IGF 2018 - Day 2 - Salle III - WS275 Before You Know It, Internet Governance will be Irrelevant

The following are the outputs of the real-time captioning taken during the Thirteenth Annual Meeting of the Internet Governance Forum (IGF) in Paris, France, from 12 to 14 November 2018. Although it is largely accurate, in some cases it may be incomplete or inaccurate due to inaudible passages or transcription errors. It is posted as an aid to understanding the proceedings at the event, but should not be treated as an authoritative record. 



>> FARZANEH BADII:  Okay.  Can you hear me?  Hi, everyone.  Welcome to -- maybe we can wait a little bit until they close the door.

     Okay.  So welcome to our session on Before You Know It, Internet Governance Will Be Irrelevant.  My name is Farzaneh Badii from the Internet Governance project at Georgia Tech.  The title of this session is provocative and maybe this is why you are here.  This is not a statement, more of an observation or question, whether it is becoming irrelevant because of this technical development that we are seeing on the Internet.

     So I am not -- this is going to be a very interactive session.  If you are very opinionated about this issue, that's very good.

     What we are going to do, you can interrupt any speaker you want and if you think that you disagree with them, you can disagree with them.  And we allow that.  There is no interaction after speakers speak.  You can speak whenever you want but, of course, general principle of not interrupting is good.

     So to lay out the problem and the issue, I'm going to go to Christian Kaufmann here and he can lay out the issue and also Geoff is online.  Geoff Huston and Christian are going to lay out the issue and tell us what the problem is or if it is a problem.  And then we can talk about how we can resolve it or what sort of Internet Governance we evolution we are seeing here.  Without further ado, go ahead, Christian.

     >> CHRISTIAN KAUFMANN:  Thanks.  So my name is Christian Kaufmann.  I guess I'm qualified or classified in this group as technical community.

     I have multiple hats, but two hats are probably relevant for today.  One part, I work for a CDN called AKAMAI in the day job and involved in different Internet exchanges and on the board of one of them is here in Paris.

     When we talk about the panel today or what brought up the panel today is a trend we have seen, I guess, in the technical community, which is the consolidation of players, especially when it comes to the content.

     Before the ten years ago you had a lot of content players and the content was very distributed from a players in the ecosystem perspective.  That was true especially with social media and the big companies changed quite a bit.  I guess right now a typical Ibon social network, the numbers might change, you have roughly ten players, 70, 80, 90 percent of your traffic.  So the content and the traffic you are getting coming from very few players.  This is one data point for today's conversation.

     The other one is we used in the past, I guess as Internet operators, two or three fundamental ways how we delivered the traffic.  One is classical transit.  You have a transit supplier, you buy transit from that supplier.  You have peering.  This developed direct PNIs or Internet exchange.  The last specific for CDN, you could have a cluster in your network from the particular provider and you got the traffic from there.

     I think what we have seen in the recent years and especially in the last one or two years is that the whole traffic which is flowing via transit and even via the Internet exchanges is actually not growing to the same degree as the rest of the ecosystem.  So the Internet still grows.  You have broadband penetration, you have more content, but the players or delivery method as in transit or Internet exchanges is not growing to the same degree.  There are actually some of them are flat lining.  If you look at the very big ones, you see that the last year is moderate or flat in comparison to what we have seen before.

     So a lot of traffic these days either goes from content owner, whoever that is, straight into a cloud via cloud on ramp.  Then it was in one of the three, four, five big cloud providers and they deliver it by their own CDN or give it to a third-party CDN and that one delivers it.

     So the traffic flow instead of going a lot through the Internet, the public Internet or the Internet exchanges is now bouncing between the big enterprise networks before it goes straight into the Ibon network.  If that continues, the trend the way we have it right now, do I bring up the questions or is that your job?


     >> FARZANEH BADII:  It's me.  But this traffic consolidation actually might have Internet Governance implications but before we discuss those implications we need to understand what Internet Governance is and what the definition is because it is very important also in this IGF because a lot of the issues being discussed in this meeting are really not global Internet Governance issues.  What we mean by Internet Governance are the policies, the best practices, and the standards that shape cyberspace.

     And the traffic consolidation might affect any of these aspects of the Internet Governance.

     So Ted, you want to add on to the definition?

     >> TED EGAN: I'm from the Internet Architecture Board.  And there is a set of tasks required to keep the Internet open, interoperable and globally connected.  Those tasks are not associated with government entities.  The governance may be in part due to the action of standards bodies, making sure that there are interoperable protocols or the market forces keeping the voluntary nature of it as part of their interaction with the market.  So given that set of tasks, I think the question that is really being asked isn't so much is the Internet Governance, is that set of tasks going to go away?  But is it going to change?  I think that the implication of the broad posting that Geoff put out and Christian elaborated a little bit, there are traffic patterns which are changing and do they involve or imply evolutions in the set of tasks that we need to take on to keep the Internet open, interoperable and globally connected?

     >> FARZANEH BADII:  Thank you, Ted.  I see Geoff online.  If you are ready?  Okay.

     You want to --

     >> GEOFF HUSTON:  Can you hear me? 

      >> FARZANEH BADII:  Yes, we can here you.

     >> GEOFF HUSTON:  What I am doing in today's Internet totally unique and I might be the only one in the country doing it right now because I am sending packets from Australia to France.  Now, you know, five years ago that would be commonplace.  The Internet is there to ship packets all around the world but that doesn't happen anymore.

     The Internet not only rules changed in the last five years, has changed dramatically.  Indeed it is hard to really understand what the Internet is anymore.  We built this network, we built this Internet and a computer facsimile of the old home telephone network, that people talked to people and computers talked to computers.

     The idea that there were these mainframes that assumed all of these kind of end roles and that any computer could talk to any other computer.  But we didn't understand what was happening in the silicon world.  When we developed computers to be cheaper, there were computers more specialized, server computers, client computers.  The network became asymmetrical.  Most of the edge of the network where you and I sit as networks are actually on the client side.  What we were doing was using a network to access a service.  But this idea that there is a service opinion, that there is a unique point where the service is available, is at best a 2000s concept and maybe a 1990s concept.  It takes a long time to send packets around the world.

     What we started to do is replicate servers all over the world.  Rather than having one point of delivering service we started to have hundreds, even thousands.  The service point was just around the corner.  Just at the other end of your access network.

     So now instead of the network bringing the user to the service, the client to the service, the service is coming and knocking on the door of the client.  What is the network role at that point?  If the whole idea was that we were meant to be building this transit network, what has essentially happened is that the network has disappeared.  The transit to deliver the service through all these CDNs has been privatised.  All the world's submarine cables being built these days, the vast majority of them are run by private distribution networks.  Christian may optimistically believe that there are ten people publishing content, my pessimistic idea is that there are only five.  If you don't use one of those five, your content is no longer assuredly available.

     The rest of the Internet is now so toxically radiated, so full of attacks, so full of malicious actions that bring your content down, there are about only five folks who can publish your network an with stand this level of attack.  We managed to aggregate the service world and collapse the Internet.

     It is a little bit worse than that.  Not only is there no peering and no transit anymore, those five players are immense.  They are larger than most governments in terms of the amount of money represented in shareholder capital.  Not only is the aggregation of content, there's aggregation of value.  And we seem to be building right before our eyes another gilded age, another age where a very small bunch of players completely dominate our world.

     What do we mean when we talk about governance of the Internet?  And the Internet is now the incredible shrinking job and becoming almost nothing.

     What do we mean when the content is largely dominated by a small number of aggregators?  Your content is no longer visible unless you use one of them.  And if none of them want to publish your content, what happens then?

     To my mine, governance is fine when this is a large participatory competitive process where there are many ways of doing the same thing, where interoperation becomes important.  This is an historical concept now.  Now it is heavily aggregated, centralized in private hands.

     What is the role of governance when dealing with entities that by an large have a better touch with all of your citizens, have a better idea of exactly what they are doing because of the way they achieve this content aggregation is by payment of an intimate view of each and every client.  The mass market, if you will, of one.

     So when we talk about governance now, I'm not sure that we understand what we are talking about.  We are certainly not talking about a network.  A network doesn't exist as a public artifact.  That is gone.  Period.  That's gone.

     The whole idea of transit and moving packets around the world apart from what I'm doing right now, that's gone.

     So what we are talking about now is an entirely different world, a world of content and a world of service where in fact there is a very narrow an constrained pipe between the folks who are publishing and the folks who are receiving.  That pipe is dominated by a very small number of private players who effectively have moved so quickly, moved so rapidly in advance of our normal world that the regulatory strictures, the entire framework of control, of participation, of account interest, seems to have been ripped up.

     These folks set their own rules as they did in the guild the age and present them back to users an governments saying here are the new rules.  I hope you like them.  There's no negotiation.  These rules are absolute.  These rules are now rules that effectively the rest of us must abide by rather than determine for ourselves.  So when you talk about Internet Governance being perhaps irrelevant in the future, I perhaps the fact is the government has done an incredible shrinking trick.  There is no Internet left.

     There is a big issue in terms of free speech, in terms of access to information, dissemination of information.  These are very big social topics.  It is actually the public space we are talking about.

     >> FARZANEH BADII:  Sorry, just before you move to your next point -- Geoff?

     >> GEOFF HUSTON:  Thank you.

     >> FARZANEH BADII:  Before you move to the next point we have a comment from Mike Nelson from Cloudflare here.  We have an interactive session.  Anyone can just interrupt.

     >> MICHAEL NELSON:  Sorry to interrupt you.  I always learn a lot when I listen to you and read what you write.  It is useful to differentiate between the social media companies, the top of the stack, the security companies that are providing certs and the domain name companies and at the very bottom companies like ours, content distribution network and below that even the networks themselves.

     I mean, our network connects 150 data centers.  We are the most peered network in the world.  Our whole goal is to make sure that everybody is connected to everybody.  We are particularly concerned not about content.  We are concerned about computing power.  So we are very worried that one or two cloud companies will control the cloud.  Our answer is to make sure that everybody can connect to any cloud they want.

     I worry that what you are saying is one nightmare scenario if everybody decides to be the ugly monopolist.  Some of us are fighting very hard to constrain ugly monopolist and to have a multi-cloud environment and connect everybody to everybody and not be a gatekeeper no matter how many governments want us to be.

     >> GEOFF HUSTON:  Thanks, Mike.  The problem is, of course, that by and large security has failed us.  And it is not the vault of the (indiscernible), it is fault of the people erstwhile fighting the good fight.

     The software systems are fallible and the fallibilities have been used to the point where we can't defend each and every website.  A bit like Norman England, if you want security these days, you have to live in a castle.  You have to live in a castle that can with stand a terabit of attack.  Very few folk can build those castles.  It is nice to think that we can all with stand these massive attacks, but we can't.  There aren't a whole bunch of folk who can build that scale of castle.  Indeed, upon this panel you have at least two, and the other three aren't far away.

     Irrespective of what we would like, the reality is if you want your content to be always available to withstand all forms of toxic malicious attack, security failure has meant that you have to live in one of those very few castle hotels of content and like it or not, the outcome is that there are very few mediums to actually push your content out these days.

     This is not what we planned, but reality seldom is.  This is the reality we live in today.  Thanks.

     >> MICHAEL NELSON:  Can I be more explicit?  Is Cloudflare one of your -- do you have a castle?  We give 12 million websites their own castle.  And we are open to all comers.  We are giving them the infrastructure to fight off all the bad guys.  And --

     >> GEOFF HUSTON:  You're one of the five, Mike, yes.

     >> MICHAEL NELSON:  I want to make that clear.  It is different to be low in the stack and high in the stack.

     >> FARZANEH BADII:  I might have to moderate you guys.


     >> FARZANEH BADII:  So I can see that there is the comment there on then we are going to go to -- you have a comment?  That's all right.

     Go ahead, please.

     >> AUDIENCE:  Sorry, Peter Tonoli, Electronic Frontiers, Australia.  You say that Cloudflare is providing this bandwidth, this platform, but Cloudflare is actually providing a bottleneck.  For example, those people who use TOR.

     So Cloudflare, whether it is on purpose or not, Cloudflare is actually contributing --

     >> MICHAEL NELSON:  You clearly haven't read the last three blogs about this.  We made some progress.

     >> AUDIENCE:  No, I'm a pragmatist.

     >> FARZANEH BADII:  Thank you very much for being so interactive.


     >> FARZANEH BADII:  but I think Ted has a comment and then, Geoff, if you want to wrap up after teddy comment, that would be great.  And so we are here to understand whether the traffic consolidation is an actual problem that, like, with regard to standards of policy rights, with regard to the various aspects and nothing is set in stone. 

      I mean, Geoff might think this is a huge problem and somebody else might think this is not a problem.  We are not deciding anything now.

     It is just like the issues are out there to be discussed by the community.  Go ahead.

     >> TED EGAN:  Ted Egan.  A couple of points to make.  One is that what Geoff is doing right now is not at all unusual.  What we see is peer to peer video and audio traffic.  He talked about us imitating the telephone network and indeed we still do.  You have a very large number of RTC web-based mechanisms for enabling peer to peer video and audio to go among different people.  I think one of the things that is interesting about that is that the traffic patterns you see there are in fact very different from the traffic patterns you would get with content consumption.

     In terms of absolute numbers, it may not be quite the same as content consumption, it is important to recognize that both in particular geographic regions and well beyond them, there is a lot of this going on.  There are also still a lot of people building small scale web systems.  We saw lets encrypt produce 100 million different certificates over the last year or so.  That's because a bunch of those people wanted to have access to higher security and were rolling their own.

     Lastly, I'll point out that although the existence of cloud based computing means that a lot of people have shifted workloads to the cloud, it also means there is strong need for enterprises to be able to reach those clouds securely and with low latency.  Those are in fact network protocols and do require the sale kind of attention from standards bodies that historically have gone to other kinds of workloads.

     I think there are certainly changes.  The Internet exchange business didn't exist 25 years ago.  It was employee tenant Aquinix.  It didn't become a business at the very dawn of the Internet.  The fact that it is changing isn't a particular surprise.  It is part of the typical evolution.

     And the push of content towards the edge is really a desire to avoid latency.  We can see these trends in Internet companies as people optimize whatever application they are running on top of the network, they may change the pattern of bits, the most important is the one doing cache fill on your CDN, but it is crossing the boundary and behaving as an Internet flow.  We have to be careful in looking at these patterns not to extrapolate too far.  That evolution doesn't mean end.  It means change.

     >> FARZANEH BADII:  Thank you, Ted.  Geoff, would you like to make a final comment?  And then any time you want to come back again, you can just take the microphone and discuss.

     >> GEOFF HUSTON:  Thanks, yes.  Now, what I would like to point out is that traditionally the network has always been the point of control.  In the telephone system it was the telephone network, not the handset that became the point of the application of control.  The point where you had the orders to intercept, the orders to open serve.  It was the common medium that everyone used.

     But the Internet has changed all of that.  And I simply just cannot believe that where we are now is similar to where we were ten years ago.  Because there is no universal service obligation anymore.  There is no care and attention to maintain universal reachability.  Enterprises don't spend 90 percent of their money solving the last 1 percent of the problem.  They spend 90 percent of their money solving 90 percent of their problems.  The Internet is replacing broadcast television with an almost similar underlying technical analogy.  All it is is an access network into those local CDN points of presence.

     This idea that each company is not there to do a universal service for all of us, that each player is there to maximize its opportunities.  There is no money left in default.  There is no will to maintain default.  The network isn't universally interconnected anymore.  When we analyze the various routing tables in each of these players, we see a disturbing trend that not everyone can reach everyone anymore and perhaps even more disturbing, no one cares.

     It just doesn't matter.  It is not a commercially important problem.  The real problem is making sure that those well-known icons of content and media, the Netflixs, the Googles of this world, can deliver their content to you and I.  That's the two or 300 routes we really care about.  Where does that traffic come from?  A mere handsful of players.

     So yes, it is not a future issue.  It is a today issue.  And this distraction of the network and replacing it with local access to a local point of presence of service is I suppose the thing we have to deal with now.  The real question is, are old tools and techniques of multi-stakeholder participatory Forums relevant to this constrained world that we managed to build?  Thank you.

     >> FARZANEH BADII:  Thank you very much, Geoff.  I can briefly wrap up what Geoff said.  Internet Governance is dead and we can just go home.


     >> FARZANEH BADII:  And so this consolidation of traffic being a few actors' hands under Internet might have an effect on Internet Governance.

     But would that have an effect on standards-setting?  Did someone want to make a comment?  Panel?  Okay Pablo?  Before I go to Ted with that question, Pablo wants to make a comment and then go to Ted.

     >> PABLO HINOJOSA:  It is just for the sake of context.  It is a good thing.  And it feels a little bit weird to see techie talk here at the IGF, right?  With all the Cloudflare and straight talk for this technical community, for reference, Geoff is in his 2AM mode in his T-shirt but he is very much APNIC, right?

     I want to contextualise this.  The notion of Internet Governance that is being worked here is interpret the as in a very strict operational sense.  So these are standards practices that are the ones that run the networks.  We are talking Internet Governance in that sense here.

     But back to what Ted said.  We are not talking, for example, about community-led policy development processes or policy processes to define sort of bigger picture set of issues, right?  I think we need to differentiate sort of whether the Internet Governance operational practices are changing in a very dramatic way and to the degree that they are not more than ones that we were used to and another different thing is at this level of communications at the level of sort of the speeches that we heard yesterday, sort of this policy setting Internet Governance principles and decisions, that will be made at sort of governmental regulatory or via multi-stakeholder kind of processes.

     I wanted to give that context and setting for the rest of the conversation.

     >> FARZANEH BADII:  Thank you, Pablo.  I think we can, Ted, you can just ...

      >> TED EGAN:  Earlier on I said that Internet Governance was the set of tasks it takes to make sure that the Internet remains voluntary, open and interoperable and interconnected.  The point that Geoff is making is that globally interconnected has a different meaning now than it did many months ago in that there are now large numbers of people who may not be globally interconnected and don't know it and don't mind.  The reason he is giving for it based on the analysis he's done, most of the flows that they exchange with their peers on the network in this case clients talking to servers in the content case he's talking about don't require many of the routes it would take to make that global reachability real.

     It's an interesting and important point but it also points to future work rather than saying we just give up.  One of the things that he's pointing to is the fact that if in this situation you discover that you want to trade traffic, let's say, from Australia to Paris in order to participate in this and you can't, then the mechanisms by which you repair that are slow, cumbersome and may not be in your control because they are in the control of somebody along the path between the two.  Making that sort of network change might be something that right now is difficult because the presumption is that you are globally interconnected all the time.  We have a couple of options here to keep it interoperable and globally interconnected.  One of which is to focus on making sure that that is a core value of people who participate in the routing sim.  The ISOC manner system might be a vehicle for that.  There are methods for doing that by reaching out to the operator communities.  Frankly, Geoff has very, very good contacts there.  If he is not doing it, there may be a reason.

     Lastly, make changes in the global stack in order to pull out a new route to activate these methods takes more time than the PGG convergence time.  Yes, this may change the standard setting we need to do in order to maintain this, but I don't see this as saying it is not a good idea.  I see it as saying the current situation has changed enough from the situation when we built the current networks or the currents set of protocols that there might be a need for further improving the conditions.  That's a point on which hopefully Geoff and I can agree.

     >> CHRISTIAN KAUFMANN:  I'm torn between the big bucket of black things that Geoff is smearing all over the place and a somewhat more positive perspective on this, which I think you just captured.

     I think it was Vint who said that the Internet always just works.  Basically the Internet always finds solutions to getting to a place where it does provide the global interoperability.

     But on the other hand, this all needs to be paid for.  And my feeling is that the introduction of new routing protocols that might replace PDP will be subject to collective action problems and I think that the underlying story with Geoff is that once there is less value in transit, because there is less value to globally connect -- that's sort of the projection I think that he's making -- perhaps there is no way back.  Looking a little bit at Michael as the economist at the table, and providing you a bridge, perhaps.


     >> MICHAEL KENDE:  Thanks for the lead in there.  I wanted to talk a little about the economics of this.  First of all I should point out, I think Ted raised this, but there is some value to all this.  I mean when I was at the Internet Society we wrote two papers about the value of locally hosted content.  It lowers latency.  It lowers cost because you don't have to bring it across the ocean.  Much more expensive intercontinental transit versus connects versus local connections.  So there's a great value in having content local.  But now we are seeing there's a lot of, always unintended consequences or often there's unintended consequences of that.

     In this case there is this idea that there is some consolidation, that there's fewer and fewer of these that are gathering up all the traffic.  But just running along the line's of Mike's intervention that there is a difference between the motion media and the CDNs or cloud providers, there's a big difference in consolidation or concentration that comes from economies of scale versus one that comes from network effects.

     So economies of scale, you know, we see in car manufacturing, a lot of industries in Telecoms, a lot of industries we see economies of scale and in many ways we benefit them.  Whereas network effects, you know, nobody wants to be on a Facebook 20 people or social media with 20 people.  You want the network, right?

     And that presents a much different set of issues, a lot of benefits of course but also a different set of market power issues.

     In terms of the clouds, there is clearly economies of scale.  If you are going to 150 data centres you can't do that for one customer.  You have to do that for a lot of customers to make it worthwhile.  It makes it harder to enter but not impossible.  You can enter with, you know, building a better mousetrap.  You can enter because some of the inputs are publicly available.  You can get access to dark fiber or build your own.  Some companies are doing that.  Obviously, that raises your costs but there is a way of entering.  When we talk about this consolidation, to mix up what is happening with some of the operators above and what is happening, the social media companies versus the cloud, not setting apart some of the other issues raised it is not necessarily the case that it will lead to consolidation and it won't be possible to enter.

     And I would also say that the Internet itself has a lot of network effects and economies of scale the way that we are used to working, that we can connect and do video conferences from Australia to France without thinking about it.  Everything we do has network effects that we don't want to give up access to this whole network.

     The standards themselves are built into everything.  I don't think we want to have to buy a computer or download a browser that only works with one cloud provider or CDN.  So there is going to be a lot of inertia or positive efforts for everyone still staying on the same network and finding some other way to compete or exercise their market power.

     >> CHRISTIAN KAUFMANN:  It is interesting that you mention the different browser for different content, but if I look at this device, we have an icon for Netflix and icon for Prime Video.

     I am not totally convinced by that argument actually.

     >> TED EGAN:  That was built the other way.  Apple came out with something new and they made it in their favor with an app store that they had control over.  That's the other model.

     Whether someone could impose that on the open Internet is a different story.

     >> GEOFF HUSTON:  My Google Home Assistant doesn't talk to Alexa.  It only talks to Google.  When you talk browser, are you talking 1980s?  What we've done now is actually created a bunch of devices that are specialized to talk to only one service provider.  It's in every household now. 

      We seem to be actually creating an entirely different world than the one you are painting where the access devices are not flavored by the service they are delivering.  I would argue that things that are coming out now are flavored by the service they are delivering and indeed are locked against a single service provider.  That's where the device market has already gone.  It is hard to maintain your optimism in the face of that kind of observation.

     >> TED EGAN:  My point was, kind of talking to the ceiling here because we don't see your picture anymore.  It is a bit disembodied.

     >> We are trying to find out how hard you are trying to maintain your optimism.

     >> Like Alexa -- my point is you don't have to have everything Amazon to use Alexa.  You don't have to have Alexa to reach the cloud and vice versa.  It is changing.  And you can see that Olaf held up his iPad or one of his devices.  That is obviously changing some of the nature.  I think that is independent of what is happening in the core of the network and the consolidation that we are talking about here.  That was my point.

     >> CHRISTIAN KAUFMANN:  I would agree with your point but there is a component missing.  That actually picks up on the enterprise connection which we had before and how potentially the clouds talk to each other or CDNs talk to the cloud.

     If you find a better way -- let's start with the enterprise example.  If you have a enterprise connection to a cloud and you use it via a layer, there is no IP.  You drive over which device you use, you can actually change that.  You don't have to stay with the IP part especially, especially if you have protocols where one cloud offers you a feature and say if you use that protocol, you change some parameters which don't work in the rest of the Internet, you choose random ones, then you have a better performance and we give you a lower price.

     If that is your main cloud, I actually think most people will follow.

     The next part is, if the CDNs and clouds talk to each other, and they find a standard, right?  Which works better for them, now you don't have to go through a lengthy process.  Or necessarily involving hundreds of people.  You send an engineer from each company and say over these connections we can do different things.  We don't have to be using the lowest numerator, which is the Internet.  We can have our own part.  To the outside we react internet IP-wise that all the clients can connect, that's fine.  For the parts that optimize, we can actually do something different.

     >> TED EGAN:  Yes.  My point was, in the first scenario you laid out, would you want to be incompatible with everything else because you are having some of your traffic going over this layer too?  Would you want to give up compatibility with everything else?  Because you have to make sure everyone else goes along with you, right?  I mean --

     >> CHRISTIAN KAUFMANN:  Not for the connection.  You have two connections -- Christian, sorry, Kaufmann.

     And I think a company has multiple connections.  You have one for the Internet and one, couple a probably band connections or MPLS to the cloud.  Now the cloud component doesn't have to take the numerator to the Internet, right?  You can Change random protocols.  You can have an Amazon cloud box, not giving them ideas, where you connect to them, right, by the dedicated protocols.  All of that would work much faster instead of going through a standardisation body.

     >> TED EGAN:  So I would like to point out in fact what you are talking about there is a differentiation versus interoperability argument.  Sorry, Ted speaking.

     In particular, what you see in that situation is you can do optimizations either like in the services offered by the cloud or in the magnet by which you reach the cloud which are specific to that service relationship, right?

     And for an enterprise which is choosing that optimization, the result of that is they are losing interoperability with moving their tasks or their service to a different provider.  And that trade-off and we are going to look at Michael again because this is economics in some sense, that trade-off is often made and regretted later, right?  People are like oh, wow, I can save $50,000 a year if I go with this hyper-optimized MPLS connection to this one provider and find out I'm spending $60,000 extra a month because I am now a captured customer and I can't go anywhere else.  People learn that lesson and tend to say hey, it turns out I want to run the workloads locally in Azure, in AWS and I want to be able to run them in whatever competitors appear in China or India or other places where I care.

     And in those situations, the value of interoperability drives you toward a standardized solution.  And it then turns out it drives the standards to move up the stack from being the lowest common denominator towards having the same features or functionality that the people wanted the proprietary versions for because they still want the functionality and they can drive interoperable versions for it.  That goes back to a point being made earlier.  If you get to the point where you only need five people in the room to come to consensus on what interoperable looks like, you have two possibilities, either consensus emerges very, very quickly.  You get some regulator coming in and saying you five people than can't be in the room together anymore.  Or you get to a point where one of them thinks they can win.  That they don't need interoperability with the others because they can win whatever this market is and be the 80 percent player rather than the 20 percent player.

     In that case, you never get to interoperability with that one.  Then the other four go and do it without them.  And the reality is, we have seen this all play out multiple times in the Internet space.  We've seen it in IM and database replication and in other places.  Over the long-term, interoperability tends to win.  And Geoff may be right that there are changes in the patterns of behavior of these content providers an changes in the patterns of behaviors of the consumers of that content, but I don't think the long-term trend has changed.

     >> FARZANEH BADII:  Thank you.  We have a comment from the floor.  Please say your name.

     >> AUDIENCE:  All right.  Thanks.  Julianna.  Considering those are private platforms and companies, who do you think should address the interoperability problem?  Technical protocols or something such as antitrust rules and state regulation as a whole?  Considering those companies are not binded to the same transparency protocols as a public service would be?

     >> FARZANEH BADII:  Okay.  This is a very good question which actually we want to address, but based on Macron's speech yesterday, it will be the governments that will regulate them.  That was a joke.


     >> FARZANEH BADII:  So I am going to -- would you like to make a comment?  But then we will go to the Internet rights which is like the exciting part of the session.  Michael?

     >> MICHAEL KENDE:  Again, keeping in mind these two layers, there is a lot of challenges for antitrust and regulation at the social media level.  You know, if something is offered for free it's hard to assess the impact of a merger or any other activity.  At least in the way that antitrust is assessed in the U.S.  Europe is a little more looking at the impact on competitors.

     But at the network level and where there is economies of scale, there is nothing changed really.  You know, the tools are very good because they are all charging fees.  If they use these maneuvers to merge their way in or collude their way in to being able to raise rates in some way, antitrust is very good at least within a country or the EU.  Anti-trust is very good at either blocking those mergers or assessing them afterwards.

     So I think at the network level there are still possibilities without having to get to regulation.  Ultimately antitrust is the means to not have to regulate.

     >> FARZANEH BADII:  Thank you, Michael.  We had a comment from the floor.

     >> AUDIENCE:  Aren't we on the verge of this already?  With Google fiber, if you have Google fiber and Google caches websites so you have Google apps, you will basically never reach outside the Google network.  You've got an Android phone which speaks to the Google network.  You use YouTube.  You have effectively, you don't need the Internet anymore.  We are already on that precipice in my opinion.  I'm interested to hear what Geoff's opinion of that is.

     >> GEOFF HUSTON:  I'm trying hard to unmute myself.  My reaction is, Android now control 80 to 90 percent of the handset market in today's world.  They certainly have an extremely dominant position.  They have gone down the path of actually taking the protocols we use and hiding them behind a layer of encryption and using a proprietary version thereafter.

     I only mention quick in passing that similar, their whole swag of technologies now where one player believes it has critical mass in the market to not worry about interoperability to actually forge an entirely unique path through the system and come out commercially just fine.  So perhaps this whole issue of interoperability, Ted, is more like a pendulum that when you get a highly disaggregated market, then every player strives to interoperate with every other but in a highly aggregated market, the giants try to take those things, those assets off the table and privatize them where the wall or the lever is encryption so you can't see what is happening, or other mechanisms that simply remove the whole issue of interoperability and create dedicated attachments to a single service environment, today's world, I must admit, looks much more like the latter model than interoperability because it is not regulatory.  Interoperability becomes a secondary consideration and the will major consideration right now is actually aggregation and (indiscernible).

     Yes, I too feel that the giants are now so big that it is difficult to understand what our leverage is.  The Sherman act in the early days of the 20th century in the U.S. might have worked for united steel but they won't work for Google.

     >> TED EGAN:  I have good news.  Google brought Quick to the IETF, it is in the middle of standardisation and committed to turning down its flavor over it but it has been anxious to do that because the standards process is slower than they like.

     Realistically the chances that somebody is going to build a proprietary protocol that they mean to run across the network and not once to bring people in, it is certainly possible but it is not what happened with quick and we haven't seen examples of it at the transport layer in my experience of the Internet which goes back aways.

     I am told that I'm interrupting the rights discussion now.  So I will stop talking.

     >> FARZANEH BADII:  Thank you.  So now we get to talk about the exciting part of this session, not that the previous ones was not exciting, but we go to -- sorry, I don't know your name.

     >> I can help you with that.  My name is Rama with Global Policy Team at Access Now and I chair the Freedom Foundation in India which works on network neutrality and I will use that anecdote to talk about regulation.  Rights, I won't go into that, you understand about privacy, freedom of expression.  Interesting in this discussion, which I think regulators and government people say, although they say it as a preference for their interests, how do you enforce it and hold people accountable on these issues?  The most interesting point here, it is the Internet Governance Forum and we understand this in a certain way and coming from ICANN and Psych Universe.

      And as some of you heard yesterday french digital President Macron, we make fun of leaders all the time.  Any regulatory societal issue, why I mention this is because of some of the questions that you raised earlier about regulators and about rights connect with us.  To give you an example of net neutrality.  The most active issue in terms of net neutrality, forward-looking thinking is the issue of CDNs now as well.  Even the most recent positive example of net neutrality regulation, positive from the rights community standpoint, I know that telecos disagree with me, for example in the Indian debate where they pass regulation on Zero Rating and then they also issued recent regulations of technical discrimination similar to what the EU TSM and the Barrick guidelines put in place and what the U.S. had previously, they should have come up with how do you treat the issue of discriminatory behavior when it comes to both technical measures and economic measures. 

      The most people ask:  Does CDN give you a massive economic advantage or economic market benefit?  How do you feel about this?

     Most people would say rightfully so, maybe you don't have jurisdiction, maybe it is already being addressed by the private sector.  Maybe it is unfair correctly to treat CDNs in the same way as licensed telecom entities, using spectrum and subject to international regulation, but the challenge, I think you have is that it does have impact on the space.  You don't have enough answers. 

      For example, for many regulators globally my initial statement to them has been to actually you should talk to CDN providers or other cache providers, other actors in the ecosystem to find out what is happening in this market.  In terms, for example, understanding the nature of how transit and peering has changed. 

      I remember your fantastic paper a couple of years ago on that topic.  The marketplace has changed now.  The challenge for most people in this space if you were a human rights activist, civil liberties go, elected law makers and others, if you are not in this room, in this conversation, you don't know how the network is actually fundamentally operating.

     This is the challenge that comes to us in terms of if you want to be able to understand how track is flowing or in the example you gave about Netflix or others, it is not just about the app ecosystem.  There is fundamentally a market post generally, things to by pass the browser.  Some entities come back to the browser but there are complicated economic interests in favor of accessing direct services.

     In that situation if you want to enforce rights or if you want to be able to call for human business accountability, how do you do this?  I am giving you questions but I have no answers.

     I recommend to the people in the Internet infrastructure space and elsewhere is actually start going to regulators about how is this actually working?  How is the transit and peering system changed?  How is the structure, what is taking place there now they don't know.  The instinct is to regulate.  Escalation is easily possible.

     Remember the World Conference on International Telecommunications and ITU and the proposal to change how settlements and payments work in telecom space.  This is still a live question to many regulators and it is particularly there.  I agree with you on the question about the nature of the stack, right?  It is sometimes hard to place Google and say a Cloudflare is in the same level.  For most telecom departments, they are saying we have no ability to know what is going on in the network.  We don't know how to favor one particular interest.  That is unfair and also different from the interest of surveillance and many other questions there. 

      The question, therefore, sometimes I pose it to the infrastructure sector, how can we address the fact that there are a lot of network traffic is off the licensed regulated directly controlled telecom providers which are subject and regulated through legal frameworks and other mechanisms.  Sometimes it is business and account interest, right?  Who is publishing, how business is operating, what is the principles that people try to apply in terms of how the network infrastructure sector is working, even the number of companies that publish transparency reports when it comes to government data requests only in the last two years there has been a massive increase in that.

     These are the questions I sometimes like to see and grapple with.  I would love conversation on this.  The main thing I keep posing to people, Internet Governance there are sometimes IGF understanding of it versus what every day citizens and their law makers, regulators, elected, nonelected -- not everyone is elected, let's recognize, they want to address the entire ecosystem of how it is working. 

      Even with these five or six players, say one player will not take it up and they want to be interoperable and they want to be active with live entities.  There are telecom especially at this times that apply.  What norms apply to you?  How are you accountable?  How can individual users seek remedy?  That is something I am looking for answers for.  I'm curious to hear from others, if someone wants to complain, saying I can't easily enter the market or I'm worried you are affecting my rights.  How do we achieve that remedy?

     >> FARZANEH BADII:  Thank you.  Farzaneh speaking.  We are talking about Internet governance.  When we come and talk about the local policy and regulating at the local level, I think that's a whole set of other conversations which is I don't think that Internet Governance Forum we can resolve that really.  So if we can have your question like at a more global level of how do we resolve this issue and hold them accountable with Internet Governance mechanisms that are there, that might be a better framework to work with.

     Collin, you have a question or comment?  Go ahead.

     >> AUDIENCE:  This is Collin Kurre, for the record.

     I like that you brought up accounting human rights among private actors.  There is an alternative to regulation or something that can be implemented in parallel which is better collaboration between stakeholders such as companies, civil society and national human rights institutes on developing kind of self-check mechanisms such as human rights impact assessments or gap analyses to try to have some sort of alternative to coming in with a regulation that could have unintended consequences.

     Then for remedy, these same kind of tools or mechanisms could be equally useful for identifying gaps in the provision of systematic and consistent access to recourse.  I wanted to throw that out there.  That's something that Article 19, Access, has done some work on that.  There is work done in that space.  It is moving slowly but things are progressing on that front.

     >> MICHAEL KENDE:  I appreciate that.  I think the way to encourage people to think about, sometimes we don't expedite these conversations or understand how can we push proactive, voluntary business accountability with human rights, the challenge is often the regulators take the worst examples and force the regulation.

     The thing I would address, sometimes it is important for us to take on board the difference between global Internet Governance and a local policy issue is super thin.  If something is a huge issue for say the French government, the fact that France is hosting IGF means it will enter the debate.  We have seen this with other governments, I shall not name them, that do this often.  I flag this because sometimes look it has not come up, it will not affect how we have this conversation at the IGF, at ICANN, in other bodies.  Not so much ICANN here, I know, but at the Internet engineering level.

     We are -- what I am trying to flag here, we need to understand that this will escalate quickly.  The people not knowing how the Internet operates is a big challenge.  I want people to talk more about what is going on there.

     I want to designate, how do you deal with human rights, who is accountable and how does that work?  Sometimes technical questions.  My client is doing that, that's totally fine.  There are lots of examples, most recent on hate speech where the Internet infrastructure companies say yes, these are the policies we apply and this is how we operate.  Can certain customers use CDNs completely?  If the customer is directly engaged in human rights atrocity, how does that work?  How do we engage with that?  That's an important question.

     The thing that is most useful also to perhaps do is be able to say this is literally how the Internet operates today.  This is how our network levels are operating.  Be able to bring that public data forward.  I think be able to publish not just transparency reports about government data requests or censorship requests but go more on these is, this is the network topology or what we are dealing with.  It will help the conversation.  Again I'm meaning to be broad because I want to prod people to see how they will respond.

     >> GEOFF HUSTON:  Should I comment on that?

     >> FARZANEH BADII:  Yes, and then we have a comment from the floor.  You go and then Ted. Go ahead, Geoff.

     >> GEOFF HUSTON:  Thank you.  I suspect there's almost a misunderstanding around motives here.  It is not that any of these service providers wish to do the wrong thing or even the right thing.  It is actually not a concept.

     What they are trying to do paradoxically is actually doing exactly what users want, precisely.  Understanding their needs and servicing them.  And indeed, they are trying to crowd each other out in trying to make sure that they have a better understanding of each and every user and understanding not just what users want but what they might want in the next second or two.

     When you talk about these related social issues around what is the social responsibility, how are they motivated to do the right thing?  I suspect that's a kind of conversation about Mars and Venus.  These companies just don't see the world that way.  They see the world in a very pragmatic view of going, there is potentially 7 billion consumers.  We wish to service their each and every need, second by second.  That's what we want to do.

     It seems odd to try and say take's the wrong thing.  They would argue that is precisely the right thing to do, to do what people want.

     How would the governance debate intersect with that motivation from this new emerging content service industry which they regard is being totally censored from doing exactly what we want rather than paddle upstream and do things we don't want.  Thank you.

     >> FARZANEH BADII:  Thank you, Geoff.

     So we have two comments from the floor.

     >> AUDIENCE:  My name is Fiona from UCL in London.  There's two comments because I've just got one from the last point made.

     The first one I just want to ask really everything that we have spoken about so far is Facebook and Google and I wonder about the other service providers in Asia, Alibaba and stuff, are they working exactly the same way and models and infrastructure.  Is it developing exactly the same way?  So the question is the -- are the questions the same.

     The second one in terms of governance, myself and Madeleine Tarhey were working on an IoT and global Internet Governance project.  So it is all very fine, everybody wanting to do what they wish at any time, but the point of government and the point of law is to prevent people doing things that cause harm to others.

     So one thing is immediate harm to your neighbor.  But there is also transboundary harm.  In terms of international space there is state responsibility between countries.  There's state responsibility to try to ensure that non-state actors do not cause harm beyond their jurisdiction.  There are legal structures in place to hold to account what is happening.  You can't build an infrastructure and just talk about it as if it had no external ramifications and that they are not connected.  They are connected.  That's what we've kind of been waking up to more and more and more.

     We have an oligopoly.  That makes it similar to the hydrocarbon economies that we had to deal with in climate ways.  So yes, that's it.

     >> FARZANEH BADII:  Would you like to take that first question and respond to the first question, Geoff?

     >> GEOFF HUSTON:  If the question was really about are Facebook, Tencent and Alibaba any different from AKAMAI or Google or others?  They are more like Google than AKAMAI.  They profile users, understand their needs, target ads at them and try to create an environment where the advertisement is not something irrelevant.  It is not spam.  The advertisement is something you want.

     Is there any difference between Alibaba and Google in that respect?  No, they are exactly the same company doing the sale job in a different language, I suspect.  I expect they are not even using different languages these days.  No, there's no difference.

     >> FARZANEH BADII:  Is there an opinion -- is that opinion or is there data.

      >> GEOFF HUSTON:  I'm sorry?

     >> FARZANEH BADII:  Is that an opinion of yours or do you have data on that?  Is this like a study?

     >> GEOFF HUSTON:  Data in this area is remarkably hard to come by.  Looking at users -- data on users is hard to come by.  The major data we look at in terms of the presence of these kind of companies is their public shareholder value and size of the company compared to other companies.  It is also looking at their reports where they have to file reports as in the case of America, to look at their reach and extent of activity.  When you say can I measure users?  That gets a little bit difficult on an ethical basis, on a practical basis. 

      You know, these are difficult issues to actually expose precisely what you and I are doing and understand our behaviors at that level.

     >> FARZANEH BADII:  Thank you.

     >> TED EGAN:  If you want to ask if others are trying to copy the bigger players in terms of the networks trying to keep users in as much of their network infrastructure as possible while collecting data, if that is your question, that is partly that is true.  I think the challenge is within China domestically that is a key place for data and also incentives, begin the way things work, I'm hesitant to look at how it happens there.  Looking the eCommerce behavior across the will region, it is not empirical.  This is opportunistic data from CTOs and network administrators and Ministers.  Some of it is public from investment calls, earnings calls.  They are all investing to keep people on their infrastructure.  There is tension sometimes. 

      We work with the larger players, AKAMAI and others, but we say we want them to access on our infrastructure.  Sometimes we have consortia including the investment firms from China, particularly Alibaba, TenCent, and others in that space.  That is taking place.

     The question doesn't go towards advertising practices of others, but how they operate at the network level, they are similar and envious of the network capabilities of some of the big companies.

     >> FARZANEH BADII:  I have forgotten your second question.  If anyone on the panel remembers?  I would like to take the will second question.

     Would you like to restate?

     >> AUDIENCE:  It was more of an observation about the responsibility in relationship to building an infrastructure.  I build a road, I put cars on it, it kills people.  The State has a role to protect its citizens.

     At the global level we have international law where there is recognition of state responsibility to not cause transboundary harm and that is filtered in through various legal, transnational mechanisms, standards, hard law, soft law to try and ensure that businesses are held more accountable for their actions.  If we are seeing increasingly, because there is more data available, more evidence, that the ways in which the new technologies are used are negative particularly around human rights impacts but also environmental impacts, that is in breach of international laws which countries are signed up to, many of them are signed up to and they have to be held actable to that.

     I wonder where that falls into your discussions about Internet Governance.  In my discussions about Internet Governance, that is in the frame, you know.

     >> FARZANEH BADII:  Thank you.  Farzaneh speaking.

     Basically, my personal opinion is that we are at a global Internet Governance is not a vacuum of lawlessness or normless.  We have also, as Ted mentioned, there are various norms within like the Internet architecture, for example.  The manners and various norms that the businesses have to -- well, they don't have to abide by, but they do order some soft enforcement mechanisms for them.  Reputation-based.

     It is not, I don't think that they are functioning or operating in anarchy.  But Ted, would you like to ...

     >> TED EGAN:  Would I like anarchy?  This is Ted Egan.  It was a good idea but it did not work.  I'll go with no, I will not go with anarchy.

     There were two questions.  The second question was in the guise of a comment.  One is, are we doing anything about that?  That's a big part of the second question and the second part of your second question that was directed to us:  Is what is actually happening falling within state purview or does it fall somewhere else?

     And to take your state purview, the answer to that is sometimes.  If what is going on on the network is damaging in a way that is harming something which the State has a duty and has taken up the duty to protect, then you may invoke the state to engage in that protection.

     A good example of this is the toxic behavior that Geoff was talking about at the very beginning, part of which is evidenced in DDOS attacks.  You can say hey, this DDOS attack harmed me because it prevented me from getting customers.  I can point because I see the source addresses and know where it came from.  I can point to where it arose from.  We will assume that we had some way of assuring ourselves that the address with return routability or something.  We can say to the state, hey track down this and make sure they don't do it again.

     You have to be at a large scale attack or a longer period of time to invoke the state for that because it happens frequently.  Most states ability to engage in that remedy is relatively limited.  They don't have the technical capabilities to do that and often the traffic crosses territorial boundaries.

     And in those cases they have to invoke very, very large scale methods to send the requests through treaty obligation mechanisms to the other states to get this done.  So it is possible but it takes a fairly large level of harm to invoke the state in cases like that.

     So it is definitely there as a mechanism, but it is heavy weight and hard to invoke.

     The second thing is, are we doing things about it?  The answer to that is yes, but that is because many of the things that we see that used to be fairly effective attacks are things that we can actually say look, if you invoke the correct filtering mechanisms on your outbound traffic you can reduce harm to the rest of the world.  We use the incredibly easy to understands argument that you should implement BCP38 because we are obscure geeks.  We apologize for that.

     What it means, if you are sending traffic to the network that you think you should not be sending, you should stop.  The other thing is, there's a bunch of other mechanisms like that that have been introduced over the years to do return routability checks and otherwise maintain that DDOS is less and less effective.  There is an arms raise there of people who control botnets but that doesn't mean we stopped our side of the race in trying to make the protocols more and more resilient in it.

     There is work in the IoT space to make them less effective as vectors and work going on in the routing space.  It is really going on all the time.  Efforts to clear things of toxicity will probably have to use both that system of technical response and occasional state action going forward.

     >> FARZANEH BADII:  Madeleine, do you have a comment on this issue?  Then we go to ...

     >> AUDIENCE:  Your comments are interesting but it veers off a little bit way into cybersecurity.  I think what Fiona was referring to was not malicious actors.  She is talking about industrial policy, about trade policy, of legitimate actors.

     I thought it might be interesting to bring in, because I know he's sitting in the front row there, Blayne Haggart -- sorry, Blayne.  I know he works on that.


     >> BLAYNE HAGGART:  Let's see what I can come up with in five seconds.

     >> No pressure!


     >> BLAYNE HAGGART:  I guess essentially the only thing I can think about and tell me if this is what you are thinking about, essentially basically who whose responsibility is it not just to deal with attacks like that but in setting the overarching framework for these things, especially for something like the Internet that is as President Macron pointed out yesterday, has become much more than kind of a technical thing.  It essentially became the ground work for all of society economically and so on.

     So it comes back to who is responsible, who should be responsible for essentially exerting kind of the structural power.  Because these rules are going to be set by somebody.  The question does seem to be who is going to be setting it.  Internet Governance isn't dead.  The question is, who is going to do it and in whose interest?

     Is that what you were going for?  I don't know.  So basically the argument seems to be that, it is that it comes back to a question of legitimacy.  When you are dealing with very specific technical issues, you can claim legitimacy there, but beyond that if you have something that is reaching so far beyond just kind of a communications network, you know, you have to go back to a deeper more fundamental democratic legitimacy in terms of democratic societies.

     >> MICHAEL KENDE:  My response to this, I have to challenge Geoff.  Democracies derive legitimacy from the consent of the governed.

     >> Yes, citizens.

     >> MICHAEL KENDE:  If the government has the active consent of the user --

     >> How do you establish that you have that?

     >> In terms of service.

     >> MICHAEL KENDE:  I'm asking, this is the challenge, I think, to argue sometimes is to say in the larger plays is to assume it is always coming from the users and you have users in the -- were forced to use this, but we haven't consented to say Facebook's discern policy, it is easier sometimes for especially people in the network connectivity system to say you are subject to sanctions law.  Show me a couple of providers that work with the authorities, I'll buy them the cake and get them bail money because the US DOJ is going to talk to them.

     >> TED EGAN:  Ted again.  The point that you are making is that there may be ways in which to structure that consent so it is acceptable to the rest of society to know that this person in deed did consent to something.  We have that in a whole variety of ways with sexual relations.  It changed over the years, what consent has to be active, continuous, all of those things.  There's a question here about whether the consent of somebody is something that can be structured in a way so it is accepted or whether we are always going to have this:  This didn't count as consent because the user doesn't understand the ramifications of it.

     There is a question of free will.  If they don't understand the ramifications of that, how do they consent to the governed.

     >> FARZANEH BADII:  I have to moderate now.  EFF in Australia, you had a comment about an hour ago?  And Pablo wants to --


     >> FARZANEH BADII:  Okay, so Madeleine has two words.

     >> AUDIENCE:  Two words, informed consent.

     >> FARZANEH BADII:  Now it is your turn.

     >> AUDIENCE:  Greats thank you.

      I keep hearing multi-stakeholder and IGF.  We had a representative from cloud play here.  The representatives from these other companies that we are talking about, are we doing ourselves a disservice by not engaging enough with these organisations?  By not engaging are we creating a self-fulfilling prophesy of Geoff's?


     >> AUDIENCE:  Out to you, Geoff.

     (Overlapping speakers.)

     >> GEOFF HUSTON:  I'm sorry, I don't think they have an interest in engaging in this.

     (Overlapping speakers.)

     >> GEOFF HUSTON:  The issue around informed consent is remarkably informed.  What we are talking about, I suppose, is as we enter this new information age, private companies are actually more efficient in understanding the individual needs and obtaining individual consent than governments.

     They are more effective at maintaining relationships with hundreds of millions of end users than governments.  And now comes, I suppose, the dramatic tension between catering to the needs of the individual versus the public space.  When private companies work around the public space to directly have hundreds of millions of relationships with their end users it prevents consent.

     (Garbled sound.)

     Is the common space that society is thriving.  Now the task for governments is the power they didn't have before is the consent of individuals.  They are quite happy to give their consent.  It is perhaps not even about informed consent.  It is about the tension between (indiscernible) and pandering the needs of individuals versus understanding and ensuring the health and wellbeing of (indiscernible).

     And I suspect in the last five years the real loser in all of this was not me.  I can buy whatever I want in my space, I'm fine.  But the public space in which I live in is now shattered, eroded and destroyed.  And that fills me with great sadness.  This is where we have failed miserably.  Thank you.

     >> FARZANEH BADII:  Thank you, Geoff.  Pablo, you have a comment?

     >> PABLO HINOJOSA:  Well, I think, I'm conscious about the time and wrapping up.  It is about breaking silos.  Part of it, I think, is that the difference between private self-interested business commercial spaces and that one that has a public effect and requires public interest considerations.

     Then whether you delegate your rights of participation to governments to regulator you go through the cost and burden to proactively have a say in the decisions, provided there is a space for that in a muscle stakeholder setting.

     So for me the answer to the question of Internet Governance becomes irrelevant because all these private interests considerations and the evolution of those practices is definitely not.  I think we really need to invent those spaces where the different interest groups will collide and define how these move forward.

     I think that's very important to start this conversation here, which you brought and that is fantastic.  But there is an endless sort of series of considerations that we need to start thinking about.

     >> FARZANEH BADII:  Yes.  Thank you very much.  I actually, the intention of this session was to start the conversation and, of course, we are going to come up with this accountable institution of the Internet.  You will all be involved.  We will have multi-stakeholder guidance.

     I believe at IGF we are here to talk about policy issues and governance problems and think about how we are going to resolve them with the mechanisms that we have.  Issues like the bottom-up multi-stakeholder approach.  If it is not multi-stakeholder, call it something else but not government regulation.  That will not be transnational.  That will not give us the interconnectedness and the open Internet that we want.

     I believe that is my personal opinion, but someone wanted to comment.

     And yes, please.  Go ahead.

     >> AUDIENCE:  Hi.  My name is Catherine.  I will try to be quick.  I want to have a quick comment to the two ladies there that asked about the operating on commercial model of Alibaba and Chinese companies.  My comment is, I think there's a difference between commercial companies operating in democratic society and then which respect of rule of law and a judicial system and the companies that operate under other countries.

     The reason I say that is because which is true that what Geoff said that when they were startups, they were also a lot of startups now in China.  They are competing for market share.  When they are trying to get market share, they only have commercial interests in mind.  Then sure, that is their only goal, right?  Maybe political sensitivity is less of a concern, but then there's also the case that some got punished by the government and had to close their whole website which is right now the top app in China.

     Then if the government wanted to punish them, they can punish and the market price in China would drop five or 10 percent overnight.  In democratic companies the state has a different kind of power dynamic with companies, especially tech companies when they are very, very big.  So this is just a comment that I wanted to raise.

     >> FARZANEH BADII:  Great.  Thanks so much.  This is the last call for any comments.

     Then okay.  So I have proposed in the session description that we are going to take a vote on whether you think that Internet Governance will be irrelevant because of the traffic consolidation and the issues that we brought up.  This is one question.

     Do you think so?

     >> This is a classic case for ITF humming, a --

     >> FARZANEH BADII:  You know, it is not a technical space.  We say words.



      >> FARZANEH BADII:  So no one is humming.  So my second question, should we do something about this?  Do you see this as a problem?  That we should pay attention to the traffic consolidation and we should like follow up and bring up the issue in the right venue?  I don't know where those right venues are?  Do you think so?  Yeah?  Yeah?  Okay, great.  I see some reaction.  Great!

     >> So just quickly, two points.  I think that I've taken from this fascinating discussion.  One is I think that we can draw up -- drop the word "traffic" from this discussion and talk about consolidation.  It is very difficult.  Every time the discussion is at the traffic layer we get into the content layer and the vertical integration.

     And then I think the other point, I think Geoff raised it, was data.  It is very hard to sit here abstractly talking about the impacts of things when we have no data.

     >> FARZANEH BADII:  It should be more nuanced in the conversation and just kind of like separate the issues out.  Yeah.

     >> I think it's this access to data.

     >> FARZANEH BADII:  Yes, I got the comment.  Thank you.

     Thank you very much for attending.