead to when they wouldn't have structure so people can adapt and that is quite encouraging so trying to be positive here.

But before I go on to you be talk about what we might do about the implications of some of this, and I am completely open to new ideas and discussions, I would like to point out automated traffic.

So earlier I talked about the operating system or the unattributable traffic was around 10%. I would add that update and back‑ups themselves, these things which keep the device running or backed up or whatever, might even be owned by default run in the (on) 15% on top, so you can kind of look at that as an overhead depending on how important those background tasks are, which were very difficult for us to tell despite the depth of our studies.

Some basic technical implicationd, if the screen is off, switch off the network. Very simple. We could try different ways of providing background noise which is lower bandwidth which might not be the full on thing. Or since we do things like include NFC and a host of other technologies, why not a digital audio tuner or FM tuner, that is not a huge cost and compared to other technologies so that could be a way forward. We do some work already collectively as a technical community about matching video bath with device capability. I also think that rather an app update every week, smaller more strategic ones we are perhaps better thought through. We can also of course shift the peak demands because updates not just streaming but also updates and back‑ups happened kind of closer to the peak than they should, 3 to 5 a.m. would be better from energy supply standpoint and think about apps that lower their reliance on Cloud infrastructures rather than raising them or assuming they are always going to be there.

There are of course a host of broader social implications of where society is going digitally and what we are doing with it. Rather than looking at filling time we might emphasise that time as maybe more traditional ways of time to relax and reflect, time to read something or something like that rather than relying on online services. That would take a broader effort, I guess. That would be part of raising awareness and encouraging reflection on how we spend our time digitally and what that means. We could even think about impact ratings for object line services or particular individual yes, sir so start to raise consciousness about what happens when you fire up a 4 K 3D video or what have you and what is involved in that. And as well we can think about engaging people to kind of work, to kind of get together more, to watch stuff, maybe it more of a spiritual ritual rather than binge watching that thing by yourself in bed. There are lots of things we could. In five years it would be nice if one of these bowls at RIPE had sustainability on it, we could put that on our badge and, people that might be concerned with these issues could get together and proclaim that interest a bit more. One of my ideal scenarios. But to conclude, I will just leave it with a quote from Chris de Burgh who posited that there is no limits to growth when it comes to the Internet, unlike roads and all kinds of societal infrastructures, the Internet can actually grow because of its fish see, because we are so good at fibre and small devices it's only limited bittener gee supply itself, it will only stop growing when energy sources run out or we impose some sort of limits and we responded to that in an article last year so if you wanted to read more about that, that is there. So I will stop and take questions. Thank you for listening.


AUDIENCE SPEAKER: Alexander from Russian Internet Protection Society. Always wanted to ask this question: Did you measure how much energy was spent during your presentation just now, for electricity, for light, for radio waves and radio mic, how you concerns to spend energy on this presentation, did you measure carbon footprint of your arrival here?

MIKE HAZAS: I didn't here. I gave an earlier version of this talk, actually, it was last year, and tend of the talk I stayed in Lancaster, I didn't fly to where the talk was but that was over 21 hours just to stream the talk. So I am definitely aware of the energy it takes to participate in these discussions about energy, absolutely, no, no, always aware of that. But no ‑‑ sure of course ‑‑

AUDIENCE SPEAKER: Numbers next time, otherwise your talk is completely spoilt. You can't measure your energy and you are persuading us to conserve energy on other things.

MIKE HAZAS: Point taken.

JIM REID: DNS guy one of the usual suspects. Interesting talk. Mike, in the UK you may be aware the national grids have in their operations centre TV screens are watching television programmes that are about to go off air because whenever one of these programmes finishes half the nation goes on or switches the kettle to make some tea or goes to the bathroom and have to try and match demand and a few minutes before hydro power stations we are going to get a spike in energy usage quite soon. You were talking about patterns of Internet traffic usage, if there is anybody thinking about joining these two things up, the content delivery networks or anybody else like Netflix or somebody and having some kind of information being fed back into national grid centres so that the national grid can get some insights about likely energy use from these activities. Has anybody given any thought that to that do you know?

MIKE HAZAS: We have talked informally and they are kind of scratching their heads, on the one side it's good you have not got as as much pronounced spike just before East Enders, for example.

But on the other hand, you have got an overall rise in the way watching happens and people are still watching at the same sort of time, it's just not as predictably synchronised. So I think maybe that sort of data, if they get warnings from YouTube, Netflix and other services that stuff is happening maybe they can fire up.

JIM REID: There is a new release of VI OS coming out, those things would be very useful.

MIKE HAZAS: And when Facebook turned on auto play, all significant events.

BRIAN NISBET: HEAnet. I don't believe a lack of numbers invalidates anything. You are saying about your travel here or otherwise, if we were actually going to be serious about this, we wouldn't have those stickers because none of us would be coming to these meetings. But that is a secondary point.

The one thing I see here and I see this in a lot of the advertising about recycling and things like that, is that people are saying, well, individual users should change their behaviour to fix a thing. I suppose I have a lot of doubt that, A, the massive humanity will ever change their behaviour in large enough numbers to make a difference. And B, that the power usage is being in no small part is being done by the Netflixes and Amazons, etc., of the world.

So while this is extremely interesting, the studies and the quotes and all the rest from were from individual users, whereas, and you have more numbers than I do so I am not arguing with the numbers here, whereas the usage primarily on the side of the content providers, aye realise they are interlinked.


BRIAN NISBET: But if you turn around and say, okay the screen is off my network is off, and this super computer becomes less useful for me, and I am not quite sure how much power it saves on aggregate versus looking at more, the serve technology, the production, etc., on behalf of the huge companies providing this content, if that makes sense.

MIKE HAZAS: Yes, that makes sense. There is a lot packed into that question. I will try to just comment on a few of the things. I suppose the, I mean, I absolutely agree with you, I think broader change comes through change of infrastructures, and of course people are going to do what they are able and perhaps they absolutely should in some way but I guess the flip‑side is that without raising that consciousness like recycling in the '80s and stuff, there isn't impetus across society to make change which might affect the big infrastructure providers. The reason I had some of those smaller implications is because sometimes people say, what can we do, we need to do something, what can I do? I put those in just for satisfaction, but I do think there is a role for policy to play and governments to negotiate with the large service providers to have a fruitful discussion about the direction of the Internet. Great question.

MARCO HOGEWONING: I am feeling really old and I agree with you there is a lot to win but let's not forget where we come from. The device I hold in my hand probably takes two hours on 25 watt charger. I come from a day where this device was this big and took about 500 watts continuously to do only half of what we did. So let's not forget where we are coming from.

MIKE HAZAS: Absolutely. With a rise in capacity and efficiency, comes a rise in demand and that is the problem we have got, instead of hundreds of thousands of devices, we have billions of them and instead of low quality streams we have very high quality streams. I appreciate what you are saying, thanks.

CHRIS BUCKRIDGE: RIPE NCC. So like Marco, I saw a version of this version of this at Euro DIG earlier this year, I am glad you have been able to bring it here and start discussion here.

Looking at the slide you had on the things people can do and the ‑‑ the stickers or the things that could show what the energy usage of watching a video would be, I think raises some interesting questions and particularly for a RIPE network operator kind of thing because we talk about old CDNs being used much more so what is the actual, saying what the energy usage of a given video would be seems like a very difficult measurement to me, it could be streaming from the other side of the world or from your, the CDN in your local service provider or anything like that. I guess sort of bouncing off what Brian was saying and I would be interested to know does your experience in talking to some of these big operators how transparent they are about that kind ‑‑ those kinds of things, about what the trends in energy usage are from say different kinds of network deployments that they do or different developments that they have. Thank you for the talk.

MIKE HAZAS: That is a great question. Yes, I guess one of the problems about gathering the evidence to make reasonable analysis of this sort of thing is that a lot of it is hidden, it's in black boxes and data centres and hidden behind corporate protected basically. So the trick is that worked in the past is to find a quite sympathetic sort of controller of media, regarding news media has collaborated with academics in the UK and that's sort of ‑‑ so, analysis we have done has relied on that. So the quick rule of thumb is basically the energy it takes to watch a two‑hour film on LED TV is equivalent to the energy it takes to stream it, roughly, currently in the UK, your mileage may vary. Double for a large TV and times ten for a small device. Rules of thumb. The trick is to find organisations that are sympathetic until there is some broader pressure for companies to do their own analysis.

JIM REID: I agree with one of the points about needing a multidisciplinary approach and I have to disagree with one of or two of the comments Brian made a moment ago. He thinks people will not change their behaviour. If you apply behaviour economic tricks to this kind of scenario can have a positive effect from minimal effect and some years ago the Irish Government introduced a charge for plastic bags in shops dropped by a factor of I think about 99%. So small changes or small incentives can make people change their behaviour and we need to figure out how to apply those kind of centres to make the necessary changes.

MIKE HAZAS: That is true. That is one of the ideal outcomes from having environmental ratings of services or particular videos. It might shift things a little bit which would shave off a couple of percent, which is actually quite big. The other nice thing about qualitative data is that it tells us what these services mean in people's lives so we actually found out although it's great for filling time, what people really value is being able to message friends in a group with notification that has been delivered and to be able to send small photos. That was what we found to be meaningful across a small sample 630 participants in total.

AUDIENCE SPEAKER: Your approach looks for me very questionable because you mixed your speech and your presentation electro energy and energy because for me looks might be particularical temporary problem distribution but in 10, 15 years the key consumption electrical will be car again and see the whole picture of energy consumption. It's absolutely differentment and all our Internet is nothing.

MIKE HAZAS: That is true.

AUDIENCE SPEAKER: It's some disbalance.

MIKE HAZAS: Of course.

AUDIENCE SPEAKER: It's not strategical problem at all still using for hidden and for transport (heating) and for industry. Large scale it's nothing.

MIKE HAZAS: So I guess you are saying if we look at the total of energy ‑‑

AUDIENCE SPEAKER: No matter which kind of energy.

MIKE HAZAS: That's right. Once you ‑‑

AUDIENCE SPEAKER: There is source of energy.

MIKE HAZAS: I take your point.

AUDIENCE SPEAKER: Interesting presentation. Malcolm Hutty from LINX. I think it's important to remember the difference between energy usage and energy efficiency. I mean, we have all got one of these things around in our pockets and you may question whether all of it is valuable and the guy from low tech may say you know what, you could go do without this it wouldn't be that bad. If I thought that I wouldn't have it in the first place. I respectfully disagree with Jim's behavioural science, try and persuade people to do what they want to do. Our job should be to help people what they want to do and we should be aiming not to change their behaviour so they use less energy; it should be to reduce waste so they can do what they want to do at lower cost and that should be the goal. Thank you.

MIKE HAZAS: I guess I think that the way the market works, it is very, very good at efficiency, and I think we will keep getting more efficient, but history has shown across sectors, that efficiency does not keep up with demand and it's that growth that I was hoping to discuss and the merits of that. I agree with that, of course, people will do want what they want, I agree with that, and tackling it from all sides.

MALCOLM HUTTY: There is a lot of energy in the universe and unless we are having brown‑outs, lack of energy isn't the problem. The problem is ‑‑ I don't actually agree with that it should be a goal to prioritise reducing energy for its own sake over the benefits people get from the consumption of that energy and turning it into useful work.

MIKE HAZAS: It would be fantastic to know more about the benefits and how that is turned into useful work. I would love to ‑‑

CHAIR: One quick comment.

AUDIENCE SPEAKER: For the future version presentation it would be really great to see also consumption of, by new technology, crypto currency, money. To understand, you know, what is going on here and here.

MIKE HAZAS: Yes. I would love that too. Thank you for the suggestion. That would be fantastic.

CHAIR: Thank you.


MIKE HAZAS: Great questions.

CHAIR: I had originally planned to have a small discussion here about where we are going doing with this Working Group, but as my co‑chair couldn't actually be here, I don't think it's actually fair to have that discussion without him. I suggest we carry on that discussion over e‑mail list. Let us all go and get some coffee.

Thank you.