Jon von Tetzchner (Vivaldi) and Laurent Ach (Qwant)

Bruce: Hello everybody, I’m Bruce Lawson and I’m a communications officer for the Vivaldi browser. I’m here today discussing generally how we can make the web better with the CEO of Vivaldi browser, Jon Von Tetzchner and the CTO of the Qwant search engine, Monsieur Laurent Ach. Gentlemen, Vivaldi’s headquartered in Oslo, Norway. And Laurent, you’re in France, I’m not sure which city, which city it is.

Laurent: Paris.

Bruce: Of course. Both cities Oslo and Paris are renowned across the world as the joint headquarters of fine cuisine and romance in Europe. As European organisations on the worldwide web, do you see that you have any qualitative differences with non-European competitors? Are there things that you do differently or you want to do differently or are required to do differently? Laurent, perhaps you could start.

Laurent: Yeah, sure. So I think there are some European values that exist and based on that, Europe is building different types of regulations. Of course, regulations are not everything we need. We need a lot of other things like investment and developing technologies and web business. But it’s very important and I think we have a kind of leader role in this domain. Because, of course, regulation might slow down a few things, but I think it’s like any rule actually. It’s like laws, it’s like anything. It should be reasonable, but we need them. And I think it’s a great thing that we managed to build that in Europe. We did that with GDPR and it has some impact, I think, worldwide, because other countries are doing similar things. We have other regulations like DMA and now AI, for instance. And I think all those things are based on values like privacy or the notion of risks of AI based on the consequences for society, for instance, and that. So it’s really a regulation that are based on some important values. So I think that that’s something to tap into account. Considering whether you work in Europe or outside. And then the other way around, of course, we have much less power financially than the US or the regions. But we have to deal with that. And this is what we didn’t prevent us from working on creating a search engine, which is kind of crazy, but we still do that.

Bruce: Jon, what are your thoughts on that?

Jon: Well, I think if you start, one of the things that I think we should be proud of is that when it comes to the web is actually invented in Europe. It was invented at CERN in Switzerland by Tim Berners-Lee. And that’s something that I’m really proud of. When it comes to us as an organisation, I mean, we are in Norway. And if you think about Norway, it’s a small country among a number of other countries in Europe, each with their own culture and language. And what that means is that you very quickly have to adapt to the rest of the world. You can’t just say, okay, I’ll do it my way.

If we tried to make a browser that was Norwegian, it wouldn’t really reach very far, would it? I mean, if it was Norwegian language only. And so in the early days, when I started building browsers, a lot of people were saying, hey, you only need to do English, because that’s initially how the web was. But we said, no, it’s not only going to be English, we should adapt to all the different languages and the cultures and the like. And I think one of the things that when you’re coming from a small country in Europe, or otherwise, you’re adapting to others and their needs. And I think that’s core to who we are as a company, that we should listen to our users, we should adapt to their needs and organise organisationally, a rather flat organisation structure.

I’ve travelled around the world, I’ve had offices in many parts of the world during the opera days. And what you notice is that there’s a much more hierarchical system in many parts of the world than what we are used to in Europe, where there’s a flatter structure. And I guess this is particularly within Nordics, but there’s a flatter structure. There’s no problem with people actually saying there’s a problem or communicating.

And I think that’s really important, again, that you’re able to communicate well inside the organisation. And I think it helps you build better products. I strongly believe that brains think better than a brain. I don’t believe in the strong man kind of leader concept for software. I mean, there are exceptions to this, but in general, I think it’s actually better when there’s more people contributing and more people coming with ideas and more people implementing them and that you work together as a team. I do think that the regulation that is mentioned as well is important. I think, obviously, I would like to see things go even further. I think we’ve let the internet kind of go astray a little bit. And it pains me because I think the concept of surveillance on the internet, I mean, that shouldn’t be there.

And I think a little bit of more regulation would help. But at least we’re seeing the regulators go after this. And I think they’ve done it the most in Europe. So that’s a really good thing. And hopefully that can continue.

Bruce: Yeah, to continue that theme, you ought to touch on something that Laurent mentioned earlier. We’re seeing some of the hardest, strongest, choose your word regulation here in the EU in Europe. I have to say that, folks, because I’m British and sadly not part of the EU anymore. But the EU’s Digital Markets Act has been pretty strident in setting down its views on how organizations, including the larger American organizations, should compete in Europe. And from people I’ve seen on the other side of the Atlantic, they’re saying, you know, well, this is limiting your competitiveness. There’s another argument that this allows us greater potential for differentiation, you know, that we are required to take more care of your privacy. We’re required to do less surveillance. So let’s make that a selling point. Does that ring a bell with either of you gentlemen?

Jon: In some ways, the Digital Markets Act is about competition. And there’s others that are more about the privacy side of things and the like. I think in some ways, the way I view it, you can try to talk about how you’re better when it comes to privacy. But I still think that we have an issue. And the issue is that even though everyone that was using Vivaldi was kind of getting better privacy, there’s still the problem with the data collection that’s happening online. And I really think the only way to fix that is regulation, because it needs to change the ecosystems.

And I don’t really understand how we ended up with the current situation. When I started working on the web all those years ago, I was kind of at the Norwegian telco, Telenor. And we were always talking about kind of how you couldn’t take any data that you got from customers and share it with anyone else. That’s just a big, big no-no. But we have now an industry where sharing data is a big part of that. And to me, it isn’t about the individual so much. It’s more about what impact does it have on society when people can be manipulated through profiling. So it’s something that for me is rather important. And I would like to see regulation fix that.

Obviously, we will market the fact that we’re better on privacy and the like. But I really think the solution is something else. And sometimes I have this horrible kind of way of describing this as if we are in a zombie movie. So you may keep yourself safe, but if the rest of the world is all zombies, it’s not great. And I don’t want to offend anyone with calling them zombies. That’s not the intention here. But the point is you want to make everyone safe and not just the users of your own product. It’s a bigger stake than just, okay, we want to get some more users using our product. It’s a bigger question than that.

And that’s really what I’m trying to say, is that we want to save everyone. We want to have a better world. We don’t just want to get more users. That’s not the core thing that I mean, for us, the question of privacy and regulation is not just some question about us. It’s about the world we live in, the fact that there’s a lot of data gathering that leads to many bad things.

Bruce: What’s your take on that, Laurent? Because as a search engine, of course, you know what people are interested in or potentially you could know, and collect stats or data on the fact that Bruce Lawson searched for zombies and werewolves and class A drugs – and that wouldn’t be a good thing. Do you feel that regulations required or do you think that you can do your product without collecting that stuff?

Laurent: Yeah, I think one of the most important notion in the regulations like GDPR and others is transparency. So it’s not like you it’s bad to use data. We need data to build a search engine. We need data. We need a lot of data. We need some users data. But we have to be transparent about what we are using on what for. And I think this is at the core of what people did when building these regulations. And so if you use some particular user’s data and especially if you sell them, they should really know. So transparency is really the most important value to me. And then of course, it’s better not to sell data to anybody from anybody. But at some point, if you for instance, are the revenues of pwn.com is mostly advertising. And so the less data you collect, the less money you get. So we don’t use much data from users. So we don’t get a lot of revenues from advertising.

But I think it’s also okay if you tell people, okay, for this service, we need more revenues through advertising, and we need more data from you and we tell you what data we would use. I think it’s okay. The most important transparency.

Bruce: It’s the honesty that you’re most valued like telling people, okay, for this service, you’re giving us this. But this is what we’re doing with it so people can make an informed choice.

Laurent: And the other important notion is what you use the data for. So it’s totally okay to use some pieces of data for fraud detection, for anti-bot detection, things like that. So the purpose, the goal of using the data is very important also. So if you don’t need to use a particular piece of data for some service, you shouldn’t use it. Especially if it’s just to sell it to someone else. So at the moment, we use really no data from the users for providing the search results. We don’t even know what the previous query was by this user. So it’s a big limitation. Maybe at some point we will do a bit more. We will use a bit more data. But as I said, what’s important to me is transparency about what we do and why we do it. Bruce: It’s interesting because you mentioned advertising. And of course, most search engines get revenue from advertising. And of course, most of the independent browsers (of which Vivaldi’s one) get money through deals with search engines where the money comes from advertising. And my salary comes indirectly from that as well. So the three of us around this table or this virtual table, none of us are against advertising per se. I think we’re primarily against intrusive advertising or surveillance advertising. But maybe you can speak to that because there’s been a lot of discussion lately about does surveillance-based advertising even work? Is it actually better, more cost effective than contextual ads in the old sense of, you know, if somebody’s on a book review page, you show them an ad for a book they might like. Is advertising necessarily evil? JON: I think if I may start a little bit. When it comes to like search, the ads that are on search results are mostly context sensitive. They’re related to what you’re searching for. And I think those ads, they generally from that perspective, they work well. And there as long as they’re relevant to your search, I think most people have been happy with that. Similarly, if you go to a tech site and you see tech ads and you go to a fashion site and you fashion ads, which is how things used to work, I think that’s acceptable to people.

From the early days of the internet, what people didn’t like was popups. I mean, which is why, I mean, we introduced a popup blocker and Opera as the first browser, I believe. And now every browser has that. And the concept basically of, okay, if you’re going to force it down my throat, that’s too much. So there’s a question of how extensive the ads are. So that’s not really acceptable to users. And I don’t think users either like ads that follow them, meaning that they know what they’ve done before.

And I think the user profiling is where it actually went wrong in many ways, with the concept of basically utilizing data from all your browsing history across different sites was then being used to show you ads. And I think that’s where it becomes unreasonable for people. So for me, when it comes to data, of course, if you’re using the data for the service in question, I don’t really think that tends to be a problem. I mean, you just need to be very clear about what the data has been used for in my humble opinion. And I don’t necessarily think that you should need to ask even; I think there’s reasonable use of data and unreasonable data use of data. And I think a lot of the popups that we are seeing, they maybe really shouldn’t be necessary, because if the data is not being used for any other purposes than running the service, and not being shared with anyone else, it shouldn’t be needed to show any kind of dialogue or the like.

So for me, it’s really it’s about the profiling of users. And I think it’s the surveillance economy. And I think I agree with you, Bruce, I do think that those don’t really work so well, because I mean, in some ways they work because people see them and click them. But on the other hand, I think it makes people really dislike ads. And I remember a time when I mean, it sounds weird today, but I would buy like computer magazines and some of the ads I found really interesting and would be reading into because they were relevant. I was reading a computer magazine, I saw related ads, I would read them, it’s fine. But I think now this concept of kind of feel feeling like you’re being followed, I think that’s unfortunate.

And I would hope that the advertisement industry also sees this and actually that context sensitive ads are a better way to do things. And I mean, that doesn’t mean that you can’t use things like maybe location and the like to kind of give certain context. But I think this idea that you’re bait seeing the ads on a long browser history, which by the way, we all know it often leads to ads that are not what you’d like to see; you buy a computer, you see an ad for that kind of computer the next month. And obviously the way they’re trying to fix that is getting a hold of your credit card information. So your credit card companies, at least in the US, they will share information on what you ended up buying, which doesn’t make it better. And this combination of utilizing Bluetooth beacons and the like to be able to follow you inside a store. Don’t think that’s great either.

So for me, you asked me one of the first questions, what was the first thing that you kind of liked? And I was kind of talking about my own site that I made. But in some ways, the big change that I noticed when I went onto the internet in the early days, was the fact that suddenly I had multiple newspapers. Up until that time, you would subscribe to one or two, typically. I mean, I was a kid, my family would subscribe to one or two newspapers. That’s the newspapers I would get. Suddenly I was able to go and visit all the newspapers. And I could read more or less all the stories, and they were for free. And the reason why they were for free is that obviously they were looking at the internet as this big and upcoming thing. But there were ads there that would be paying for this. And I think the model was actually working relatively well until the change of the ad model.

Because up until that time, this was called the place, to place your ads. You wanted to get tech audience, you would get them through a tech fight that would pay for the content that was there. But when the ads utilized the fact that you’ve been to that tech site to place tech ads on other sites of lower quality, that changes the business model. And I think it changed the business model to make it really great to create content at a very low price automatically with very little effort. And the price differences changed dramatically.

A lot of this is kind of technical, but I do believe that the change of the business model from context sensitive ads to surveillance based ads has had a lot of other consequences that people don’t necessarily realize. And I think reverting that would be a good thing. And sometimes reverting isn’t going back. It’s actually just realizing we took the wrong turn and we need to change and go back to the straight and narrow. And I think that’s what I would like to see.

Bruce: What about you, Laurent? How does Qwant square the circle between needing revenue from advertising but not being willing to participate in that data surveillance thing? What’s your take on that balance between advertising and privacy contextual versus surveillance ads?

Laurent: Yeah. I think advertising can be a part of the results of a search engine when it’s really irrelevant. It can also be annoying. And as a developer of a search engine, there is another thing we have to take care of, which is advertisement from the website we index. So because when we crawl the web and we gather a document from the different pages from many websites, the content of the pages should be interesting to the users. And if it’s advertising only or just maximizing traffic, we shouldn’t even put them in the index. So it’s also something very and a very important topic when we crawl the web, not only when we use advertising for revenues. Bruce: This allows me to quickly segue into a less political and more techy subject area, because I’ve never spoken to anybody who actually runs a search engine before. So presumably you have legions of servers looking at web pages all the time and ranking them and working out what they’re talking about, etc. And I’m not going to ask you for your secret algorithm to determine which website is better, but how do you even go about achieving that? Because the web is so vast and so many sites change so quickly. And we know from having ad blocking in the browser, there’s a constant arms race between even identifying ads and ad snooping by, how do you even manage to keep searching relevant and how do you even manage to keep crawling the right stuff?

Laurent: So we use various techniques for that, different in different parts of the search engine. So the first thing we have to do is to crawl the web. That means following links to find documents. And so at this very initial point, already we can select what we start from. So we have what we call seeds. So we gather more or less manually. It’s not really manually because it’s already millions of pages, but we have like a database of seeds, which are important websites, useful websites that are well known. And this is our seed for crawling. So starting from that, we can crawl through the links in these pages. So it’s already a way of having some quality in the document. Then the further we go, more random things we find.

And so we have to use other techniques to ensure the quality of the pages. And so there are different features we can use for that and machine learning models we train for that. A very well known techniques comes from the early days of Google. It’s called the PageRank. It’s still useful. It’s a way of evaluating how important a website is based on the different links pointing to it. So it’s a useful feature. There are many variants of that that we use. Then we can use the history of what we saw from the users on our website. Like has the page been clicked one day or not, and based on which query. So we can use that. We can evaluate the quality of the page with using machine learning models just by gathering different input features from the document.

And the first thing we have to do, which is very important, is to extract the HTML content. So the pages we scroll are mostly in HTML. And it’s not so obvious to extract the relevant content from it. And the different part, what is the title and the content and the data. And which part of the content is interesting. Which links are useful to follow. So for every part of the document, we do some machine learning. And we evaluate the quality. We try to detect spam and phishing. So we really use a bunch of different techniques to select the document we’re indexing. And then comes the part of using that index. And this is another story of it. Bruce: How do you determine your initial seeds? Because it seems to me, for example, like if I don’t know, I’m making up this example, but if you had 10% of your seeds were in the Hindi language or in Mandarin Chinese, you would, because those are the seeds from which you began crawling, you might see far more Hindi or Mandarin pages in the search results, which would be very, very useful for Hindi and Mandarin speakers (but not so useful for me, who being British can barely speak one language). So it strikes me that the initial choice of the seeds is the most important starting point. How do you choose them?

Laurent: So it’s, so the direct answer is mostly most visited websites. And based on known statistics, or based on our own statistics with our users. So that’s the simplest answer. Then what you’re suggesting is that there is a bias at the beginning, which is totally true.

Bruce: No, I’m not suggesting that.

Laurent: No, I don’t think it’s me. But there are a lot of biases everywhere. It’s very difficult to come up with very neutral results. You cannot avoid bias in general, but you can try to limit them. And so, as I said, using statistics, it’s a way of selecting the website. It doesn’t mean those websites are the best ones, but it’s a starting point. So we just need to start from something.

And then we have many other, the other techniques I mentioned to select what we will put in our index. But our goal is to gather a maximum of useful pages in the index. At the moment, we mostly select documents in French. And so we will, following the links, if we are good at selecting the useful links and starting from quite good website as a seed, we will eventually find all the interesting websites.

Bruce: I’m interested to know, obviously, I’m not going to ask you what your initial seeds are because that would be a trade secret. But there must be a point which you follow all the links and then it stops. So what is the URL of the end of the internet?

Laurent: So there is no end. We can’t crawl forever. So we can gather billions of pages. The number of useful pages is not so big. It’s big, but it’s a portion of the total number of pages. So yeah, it’s a problem we have to deal with: when we stop? But the most important is how we select the document and the links. There are a lot of links in the document, sometimes 100. Some of them are not interesting at all. And then, so one question is how many documents we need in the index, but then how to retrieve them and render them, which is even more important because you can have billions of documents that you will never find in your results or that are bad quality and you you don’t want them in the result. So retrieving and sorting, ranking the documents, then that are part of the search engine, which is very important.

Bruce: Yeah, I guess if you’ve got a massive database of billions of documents, but it took four hours to search it, it’s not a useful product. So you’ve got that end as well.

Laurent: Yeah, it’s not only a matter of how long it takes, it’s also the quality of the document. So there is a trade-off between the number of documents and the coverage of what we want to find in the results and also the level of quality in the index and how you’re able to deal with different, with documents of different qualities when you have to retrieve them based on the query and rank them. And at every stage, there is quite a lot of machine learning. Bruce: Fascinating. And, Jon, now I’m going to ask you to give away some technical secret sauce, given that Laurent’s told me how a search engine runs. So Vivaldi has a built-in tracker blocker and ad blocker. And in the same way that I mentioned to Laurent, of course, there’s a constant arms race between people who want to identify ads in order to remove them and people who want to show ads and not be identified. How does the Vivaldi tracker and ad blocker work in order to make stuff better when it’s presented to the user? You don’t have to tell us everything, just the super secrets.

JON: I guess I wish I had Julien with me or someone else to kind of go into the more technical details of this. But obviously, a lot of this is about basically blocking connections to known trackers, to block, should we say, known advertisement sites and links and the like. A lot of it is about basically having a link of items and then you go to the next level thereafter. But a lot of it is basically a list of known content that is being blocked. And then it gets to be more complicated thereafter with regards to potentially removing parts of, not only removing the content, but removing any placeholders and the like. And that’s where it gets to be quite a lot more complicated.

So we’re basing our lists. I mean, we’re using lists from Duck Duck Go and from Adblock Plus. We’re working on kind of expanding that to include more of the Ublock blocker lists and the like and being able to handle more of those. But a lot of it is basically working with lists and blocking content based on that. It’s a fair amount of work. And then like you said, there’s a little bit of an arms race at times. On the other hand, I also think that it’s actually, sometimes I feel that we end up blocking too much. Because again, depending on what you would like to see, if you’re actually, for example, saying, okay, I’d just like to reduce the tracking. I don’t mind the ads. And then the lists actually block more than you would like to see blocked from that perspective.

So it’s complicated getting it right. And because at least for me, I do believe that ads have a purpose on the internet, because the consequence of not having them is less free quality content. You may have still a lot of content that’s available.

So it’s all about a balance that I find really hard to deal with. And we’re trying to find that balance. And I think a lot of others are also trying to find that balance. Because again, not all of us, and I’m one of those, I don’t mind seeing ads. I would just like to see less tracking. When I mean say tracking, I mean profiling of me as a person. If someone clicks an ad and you can track that that happened, that isn’t necessarily a bad thing.

Bruce: You say it’s hard to find the balance. And I think once you’ve found that balance, it’s balanced for precisely 39 seconds, and then somebody invents something new, and you have to recalibrate, which brings me on, (because I know I’ve taken a lot of your time, gentlemen) brings me on to the last thing on, to touch on. Obviously, in tech, things move fast. We live in, we work in very iterative industries. But how important is it for Qwant -and then subsequently I’ll ask Vivaldi- how important is your user, your customer feedback? Do you have a process by which you solicit it, act upon it, iterate it back into your products?

Laurent: Yeah, so for search engine, the goal is to show relevant results. So relevancy is only defined by, can only be evaluated eventually by a user. There is no way to evaluate the relevancy in a very objective manner without users. We have many ways to do it, using proxies. So for instance, we can consider that if a user sends a query and click on a particular URL in the result page, it means the page is relevant. So it may not be, because usually people just click on the first result, for instance, or there is a lot of noise in this data. So it’s a signal we can use.

And we can also compare our results with other search engines, for instance. So we have many ways to evaluate how relevant a result is based on a query. But the final evaluation has to be done by an actual person, a user. So we can do some of that internally within our team. But at some point, we need actual users because that’s a result of bias in what kind of users we are. So at some point, we need user feedback from many users. And there is still a bias because people who answer, who give feedback are not any user, but you always have some kind of bias anywhere. But still, this is a very useful information.

So we have ways for users to give feedback on the website, especially when we were in a beta test phase. We also have a particular interviews with some users at some point. And we have chat, like Discord groups, to discuss about the product. So we have many ways of doing the evaluation of the quality of our search engine, which are totally automatic. And but at the last stage, we need the users.

Bruce: Actual people telling you.

Laurent: Actual people, yeah. Human beings… preferably. So we can also use a Generative AI for that, because it’s also a proxy for having a kind of something looking like user human feedback. And until some point, it can be used for creating data sets, for instance. We use that. But again, this is not real users. So at some point, we need them.

Bruce: And then, Jon, what about Vivaldi? How does the Vivaldi find out what its users think and, more importantly, act upon the user’s thoughts?

JON: Well, I mean, we have a rather special organization when it comes to our community. And I mean, we have a very strong bond with our user base. We have people that are volunteers for the company, helping us test, helping us translate the application, helping us spread the word. And it’s very special. And I mean, it makes you kind of humble and thankful because without them, we wouldn’t be able to build the products that we do. We get constant feedback. We get the feedback through our bug tracking systems, through our forums, through when we have blogs and releases. There are tens of thousands running test builds as their primary browser even. And they give us feedback about what they like, what they don’t like.

We don’t collect information from users. So we’re not tracking how the user is using the browser. So it’s all based on the feedback that we get from users. And what we find is that people have different opinions, which is why we end up having a lot of options and a lot of different ways to do things. Because again, if you look at how the different browsers look, because a lot of our users are sharing pictures of their browser, they look like totally different applications. Because again, you can customize it so much. And people do, they put the tabs on the left, right, top, bottom, they have different backgrounds, they assume the interface, they go without an interface. I mean, there’s a lot of variations on this. And our job is really to listen to their feedback. And in our forums, people can vote up certain features. And in a way, the organization is kind of sometimes it’s unclear almost where the organization ends and the user starts. And sometimes obviously, we also have users that are volunteers that become employees. We have also former employees that become volunteers. So it goes both ways.

And that’s one of the things that I’m really proud of, that the fact that people that have worked inside the organization, or inside my former organization, Opera, that they’re actually willing to be volunteers and help us improve the product. So for me, this is the reason for us to exist as a company, is our community. And our job is to listen to them, do our best to give them what they want. And it’s for me, it’s the reason why we are doing this. It’s what makes it fun. It’s what makes it so giving is to listen to what they have to say and give them what they want. Now, sometimes the feedback we get, if we do something, if we break something, sometimes they’re a little bit angry, but normally it doesn’t last for a long time as long as we listen. And we learn from that. Again, if we put out something that they didn’t like or change that we did at the very least, we’ll put in a way for them to get it the way they want, because we respect that there is a difference in individuals.

And I think, again, this relates back to who we are as a company, coming from a small country in Europe. We can’t tell everyone how things are supposed to work. We do our best to build a great experience. And then we make it so flexible that no matter how you want it to work, it will work for you. And that includes, okay, we put in kind of keyboard shortcuts, we put in mouse gestures, we put in quick commands, we do all those things which are maybe not something that everyone wants to use. But the people that do use it, they love it. And that’s some of those are the reason why they’re using Vivaldi, because there’s this thing that we did for them that they really love. And getting that feedback, and when people are happy with what we’re doing, it’s this amazing feeling for all of us. And I know it really makes the developers stay when they get that kind of feedback that they did something and people loved it. And again, they hate getting the negative feedback, but it’s great to get it as early as possible so we can improve on things, because that’s what we’re always trying to do is just to improve and do the right thing and do everyone, I mean, seeing everyone as a user, seeing everyone based on their needs. And so the community is just a core part of who we are as a company.

Bruce: Lovely. Very briefly, Laurent, you mentioned that there’s lots of new stuff coming up in Qwant. What can we expect coming up in the near future? Can you give us any sneak previews of what we can see over the horizon?

Laurent: So one thing that already went to production is, as I said, the short answer using the generative AI, which is very well integrated. And we’re working with, so maybe you heard that we were acquired recently by the founders of OVH, which is a cloud provider in France. So we work together also to use language models that are hosted in France. So that’s part of the work we do to develop European solutions for different services. So this is one thing that is already available in production. But the main thing we’re working on at the moment is the core of the search engine, because until now in the history of Qwant, we have never been able until today to be really autonomous with our own technology. We are using Bing, the Microsoft search engine. Now our goal for the next months is to come up with our own, we already have our own technology, but to bring it at the level that we can use it 100% for all requests in France, for French documents to start with, and then in the world.

And so that’s a huge effort because we are one of the few companies in the world building our own technology for a search engine. And so that will be very useful to ourselves, but to other alternative search engines, because there are a few of them in the world, and very good ones. And so we’re also working with some people who are interested in that.

Bruce: Fascinating. At which point, gentlemen, I shall let you go back to the important work you do.

I want to say, A, thank you to our listeners. B, thank you very much to Laurent Ach, who’s the CTO of Qwant, the search engine, and to Jon von Tetzchner, who’s the CEO of Vivaldi. Thank you both for running companies that offer us an alternative to big tech and therefore helping make the web a better place. Appreciate your time. Thank you, gentlemen.