 We will see the whole interview somewhere. Okay, yeah, Yves Alexandre is there. Okay, so, so we, hi Yves Alexandre. So, in terms of of the timing, we start with Lorde La Roudière, who is the chairman of the regulator. They are set in France of telecoms and the parcel services. Then we'll move on with Yves Alexandre de Manjoir, who is at Imperial College, but I think joining us from Brussels, I believe. And then we'll move on to Philippe Val with us, as you all know, CEO of La Poste. And finally, we'll get Pierre Régibot, who is a chief economics economist at GG.com. So, without further ado, I suggest that we start with Lorde. Each participant will have about 15 minutes, and then we'll have an open debate between them and also with us in the room. No? Thank you. Mr. President, ladies and gentlemen, thank you for inviting me and giving me the opportunity as president of the Autorité de Régulation des communications électroniques postales et la distribution de la presse de participer à cette table ronde visant à s'interroger sur les voies et moyens de la politique de la concurrence et de la régulation afin de promouvoir une économie numérique loyale et inclusive. C'est un honneur pour moi de participer à cette table ronde car l'arcebe, c'est beaucoup investi sur ces sujets. Je suis donc heureuse d'avoir l'occasion de vous échanger avec vous sur nos réflexions. D'abord, dès 2018, l'arcebe a produit un rapport sur les terminaux que nous considérions comme le maillon faible de l'Internet ouvert. Et nous montrions en fait que le règlement européen visant à garantir la neutralité d'Internet sur les réseaux et aux fournisseurs d'accès à Internet ne s'appliquent pas sur les terminaux, laissant en fait une faille dans la neutralité d'Internet. Pour offrir un service sur un terminal, vous le savez, il faut passer par un magasin d'application du fabricant absorb pour l'iPhone qui vous dicte ces conditions et vous n'avez pas le choix si vous avez un service à offrir de passer par un autre magasin d'application. En 2019, nous avons produit une note sur les plateformes numériques structurantes et cette note identifie les défaillances caractéristiques des marchés numériques et met en exergue des problèmes d'ouverture de l'écosystème numérique. En 2020, nous avons répondu à la consultation publique de la Commission européenne sur le Digital Market Act et en 2020, nous avons rédigé une note sur les remèdes aux problèmes posés par les plateformes numériques structurantes où nous dressons un panorama des remèdes susceptible de traiter les problèmes identifiés. Nous mettions en avant par exemple l'intérêt de la portabilité des données des utilisateurs permettant de renforcer la liberté de choix du consommateur ou encore l'ouverture des plateformes logiciels des terminus à d'autres acteurs que le fabricant du terminal lui-même. Ces remèdes sont d'ailleurs inspirées du cadre de régulation des télécommunications. L'ensemble de ces travaux nous amène à penser qu'il est pertinent pour l'économie numérique qu'elle soit soumise à une régulation excentée, assymétrique. Pourquoi cette régulation excentée ? D'abord parce qu'un nombre restreint de grandes plateformes numériques deviennent incontournables dans la vie des citoyens et des entreprises et proposent de nombreux services qui font partie intégrante de notre vie quotidienne que ce soit Google, Apple, Amazon. Ces acteurs mondiaux sont en mesure de déterminer quel contenu et services peuvent être mis en ligne et à quelle condition les utilisateurs peuvent y accéder. De plus ils s'organisent pour maintenir les utilisateurs captifs au sein de leur écosystème. Alors par ce biais la liberté de choix du consommateur est finalement bridée. Les effets de certaines de leurs pratiques sur les dynamiques concurrentielles et sur les acteurs du marché sont souvent irréversibles. Il est donc crucial d'intervenir avant que les problèmes concurrentiels ne se matérialisent. De plus, compte tenu des enjeux, notamment en matière de liberté de choix du consommateur et dynamique concurrentiel, des caractéristiques du secteur fait de positions hégémoniques de quelques acteurs sur un marché, des pratiques des plateformes d'ailleurs sanctionnées, ces pratiques, mais sans que les sanctions aient profondément changé les comportements des acteurs, nous pensons qu'il est pertinent, qu'il y ait une régulation excentée, assymétrique. Donc nous saluons l'initiative prise par la Commission européenne avec le Digital Market Act qui représente une avancée majeure. D'abord l'échelle européenne nous paraît le bon niveau d'intervention et l'Europe écrit précurseur comme cela a été d'ailleurs en matière de régulation pour protéger la vie privée avec le RGPD. Et puis c'est une régulation que nous appuyons de nos vœux, excentée, assymétrique, s'appliquant donc à un nombre limité d'acteurs principaux et nous pensons que c'est un bon principe. L'ARCEP a d'ailleurs largement contribué avec l'autorité de régulation européenne, le BREC, les télécommunications et avec le GREP aussi sur le DMA, et nos positions sont alignées avec celles des deux réseaux que sont le BREC et le GREP. Le BREC et le GREP ayant pris des positions communes et collectives sur le Digital Market Act. Avec le BREC, l'ARCEP note néanmoins trois points d'attention tout particulier. Le premier, et c'est crucial, il convient de préciser les détails techniques pour que l'application de certaines obligations soit efficace pour éviter le contour nevant des nouvelles obligations qui sont incluses dans le DMA. C'est essentiel. Si je prends l'exemple de l'interopérabilité, l'interopérabilité c'est la capacité du système à communiquer, à échanger des informations avec d'autres systèmes ou produits. Les détails techniques des interfaces ou des standards qui permettent cette interopérabilité doivent être clairs et transparents, partagés entre les différents acteurs pour que cette mesure d'interopérabilité soit efficace. La correcte spécification de ces détails techniques qui découlent d'une analyse est donc absolument crucial pour que l'intervention de régulation envisagée atteigne ses objectifs. En second, second point, nous avons préconisé, donc le BREC et l'ARCEP ont préconisé la création d'un groupe de haut niveau de régulateurs du numérique pour assister et fournir de l'expertise des avis, des recommandations, des études à la Commission européenne qui reste la seule responsable de la mise en oeuvre du DMA. L'idée est bien de requérir le maximum d'expertise pour alimenter la réflexion de la Commission dont les moyens sont nécessairement limités en nombre. Or l'expertise recueillie par les réseaux des régulateurs du numérique dans chacun des pays de l'Europe, de l'Union européenne et leur analyse et leur expertise permet de compléter finalement les moyens de la Commission. Cette proposition qui a été aussi portée par le Parlement européen fait partie du compromis final à l'issue des triogles. En troisième, nous insistons sur la nécessité d'une bonne articulation avec les cadres réglementaires existants. Ce matin, j'étais à Dauphine. Il y avait aussi un colloque sur la régulation du numérique à Dauphine, notamment sur l'intelligence artificielle. La CNIL disait la même chose. La CNIL disait aussi attotions a bien articulé l'ensemble des réglementations européennes touchant au numérique. En ce qui nous concerne, nous sommes attentifs au fait que le DMA amène une nouvelle régulation des messageries instantanées, en tout cas de celles qui sont retenues parmi les gatekeepers. On pense naturellement à WhatsApp ou Messenger, par exemple, que cette nouvelle régulation s'articule bien avec le Code européen des communications électroniques qui permet aussi la régulation des services de messageries instantanées. Il sera crucial d'assurer que les dispositions du DMA et du cadre des télécoms soient appliquées de façon cohérente au niveau européen. Naturellement, le BREC, tout comme les régulateurs des télécommunications dans chacun des pays, sont à la disposition de la Commission européenne pour pouvoir travailler sur cet enjeu. Toutes ces propositions sont portées par l'ARCEP et le BREC. Nous les avons soumis à consultations publiques, comme nous le faisons habituellement. C'est la façon de travailler des autorités indépendantes. Et nous l'avons testé auprès des différentes parties prenantes, non seulement des gatekeepers, mais aussi les concurrents des gatekeepers, les utilisateurs professionnels, les associations de consommateurs, les représentants de la société civile ou des experts. Enfin, en ce qui concerne le secteur postal, l'ARCEP et les travaux du GREP, qui est l'autorité de régulation des services postales au niveau européen, préconisent de tenir compte des aspects concurrentiels liés aux plateformes de e-commerce. En effet, le développement des propres réseaux de distribution de colis des plateformes de e-commerce est de nature à avoir un impact significatif sur le secteur postal et son évolution. En effet, ces acteurs peuvent être amenés à développer des stratégies dits de charity picking consistant à distribuer eux-mêmes leurs colis uniquement dans les zones rentables. Et de remettre aux opérateurs postaux et en particulier aux prestateurs du service universel, les partis non rentables de la distribution des colis amenant ainsi une situation concurrentielle sans doute aux désavantages de ces prestateurs de service universel. Dans ce rapport sur l'impact du développement des plateformes de e-commerce sur le secteur postal publié très récemment en décembre 2021, le grep a soulevé la question de la qualification juridique des plateformes de e-commerce actives dans le secteur postal, considérant que les plateformes de e-commerce actives dans la distribution de colis devraient être considérées comme des opérateurs postaux et donc soumises aux obligations des opérateurs postaux. Le grep poursuit cette année ces travaux sur le sujet et devrait notamment analyser comme nous l'avions d'ailleurs proposé l'impact sur la concurrence entre les plateformes numériques de l'existence de plateformes verticalement intégrées sur la distribution de colis ou non. En conclusion, les enjeux associés à la digitalisation et au développement des plateformes numériques sont nombreux et pressants. Ils font régulièrement l'actualité. Ils font régulièrement l'objet de décisions d'autorité telles que les autorités de la concurrence ou l'acnil. Mais ces sanctions n'ont pas forcément un effet important sur leur comportement. Et donc la proposition de la commission et les contributions du Parlement et du Conseil sont à la fois ambitieuses et vraiment intéressantes. Et nous les soutenons comme constituant une première réponse. Nous sommes le réseau des régulateurs compétents en matière du numérique représentant un soutien précieux pour la commission et nous pourrons fournir donc une expérience solide. Dans le domaine des téléscomes, ça fait 25 ans de régulation excentée depuis l'ouverture à la concurrence du marché des téléscomes. Multisectoriel, si on prend l'ensemble des autorités concernées par le numérique aujourd'hui, qui est crucial dans la régulation de ces écosystèmes de produits et services très variés. Dans ces secteurs dynamiques et innovants, la surveillance et l'analyse des évolutions du marché et non seulement des gatekeepers eux-mêmes mais des gatekeepers sur leur marché et les pratiques qu'ils ont sur ces marchés sont des éléments clés. Viall-Bérac, l'ARCEP contribue également à la rédaction de plusieurs rapports sur ce sujet. Nous comprisons d'ailleurs un groupe de travail parce que vous avez compris qu'on était très à l'an sur ces travaux au niveau européen au sein de l'autorité de régulation des européennes des téléscommunications. Nous avons un premier rapport qui va sortir en juin 2022 sur l'analyse de l'écosystème de l'Internet afin d'identifier les enjeux de demain. Une analyse aussi technico-économique de la mise en oeuvre des mesures d'interoperabilité concernant les services de messagerie instantanée. L'idée est de savoir là comment on utilise les nouveaux pouvoirs confiés aux régulateurs en matière d'obligations d'interoperabilité concernant les services de messagerie instantanée et d'ailleurs à ce titre-là ça sera très intéressant de voir des discussions sur l'application du DMA sur ce point précis. Permettez-moi en conclusion aussi de rappeler qu'à côté de ces principes fondamentaux d'ouverture à la concurrence qui sont à l'origine de la création et qui sont la base de travail des autorités indépendantes de régulation des communications électroniques et des travaux que nous menons pour assurer une plus grande liberté de choix du consommateur l'ARCEP a ouvert un nouveau chapitre de la régulation un nouvel axe que nous initions grâce à la compétence juridique que nous a confié récemment le Parlement il s'agit de concilier le développement des usages du numérique et la réduction de son empreinte environnementale nous sortirons prochainement notre première enquête pour un numérique soutenable c'est un axe de régulation que nous avions souhaité porter parce que c'est une ascente forte des citoyens qui c'est d'ailleurs invité dans le débat public au moment du lancement de la 5G et nous pensons qu'en matière de régulation nous devons prendre en compte ces enjeux pour pouvoir fournir par la régulation par la donnée des informations aux parties prenantes afin qu'elles puissent se positionner et ainsi nous les incitons à avoir un comportement plus vertueux Je vous remercie Thank you so much for this very nice talk We are going to move on to Yves-Alexandre and I'm sure we'll get back to interoperability and data and privacy and topics like this Am I right? No, absolutely actually Should we do it in French or English? If you have any preferences It's in English, yes English, English, sounds good Perfect, perfect I have prepared the remarks in English so that's great So first of all thanks a lot for the invitation to participate in this panel and my apologies for not being able to be with you in person in sunny Toulouse It could have been a pleasure So basically I just want to make sure that I'm leaving enough time for a discussion which I think is going to be the most interesting part of the panel so I really try to keep my remark quite short five to ten minutes and basically a few, you know kind of technical observation from my perspective since the publication of the report to Commissioner Versteyer that we did with Jacques Aminot Samar Yeah Do you mind speaking up a little bit? Oh no, of course Is this better? My apologies So yeah, so I just want to make sure that I'm going to give you enough time for the discussion so I make sure that I keep my mark to be to be very short and basically a few technical observation from my perspective since the publication of the report to Commissioner Versteyer with Jacques Hemaire and Nika Schweitz I should also emphasize that obviously it's all of this is only my personal opinion as I'm currently a special advisor to the Justice Commissioner on AI and privacy So basically I think I mostly have five observations since the publication of the report that I thought would be of interest for this discussion The first one is linked to data and machine learning models I think even more than before and even more than in 2019 we've seen the most recent machine learning techniques and in particular self-supervised learning models requiring more than ever extremely large amounts of data I think this is something that we had already emphasized in the report and it's something that's even more true today just to give you a note of magnitude language models I'm sorry but at least for me it's a little bit hard to hear you it's not enough maybe My apologies do you want me to try I can switch maybe turn on the sound of your microphone or something because we have a lot at least I have a lot I'm following Okay, my apologies is this better? Hey, Alexandre, can you hear us? Shall we switch the order and then get back to you? Philippe Okay, so I suggest that we change the order while this technical issue is resolved I'm sorry about that so we are going to have Philippe Val actually speaking now so why don't you go ahead and then we'll go back to Okay, okay, I will do that Thank you Ladies and gentlemen, in fact four years ago in the same town and in another room of TSE we were addressing two questions the first one is should digital platform be regulated and if yes, how to do it those two questions are very interesting because they are showing that the world has changed a lot why? Firstly, the world is now far more digitalized and on the regulation side we have major events which are occurring on Thursday, the 24th March European Parliament and European Council agreed with the EC the European Commission Proposal about DMA, Digital Market Act, which is a major type of ex ante regulation this is a major change and Thursday last week in Washington DC the American Innovation an online choice Act which could be named as a U.S. DMA has been able to have a new hurdle a new step in the process before becoming a law so those two examples could can show you that we have made significant progress on the awareness about the digital platform and on the regulation especially with this ex ante regulation type significant progress however, this is not the end of the question why? Why? Because those platforms are becoming more and more important in the economy and in the society in fact, we could say that the GAFAM or the BATX Chinese form of GAFAS are not only becoming dominant players they are at the heart of the economy and at the heart of the society they are firstly, I think, working very well and bringing us infrastructure services and for instance, in the past we are seeing those operators as benefiting both to consumers or to companies and from my own point of view I am seeing them as major factors of change major factors of acceleration in growth and in change but there is a dark side of this influence and this dark side is that all those platforms have now established dominant position through killer acquisition which does mean acquiring potential future competitors with very high prices using a monopsony or a monopoly position bundling strategy to look up the consumers or planned obsolescence of their network or devices so at this stage, I think that we could say that their position of market domination is an issue for everybody which leads me to a second conclusion after the fact that we have to recognize that major change happened and that we have made progress my second conclusion is that we have now to prepare a second step in regulation we can't just be satisfied with DSA, DMA and the new coming American law about those platforms because I think that those platforms are becoming more and more powerful in the economy they are stimulating, challenging not only industrial operators like La Post and I will give you an example of that they are also challenging regulation authorities how are they stimulating a very old company as La Post? very old because we are now in our sixth centuries of activity which could be qualified as very old and we are stimulated we are challenged by them especially by Amazon as you are working on the competition regulation as you are teaching economics what is our situation with Amazon? Amazon is our first client and at the same time Amazon is our first competitor I repeat it because it's not so easy to understand let's imagine the situation when your first B2B your first client is also your first competitor so in fact for now 10 years I think that Amazon has been a major factor of change for our organization which is good which is good if everybody in the market is respecting a same level playing field which is precisely the issue of regulation so La Post and its 250,000 employees are committed to be challenged and to adapt to this situation and to the market power of Amazon I think that is exactly the same for the regulation authorities and that's why I was speaking about the second step of regulation I think that self-regulation is not far from being not very efficient this was an English understatement secondly, code of conducts are not very efficient the traditional competition regulation is very good is very good and monitoring market share by judges or by the EC the European Commission is very effective it is not sufficient but it is very effective DMA is a great hope of effectiveness of the regulation however I think that we should work on new forms of regulation I think that both the regulators and the political authorities have to adapt themselves like the companies like La Post it does mean that they have to accelerate their own production of rules and of regulation because the context of digital platform is that everything in the world is moving far most quickly which is leading me to my last conclusion the digital platform power represent a permanent challenge for regulation authorization the regulation authorization should change themselves at the speed of the market and at the speed of all the operators which is a big challenge for such organization because if it is a challenge for Amazon and GAFA's competitor to adapt it is also a big challenge for the institution and for the regulation authorities to adapt which does mean at the end of the day that we as competitors to those platforms have to create a new business model a new paradigm but at the same time that regulation authorities or political authorities should also invent and create their own new paradigm to be able to manage those digital platforms well thank you very much Philippe it's very interesting and we have the message we all have a lot of work people in this room both academic and regulators have a lot of work to do you love the first episode and we are happy to offer you now episode number two with alexander return thank you hopefully it's gonna work better this this time so as i was saying basically i just want to i i'm going to keep my remarks to be fairly short a few minutes as i really want to leave time for the discussion and just limit them to a few basically what i see as technical observations since the publication of the report was the first one basically relates to the recent evolution in machine learning techniques i think in the report we already emphasized the the fact that machine learning models are largely dependent on the availability of large amount of data i think it was true then i think it's even more true today was the most recent self-supervised learning techniques that require even more data than before just to give you a sense in order of magnitude when you look at you know modern you know last year basically language model that were trained by microsoft and nvidia the megatron turn basically it's a it's a model that has inside its it's architecture five hundred and thirty billion parameters there's five hundred and thirty billions variable within that model that are basically learned from data and in a recent paper google be so sure okay even if somebody you're stuck uh we see that's a rule that the connection this time which is terrible oh you're back i'm back okay you're back okay sorry about this i tried to switch to five g so hopefully it should be more reliable um where was i stuck i don't know yeah you had a very large number okay a very large number uh five hundred thirty billion is is the last one probably um so yeah i think basically so large-scale language models uh today have an extremely large number of parameters and again i think the five hundred thirty billion just helps you know put things in perspective in terms of the flexibility of these models and the amount of data that is going to be required to train these uh models as well as the deep pockets that are required to basically create these models and importantly i think such model like extremely uh general they're self-supervised and can be then applied to a very broad range of applications so i think this makes a lot of the comments we made in the report and the importance of the availability of data uh even more pressing today than it was in 2019 the second one has to do with uh what we called back then protocol interoperability basically ensuring that two systems can fully work together and that complementary services uh can still be can be provided that is at least in my opinion still a challenge in practice today it is a challenge when it comes to iot devices but it also is a challenge when you look for example at browsers uh which you know today we're really seeing kind of a fragmentation in what kind of browser are for example being supported by specific uh services the third one has to do with data interoperability uh which is something that we talked quite a lot about in the report and that i think has been taken very seriously by the commission uh in particular when combined with a ban on self-referencing i think this is this is really great uh however i think the challenge moving ahead as pointed out by madame de la rodière um will be how to implement uh the requirements in ways that are going to be meaningful how are we going to make sure that apis can be relied on can be accessed on reasonable terms um that we can ensure that the important data is indeed available and accessible through these apis and all of this while ensuring a high level of security and privacy both in theory and practice and i think this is going to be a major challenge uh moving forward the fourth point is relates to last minute changes that i think we've seen from the european parliament um to include what we had called in the report full interoperability basically the ability of competing services uh to interoperate more specifically in this case messaging systems that will now have to allow you to basically send messages from one system to another this is something that at the time we are hand emphasized requires much deeper integration and standardization than protocol interoperability i think we've seen quite a lot of backlash in particular on how to ensure interoperability while maintaining high level of privacy and security in particular with regard to end-to-end encryption and then finally i think a lot has been happening in online advertising and digital privacy cookies are being phased out by browsers new privacy preserving uh systems are being proposed google announced topic api basically an on-device mechanism for targeted advertising while meta and mozilla announced a privacy preserving attribution mechanism based on security party computation these are really interesting developments to me they really reflect the increasing social unacceptability of mass surveillance practices as in fact we're emphasizing professor sterling's recent paper basically aiming us aiming at making us click on ads and and we are so sensitive and identifiable these days and i think these initiatives because they really touch at the core of the advertising industry targeting and attribution are likely like the privacy sandbox before them to raise competition concerns and to require a very detailed technical analysis of the trade-offs and ultimately the choice being made so these were just basically five observations from the field that i've seen happening since the report that i thought i would you know throw in for discussion thank you and my apologies for the technical glitches thank you so much even example that was very interesting and now we conclude with pierre gibo and we started with a regulator we end up with a regulator as well well thank you for having me and of course everything i say is just my own opinion i don't know a single of my colleague who would trust me to toss a line so that's that's my home uh we all have agreed so far this is the need for regulation so what i'm going to do is talk about the specific regulation we're introducing the dsm and especially the dma and ask you know why are we introducing it and what is the logic behind the approach how do i think that we can implement it and what are the problems we are likely to run into well i mean in terms of why we need a regulation that's not that surprising you know in many other fields we've got some complementarity between competition law and regulation for a long time think about the pharma sector so that's that's really not new and we know the theoretical advantages of regulation you know ex ante what the requirements are so it contributes to what they call judicial or juridic certainty which by decrease uncertainty linked to investment and so on and also maybe the delays for review and decisions are kind of shorter especially than in antitrust where the delay can be very long as point out or whether that those comparisons must be careful because when you look at the delay in antitrust for example a big part of those delays come from the litigation process which in itself comes from the parties having rights and rights of the parties and the rights of the parties tend to be a bit less extensive under regulation than under the review of competition policy so there's no free lunch you gain something but you kind of you look something so beyond this kind of traditional advantage of regulation versus competition policy what are the rational one part of the dma is very simple we do not at the EU level have a very strong mandate to look at what is called unfair practices local authorities local competition authorities routinely if those unfair practices is part of the remit it's very vague under EU law whether we have this remit or not so introducing a regulation that can deal directly with those kind of unfair practices you know transparency in dealing with the people on your platform and this kind of stuff which really do not require the depths analysis of market structure or even behavior or effective behavior is just fair play that's something that really should be done through regulation and that that fills a hole if you want another thing is that you know we we do examine cases since then sometimes determine that in a given case a given type of conduct was undesirable could have anti-competitive effect or had anti-competitive effect but then what if somebody else does it in the same industry or another industry we have to run another case again in another case again right and they might even know that they're going to lose eventually but you know it takes a while for us each time to run a case so once we've looked at a given question for a number of time it seems efficient for all parties to kind of enshrine our previous review into a principle you cannot do this there's always of course a trade-off between time what did that do errors but when you've looked it's the same practice for a number of time I think the trade-off it says that you should move it to the regulatory realm okay so these are the obvious reasons there are however kind of deeper reasons why we also need something like the DMA and that has to do with the structure of competition law whereas you know our main two articles are one or one and one or two and both rely you know on dominance or at least significant market power well we know that you do not necessarily need to be dominant on have extremely significant market power for some conducts to actually create damage in the market especially if you are big in absolute term even if you don't have very very large market shares isn't a member of operation so in that sense it's again a compliment we're not saying the competition policy is wrong we're saying the competition policy has been designed to deal only with situation of very significant market power and of dominance and besides this is a range of situation where damage to the curve of the process can arise and we have to deal with those and that's where we get this concept of gatekeeper it's not defined in term of market share actually markets are barely defined in the DMA is defined in term of being big absolutely the european economy and being tentacular in the in the sense that your presence in a different set of sectors of additional activities and the very fact that you're big means that even if you're not dominant what you do by matter and this simplified by the fact that you're tentacular you can in fact not just one sector but many sector of activities so that in my view is a main reason for the DMA so also and theory reasons that have more to do with some difficulties that we face in enforcing 101 and 102 and in doing this I never want to suggest so I return to the DMA because we're not able to prove our cases through 101 and 102 I'm not saying that once we establish dominance we have a case we really don't like this conduct it smells bad but somehow we cannot satisfy the burden of proof this is not something we try to get around this not something we should get around so the burden of proof that we should uh that we should be faithful to but when we've run into trouble is in the definition of dominance itself and there are essentially two kind of troubles so first one is to do with market definition as you know the definition of dominance is actually the legal definition is not in term of market share but de facto for judges it's in term of market share which means that in order to argue dominance you need to compute markets and we already know that for economists it's not always you know that obvious as you define the market meaningfully where is a single relevant market whether you can define it in different ways all of the variable those problems I even worse is the digital world why because you've got those platforms that are decided there's more room for price discrimination different sellers have different information about the same kind of buyers all of those create really big difficulties to define markets in what we would think is a meaningful way so because of that we might not be able to easily prove dominance even if we feel that there is dominance there are other reasons like tipping we know that some markets with their work effects tip if you know that after you reach a market share of 25 percent the market is going to tip then it's probably fine and we should define dominance from 25 percent on so courts will not follow us on this then finally although we can in theory base a case on joint dominance this is an area where the courts have not followed us easily and that gives you some difficult situation because if you look at apple versus android for example they're each pretty big together they're certainly dominant but if you cannot argue joint dominance finding one of them you know a single dominance in all or majority of the european market is not this is going to be something easy so that's another more fundamental reason why we do need something like the dna now another interesting aspect to the dna goes back to what i told you about this this gatekeepers being tentacular and tentacular gets back to something we've discussed a lot in this conference which is this system competition this ecosystem right so i in my mind another very fundamental aspect of the dna is that it looks at the digital world at least the part of the digital world we're looking at with a big platform it says it's a strange landscape this landscape we really haven't seen that much before where you have a few constellations i call them death stars that are the system the center of the system whether it's a completely directed system or not might depend from one to the other but this is the center of systems and then floating in this space either smaller stars or you know atoms or molecules so people are very good at one thing and what we try to do is create an environment where both kind of models of business model both kind of players can compete on using an expression that you already use on a level playing field and that's in just sense that i personally understand the concept of constability which is at the core of the dma is to make sure that without certainly discouraging at all the behavior and the development of those big stars and actually encouraging them to compete with each other we also want to make sure that there's room in this space for the smaller fish to fry to to to to fry not fry they get fried we want them to just a little difference in words so that's that's that's kind of my general view of where of why we need to uh the dma now how is it likely to work well you guess this is good as mine but clearly we even don't know who's going to enforce it it's going to be some combination of probably dg connect and uh and dg comp and whoever gets kind of the central power on it there should be a dialogue between the two because we have different expertise so what do i see things i see things where really the dma can be applied without talking to dg comp or to the traditional part of dg comp and again this is kind of this unfair trading those things we've looked at again and again and again so also things that are fairly fairly simple and straightforward to apply look at the bundling usually i would say bundling that's a tough question that has to be looked at by conventional authority but here it's just been defined the only kind of bundling is bundling between core platform services and there i think you know these are such big things that finding a situation where you could credibly argue that uh compensating efficiencies to this kind of buddling would be hard so i'm fairly comfortable to let that be dealt with a dma without further references to us now let's move to things like interoperability we've talked about okay again it's both a technical issue but it's also an economic issue in the sense that it's a different kind of interoperability right you can have interoperability of data that's something i would encourage what does that mean that data would have to be kept in a number of preset format so that is a situation where they need to be exchanged and so on they can do so they can be done readily that's fine because this does not principles access to somebody's infrastructure also kind of interoperability does principles access somebody's infrastructure i can you know use my software with windows right so i get access to the windows infrastructure that's an economic issue because windows is going to be a trade-off between granting this access making sure that the build of the infrastructure structure is properly rewarded and also making sure that the others keep sufficient incentive to create their own infrastructure so that we also have infrastructure-based competition but that's that's pretty standard stuff and that's the kind of stuff that regular regulators have been used to dealing with well or not we can judge for a long time so is that too i'm fairly comfortable to let to the dma but when you get to things like you know self self preference we have one case that's seeming to be appeal that defines self preference in a very very specific situation i personally wouldn't want whoever gets charged with the dma dg connector a special part of dg gum to extrapolate directly from this one decision and say well a new case is going to be in this and this and this if they can find an agreement with the companies maybe but i think that in that case there will be a lot of payoff for having kind of further competition law case so look at this in detail final point does this all mean that there's never going to there's not going to be much enforcement of competition law as opposed to the dma is a digital sector so turn it first this is just a small part of the digital sector is going to increase a few years from now when we finally designate la post as as a gatekeeper we already planning on it so so so so so prepare your defense but we only look at one part of the digital world so all many other parts including you know ai algorithm and so on where the dma is not going to have any impact so in those parts of the digital world we're still going to be relying mostly on a competition policy we also see responsible of mergers there's really nothing about merger uh in the dma also the dma is really most it's about unilateral conducts it's not about you know collusion to finish the first thing that gets it's on and then finally i don't think the dma should be applied directly when there's a brand new conduct that's a brand new conduct and those will emerge directly because of technical change indirectly because you know once you make something and don't know people find way around it i think very new conducts should always start with an in-depth investigation and therefore under competition policy thank you thank you Pierre excellent so we had a great dream team and that showed uh have lots of questions including only well of course we want to know about La Poste as a gate keeper but also the you know interoperability and the other friend requirement for example contained in the dma going to work even that we know that friend doesn't work that way itself what can we do uh what about privacy which was mentioned by ivel exon on encryption there was an article in the ft two days ago on exactly this issue there are issues about the burner of proofs including for killer acquisition i guess there are issues about unfair practices and should they be left to national authorities there are issues about the location of of authority i mean basically dg converse dg connect and you have the same debate actually in the national authorities and for example the uk and many many more questions i mean this was very exciting so what rather than ask a question i suggest that you first we have a very brief sequence in which you basically comment on each other if you have questions about each other's uh presentation and then we open the discussion to the floor um so i don't know who wants to start for the first sequence yes i will i will answer the question of la poste as a gatekeeper um it is not our strategic objective however if it would happen we would be a very good gatekeeper why why firstly and i'm very kind to share it with you we are looking for profit we are not looking for the maximization of profit which is changing the behavior of a company secondly as having the responsibility of for public services mission we could be a good gatekeeper thirdly we don't try to develop our own measure of domination we are respecting the regulation environment and we be followed by regulation authorities kpi or criteria so that is why i'm saying that we would be a good gatekeeper and that is why i am very serene if it would happen one day so as you know that is that two two ways into being a gatekeeper one which is automatic based on numbers and there if you fulfill these numbers you have to declare yourself fine and frankly that's not that hard to monitor but not so many companies who are overreacting around those numbers uh if they don't declare themselves we will seize them for the rest these are mostly you know again numbers matter if you're not big but you're talking more about companies that are only really really big on one set of core services right think about amazon for example right and when there you have more of a quantitative uh judgment okay so again normally normally this is supposed to have declared but okay you know we do monitor the markets over time you know we do monitor markets looking for cartels which is much harder than looking for markets for our companies and that be you know close to the limit so that we can send them requests for information so that's what is that the way we would we would proceed if without you know intimations that they haven't done anything uh uh wrong send requests for information for to determine whether they've become a good kid or not maybe also on this point and especially on the x anti-regulation why is it so important because of the question of agility of the regulation if you are following traditional way of regulation you are always following the behaviors of the operators taking the x anti-regulation you have more agility to prevent any misconduct so I think that as everything is about speed you are able to be more efficient more quickly so I think it's something very important to have this type of regulation maybe a comment on this I generally agree but there is a trade-off in the sense uh having speed so that you get of orient the industry before this kind of bad behavior appear implies that you're pretty sure that those behaviors are bad right here so let's give access to the to the platform to Eva-Alexandre you have something to add uh yeah actually I wanted to jump in on your comments on privacy and security uh which is you know kind of what my my day job as a computer science researcher uh is um I basically see two things the first one is on data interoperability I think they they they have been a lot of talks that somehow data interoperability would run counter to GDPR and I do think that this is really a misunderstanding because again if done under the control of the individual you do have a very clear GDPR legal basis which is the consent of the person so really from a from a legal perspective I really do not see this to be an issue I think the the challenge uh when it comes to data interoperability and privacy is really going to be how do you prevent uh the issues we're seeing today with subject access request of how do you authenticate people how do you make sure that you do not give data access uh from one person to another person how do you prevent the system from being game or used for hacking into people's account um and how do you prevent you know another Cambridge Analytica uh how do you make sure that when the data is is transfers there is still some level of controls so that the individual can give consent for the data to be used but does not have to himself or herself go and check the details of every single website that's going to ask access to uh the data and that is a really tough question and the good is really to prevent you know another Cambridge Analytica uh but also again it's it's a tough one but Cambridge Analytica was also not prevented at the time since then we've been developing quite good you know solutions auditing and others to make sure that the data was used properly on full interoperability which I know to be kind of the traditional interoperability in competition and into an encryption uh I think you pointed this one out I think it's a really hard one I think we've seen as you said a lot of people you know taking to social media writing op-eds to basically complain that this was like you know another example of brassards not understanding technology even before the reading the text in details I think as we said back then in the report it's it's much harder it requires a level of standardization uh that you know goes much much much deeper than protocol and data interoperability some people seem to believe that you know a centralized you know some solutions might exist that would preserve and to an encryption fixes have been proposed but I think you know it's not guaranteed that they would keep the same level of privacy than what we have today with uh with internal encryption so this one is a is I think a really tough one uh and might require you know quite extensive standardization before we can do it we can do it right okay thank you I suggest we open the discussion to the floor so if you have question raise your hand and you can give your name too um even if you don't give your name that's all right I have another uh you you have also said that uh the royal practices remain at the level of the authority of the competition the authority of the competition produces sanctions or we could have imagined to have a regulation the most agile and the most flexible that it is to finally have a mechanism of different regulations to produce in a dynamic way of new rules according to events that can and practice new royals finally and as a result of producing rules that would adapt that would be implanted on the whole European market why this this type of tool hasn't been retained but I don't know if it hasn't been specified but I don't know so personally we're not disturbing at all to have the dma group whoever that is doing not only with the unfair practices that are recognized in the dma but we see update of these unfair practices I think that's part of the dma it can be updated without taking a trip back to competition policy since that European level right not the national level since that's a competence we barely had and never really exercised anyway Tony you have no question yeah so sure I mean we're going to speak among ourselves but oh here thank you so much okay thank you very much for the great discussion I have a question I don't know if it's a little bit aside of the of the main topic but in in in the context of interoperability there there's been a great discussion these days about also like the the limits of interoperability in in in for instance like freezing the state of the art of given technology for instance like the for the encryption or or that encryption or for other possible service services that somehow in my mind is also conflicting is conflicting with the fact that these are not symmetric regulation that mandates like obligation for gatekeepers but not for like competitors that have all the incentives to kind of differentiate and to invest and develop this kind of technologies even further and possibly also like keeping like this multi-homing service without making full benefit of the interoperability benefit that they are and I was wondering whether you have any kind of yeah my opinion or or or or what's about this discussion so you have this this question also about special obligation for the gatekeepers which are designated as gatekeepers yes and that is smaller one I mean that goes back to pierce a broader question about the treatment in DMA of course is going to designate some large gatekeepers but of course smaller firms can do lots of damage to I mean the case of most fever nation closes is a good example of behavior for which a big gatekeeper don't have can do not much they can do harm but as much as small gatekeepers so difference with a competition policy which encompasses everybody I mean here it's limited to actually it's closer to exalted regulation which allows you to get data and collect data but it raises the issue about who is big and who is small and there is a very thin line given that we don't want to I don't know maybe yeah you should answer that question sorry just let me just rephrase just a little bit it's when it was indeed whether small firms can actually be benefit in this beneficial in this case by developing on lines that are probably freezed by the standardization like if having a standard kind of impede the gatekeepers to even further develop those their technology on that encryption or other services then you have like these fringe competitors that might actually keep on doing that and I recently assist you to a great discussion about the limits of interoperability as harming innovation that seems like well debatable and I was wondering whether you had some it's a good concern but I don't think it's a good answer right interoperability in industry where technology is mature is certainly a very good remedy in industry where technology changes fast it depends on where does the technology change because in the industry you can have lots of changes in technology but you can still keep the same interfaces so even in digital what we would impose would only affect some type of innovation so maybe encryption where progressing to safeguarding the security on your platform might also make interoperability harder right but so I want to point out it's only for this specific type of innovation that the issue would arise and I don't know enough about what the technical constraints are there to see anything more than that thank you yeah maybe if I can jump in for a minute I think I agree with Pierre it's it's really I I don't find it extremely useful to discuss you know interoperability or even full interoperability as a as a binary thing as a like you know do we have or do we not have interoperability I think it's you know there's there's a very very broad spectrum of things you can do ways you can do it and and what you want to interoperate and what you allow each platform to keep doing on its own it's also a question of what do you freeze in the standard and how do you allow the standard to keep improving you know even in in telecom right we've not you know we've not been stuck back with 1g because they have been you know standardization effort and interoperability requirements so clearly I think encryption is a difficult one but again that does not mean that there isn't like you know a spectrum of things that one can do and in ways you can implement interoperability while still allowing for innovation but it really requires to to get down to the details and this question about friend and who is going to enforce access pricing I mean friend is a very vague thing that usually is enforced by courts but now we are going to get a regulator who is going to be able to do some access pricing also I'm not sure on what basis it will do that but that more journey raises the issue the other issues which was mentioned on this is now we we have to allocate between digi comb and digi connector my own preferences but it's clear that there is there's a lot of technical issues to be solved so we are going to have some kind of mix between the two but but still the choice of where it's okay it is not going to be an occur so I don't know if any of you have some comments on what should be done I can take the question I mean I can ask the question after you and yeah well so my name is Sara Lema I'm a PhD here at TSE and I wanted to have a kind of your feeling about you know do you think that the DMA is going to decrease the number of free services because in a way you know companies won't be able to justify anymore that you know oh but we have this free thing so we I mean for instance we have this free OS so to finance it we need to do this and that so do you think it will have kind of a long term impact on you know the number of free things that's a very hard question to to to answer is an economist because as you know in a platform everything is related right your practices of the advertising market the price that you charge on one side of platform or of the other if you ask me oh you do something drastic on one side of the platform what would happen on the other side I could answer but if you say look a bit in the way you deal with your platform you choose this one class even if it makes you less profitable it would not necessarily directly implies and then you would raise your price to consumer right because that says that the says that the reason this price is zero in the first place is not a matter of whether you have higher or lower profits either it's a matter of optimal structure of your different instrument on top of it when you see zero prices tell yourself that this is a corner solution you probably not just at the corner there's probably kind of room for things to change before you leave this corner so you know it's very speculative answer but based on this kind of reasoning I would say no I would not expect this regulation to have a huge effect on the amount of stuff that is there freely any other opinion on sarah's question yes it's better and the regulation will from my point of view once come to this free services issue it will come because there is something wrong in offering very vastly provided free services because somebody is paying it so I like your question I think should be an exam question this is a very so you use it as an exam and then with the student complaint you say oh you know you're from a classmate came in yeah but sarah has passed those exams a long time ago so next generation yeah since this is a postal economics conference I'd like to know more more about the challenges from amazon to the to that post and then what should be done to create a level playing field in fact it is very simple that this operator which is our first client so we like him has just the same obligation as ourself so the same service the same obligation that is clear respecting the same social and tax rules you know so you know I told you just before that we made significant progress I will tell you a story in October 2014 we had with other postal services CEO in Europe a meeting with the former commission European Commission and I was in charge to try to explain what was the same level playing field paradigm and the commissioner was very surprised because it was not an issue for him at this stage not an issue he saw us as old monopoly incumbent just asking for more protection which was not the case our case was not to be protected we don't have any monopoly today and everybody is switching from the letter to the email or to the sms we try in 2014 to explain that the same level playing field would mean or will mean that everybody is respecting the same rules and the same duties and I have to say that today with the DMA or with these future American law we have made significant progress because both the politician and the regulation authorities are trying to put the same duties and obligation to all the players in the market so for us it's a big progress we are not we don't need protection we are knowing how to adapt ourselves we just want the same level playing field which does mean everybody is respecting the competition rules am I clear and that is key that is totally key because when you are competing with a group as Amazon which is one of the most powerful in the world you need that they have the same duties and the same obligations that yourself that is a very simple demand and that is why I made this first conclusion we have made significant progress but it's not over sorry I don't have all the words in English but the Amazon don't have any postal obligations and they deliver packets exactly like La Poste and for instance we don't regulate Amazon we have a regulation on La Poste so I just want to check if you have something to be added to what was said not necessarily I think this is this is outside of my error of expertise we are tired after all those days of conferences you know last words the panelists yes maybe which is about challenge these digital platforms are bringing significant progress in both the economy and the society and the society they are stimulating us and they are in fact stimulating us to change that is very positive at the same time they are creating domination effect which are issues not only for La Poste or all the postal services but for the whole economy and I think that the regulation authorities are in the process to adapt their rules and their policy to these new economy and I am finding that very positive and for that I think that those authorities like us will need more economic and psychological research on the behavior of these actors so it's not only a business policy matter it is also a research matter as it is also a challenge for you as as academics professional that is my conclusion well thank you Philippe and on the behalf of the TSE platform I would like to thank the finalists and also all the participants to that conference that was a great panel I really appreciate very much and Elmut Kremer is going to say last words yeah I would like to just to say a few words of thanks there are the list of people to thank would be too long to to mention and so I will be brief first of all well I would like all of you to come and in spite of the difficult context it was nice to have a almost in-person conference for a change so this was very nice I thank the group La Poste for making this conference possible via their support I also thank specifically for the group La Poste Claire for being a very she was extremely helpful for both the scientific and practical part and so I don't know what I would have done without her and talking about female support I don't see Christelle but she's hiding I think but I would like to to mention her it has been this conference has been very demanding on her and so Claire works 24-7, Christelle had to work 2026 for making this conference possible and so she was very brilliant very efficient organized everything from A to Z and with the help of other assistants of TSC so Marie-Laine was also very active I should also point her and but I cannot point the whole list I should also point out our computer persons who are very helpful because in most places I've seen most places IT people usually live in their own world and they are not useful to the users but which is not the case here so they have been very helpful and worked out everything so I'm forgetting people I'm sorry but I really had to do this for the main persons with whom I wouldn't so for me the job was actually rather easy because everything was like an automatic pilot because of Claire and Christelle and so thank you very much and so thanks so much