podcasts

How to beat Surveillance Capitalism, Brendan Eich of Brave

podcasts

How to Beat Surveillance Capitalism, Brendan Eich of Brave

June 2020

Posted by

Jamie Burke

CEO and Founder

As an early investor in Bitcoin and Ether Jamie went ‘all in’ during 2013 founding Outlier Ventures, Europe’s 1st venture fund and platform dedicated to blockchain and Web 3....read more

Brendan is Founder and CEO of Brave Browser. As the creator of JavaScript at Netscape in the 90’s, he talks about how it’s success and ubiquity alongside cookies and behavioural tracking led to the business model of Surveillance Capitalism that has come to dominate The Web. 

Posted by Jamie Burke - June 2020

June 2020

Posted by

Jamie Burke

CEO and Founder

As an early investor in Bitcoin and Ether Jamie went ‘all in’ during 2013 founding Outlier Ventures, Europe’s 1st venture fund and platform dedicated to blockchain and Web 3....read more

We discuss how the industrialisation of the Ad Tech ecosystem, and its layers of intermediation, has not only led to erosion of user privacy but also a poor web experience and how Brave’s Basic Attention Token incentivises a better outcome for publishers, users and brands.

Listen on iTunes 

Transcript:

Welcome to the Founders of Web 3 series by Outlier Ventures and me Your host Jamie Burke. Together we’re going to meet the entrepreneurs the backers and the leading policymakers that are shaping Web 3. Together we’re going to try to define what is Web 3, explore its nuances and understand the mission and purpose the drive founders. If you enjoy what you hear, please do subscribe, rate and share your feedback to help us reach as many people as possible with the important mission that is Web 3. 

Today, I’m really happy to welcome Brendan Eich, Founder and CEO of Brave browser, basic attention token creator of JavaScript and Co-Founder of Mozilla Firefox, all round web pioneer. Welcome, Brendan.

Brendan: Thanks for having me.

Jamie: So whilst it Netscape, you famously built JavaScript in 10 days, which is now for better or worse, ubiquitous across the web. Alongside HTML. And CSS. JavaScript is one of the kind of cool technologies that drive the web and enable much of the interactivity that people just take for granted. For the benefit of listeners, including loading new page content without reloading the page, animation, interactive content, like games, videos, web forms. And of course, I guess the thing that you’re most concerned about and trying to resolve or solve for now, which is users behaviour, analytics, tracking, and this kind of personalization through cookies. So in a way, it’s the success of many of your innovations and perhaps the unintended consequences that have brought about Brave. So I’m really looking forward to exploring your journey as a web pioneer, and going a bit deeper into Brave in this mission.

Brendan: Yes. So you know, back in the 90s, with Netscape we were trying to make the web easy to use for average folks. Earlier browsers, notably NCSA mosaic had already started taking off and people could see the value of unifying a sort of of information. Under this, click on a blue link and get a new page model and form elements come in NCSA mosaic under Marc Andreessen. Eric Bina added the image element in 1993. Netscape in its first year, I think, for lumen two, he added the cookie, which is somewhat Miss named I think it was inspired by another use of cookie the magic cookie, which comes from Unix and it describes the first few bytes of the traditional. First four bytes of a file a binary file format, which can contains a number that should be unlikely to occur randomly. And that can be used to identify the file type the cookie on the web, the browser based cookie is a bit of storage, one sort of blob of storage can be fairly big. I forget the current limits. But back then we considered anything that was like over a kilobyte big or 8k, certainly big per, let’s say website, if you think of it that way. There’s a more technical limit to it, that equates all the sort of top level plus one domains. So you know, bbc.co.uk, or NY Times.com would both be identifiers for cookie jars, or cookies really, and the cookie is associated with that domain, whether the content is the page you’re looking at, or something embedded inside the page. And that means that the cookie gets stored based on what the server sent, and it gets sent back to the server on subsequent fetches, so that you can have the server updated based on its previous value. So the server doesn’t have to keep track of you through some other means because there really isn’t any other means to do it and or there wasn’t been any way in HTTP. It was so called stateless, which meant, you go to your bank and you log in using a form, the page that authenticates you is the only page that knows about you, if you go to a different page, you’re gonna have to log in again, which would be a bit of a drag, so cookies relieved users. So having to do that, just as embedded images for lead you from having to get a plugin of some sort, or go to a full page image. So these innovations were meant for good reason to make the web more usable by average people, but they had unintended consequences even before I did JavaScript. The combination of the image and the cookie if you digest what I just told you, it makes tracking possible. You can have what can be called a pixel because it was used A small image you wouldn’t notice maybe it was transparent or translucent, semi opaque and it was one by one pixels on your screen, it was somewhere tucked away on the age. And it was embedded on two different publishers sites. So let’s say NewYorkTimes.com and espn.com. But the same pixel was being fetched from a tracking service that hosted that that little image, tiny little image, well, then, through cookies, that tracking server could see that you’re a new user on the New York Times, it hasn’t got a cookie for you. it assigns you an id say 1234. And it sets that in the cookie for the pixel at its web address. And the way it’s embedded is not just the address tracking server.Com. Let’s say it’s got a further part to the right in the URL, the web address and that part can be used to identify that it’s a New York Times embedding of this image. But when you go to ESPN, the same image, possibly with a different decoration on the end of the link to identify that it’s on ESPN is fetched. And because it’s fetched from the same tracking server.com Well, if there’s a cookie that’ll be sent back, if there’s not a cookie, the tracking server will assign a new user ID. But since you went to New York Times, first, you’re known as user 1234. And now after visiting ESPN tracking server knows you’ve gone to New York Times, and then ESPN and it can build a dossier on you. And this is done all over the web, you know, to cut to the chase JavaScript added fuel to the fire, and made it possible to get rid of the image altogether, you can just use scripts, which are also embedded elements that can be fetched and can set cookies. So now with this sort of torque of image cookie in JavaScript, the whole sort of primitive fundamental ad tracking and surveillance and analytics as you say was was ready. It took a little while.just realised I remember I was at Netscape in 95, doing JavaScript in a hurry, sometime in late 96. I think, Lou, until he wanted to get rid of third party cookies, the so called embedded ones, the ones that aren’t set by the whole HTML page, you’re looking at the New York Times news article, but rather by the tracking server, with its little element hiding there.And it was too late. It a lot of things that get out of the lab onto the web, it gets such wide adoption that they set a de facto standard. This happens all the time. Marc Andreessen told me that when they had only 80 servers on the early web, and they made some sort of change to mosaic that threatened to break some of the content, they quickly reverted that change because they realised that it had become a de facto standard, even with a what we would now regard as a small population. And it would have been easier to change them anytime after but even then, they didn’t want to change it because of the incentive to not break the web as people say Don’t break the web.

So you know, a lot of these innovations once out, could not be recalled the genie was out of the bottle. And tracking was done in the 90s. It was not widely understood how it worked, I think certainly not like the average users, the bulk of users, but you know, there were companies doing it. There were a lot of ads that were just banner ads, and they were like, takeover or sponsorship, they didn’t have any targeting capability. But as the tracking powers built up their their expertise with cookies and scripts, they started compiling dossiers on on people, I think the ad server that became double click, which Google bought in 2008, was started in the late 90s, early 2000. I forget the exact date. And it was called the dash server, I believe at first, and it was one of the early ones that I think did tracking. There were others. There are people who sort of mixed this up with advertising. They say advertising was the primal sin of the web because it conditioned people to expect free content. I take a different position. I think tracking was the problem. And if you have a pure sponsorship ad or an ad that’s getting you to contribute to a charity and nothing happens to you click on it. And if you change your mind and don’t fill out the form and donate money, nothing bad happens, then I think there’s nothing wrong with that kind of sponsorship ad or whatever you want to call it.

So you have to separate ads from tracking. It’s as if you went to your local, you know, grocery store, and instead of, you know, flyers on stands or at the door, the flyers detached and became little paper drones that follow you home, it’s spied for your windows and did things to you. So, you know, it’s not the whatever the promotion is on the flyer that’s bad. It’s not the ad as a call to action or a piece of brand. collateral, it’s really the tracking. And so, you know, long story short, by the time I’d say maybe 2006. Firefox have come out and we pioneered extensions, we call them add ons. Eventually the the variant, the chrome supports sort of swept the field. And everybody who’s doing extensions now kind of supports that, because that’s where most of the extensions are in the Chrome Web Store. But Firefox, at first had add ons and still has quite a few. So it’s deprecated, or retired it’s old extension model. But when we built that extension model, we allow for things like turning off JavaScript across almost all pages. That’s the famous no script add on. And I think that was number one at the time. And then ad blocking was done as an extension. It was originally, I believe, called adblock. And then there was some fork and it became Adblock Plus, that went from being a Firefox add on to being an extension on all browsers, and also having its own browser on mobile on Android where for some non coincidental reason, the Chrome browser doesn’t allow any extension. So you can’t have an ad blocking extension on Chrome for Android. So users liked ad blocking publishers hated it. And this started a contentious period, there was even litigation in Germany. I think the maker of Adblock Plus IO won almost every case, every case in, in summary, and or on appeal. And users have a right to do this when you’re using the web by design for accessibility if you’re handicapped, or if you’re on a limited bandwidth device or using an old terminal, where you use a text browser, you’re allowed to take pieces of the content that are served, provided you have authorization to get to them and you know, most of the web is freely open. You don’t have to log in at all to get to it. You can download pieces of it and block other pieces from being downloaded. You can mash up things you download, you can make your own local combinations, do your own local edits, if you wish. Now there’s copyright and trademark law would bind you from publishing your changes to somebody else’s material if they have copyright and so on. But the web by design allows the browser user to mix and match and block and this was pretty well settled in not just Germany with IO, but in other jurisdictions. and so ad blocking took off. And publishers hated it because they thought they were free riders out there denying them revenue. And they were right. But the user has the right to do it. I think you have to respect that. publishers have been under pressure, especially the newspaper business for decades from, you know, prior technological shifts, but didn’t get an easier when Craigslist took down classified advertising revenue. And then, you know, the rise of Google and Facebook is, I think, widely cited. And I think this is supported, well supported as damaging publisher revenue. Publishers had so many, you know, abuses and insults to their revenue that they didn’t like ad blocking. And for a time, they thought not only lately, but sort of with users, they would put up anti adblock dialogues. When you try to go to their site, they’d say I see you running an ad blocker, you must turn it off, or subscribe. Generally overpaying for content when you only want one article and don’t have good options for buying just one article, or, you know, go away. We don’t we don’t like you. I think wire did this and a bunch of other sites did start in 2015. And it generally failed users just went elsewhere and the further commodification of news meant they could find stories elsewhere. And Alexa ranks of all these sites like wired the tried this sort of user hostile banner against ad blocking went down. So you know, there was there was a an opportunity there, I think Brave we started in 2015. And I’ve already seen it in Mozilla from the rise of adblock, and Adblock Plus as extensions as Firefox add ons. I seen the the opportunity by the Mozilla we didn’t do much about tracking we were busy, I think growing Firefox and restarting standards bodies. In hindsight, I regret this.

And over time, you know, with Google brain chrome out, it got harder to try to change the defaults to block tracking because sometimes he would break a site that had tied some inappropriate knowledge. So that if you blocked tracking the site malfunctions, if you blocked tracking in that blocks, ads, the site malfunctions. And so it was hard. I think for browsers that hadn’t done anything from the start to start doing things now. Apple Safari came out a year before Firefox one Oh, there was a earlier version of Firefox at the same time, but we didn’t do one oh till November 2004. And 2003. Safari came out and it had a third party cookie blocker that kind of blocks those cookies set by those embedded elements like scripts and images that can track you. So Apple got it right from the start, relatively speaking. Now, they had a loophole in that cookie blocker in that if the domain trying to set the cookie as an embedded element, had already successfully set a cookie as a page you visited so called first party cookie, then it was allowed to track you. And so you know, the advertising companies, some of which you may never have heard of, such as app Nexus figured this out. And they would cleverly do things where if they got the chance to handle it. cliques and MCs were accidental. Or if they got the publisher for whom they were serving ads or delivering ads to arrange it, while clicking on the ad or otherwise interacting with the publisher might take you to App Nexus com as a full first party page very briefly. And that would set an app Nexus com cookie and Safari. And then thereafter they were whitelisted for tracking. And this is called a redirect bounce because of course, you didn’t want to go to App Nexus. So as soon as you landed there, and they set the cookie, they immediately redirect you back to wherever it was, you were supposed to be going. Whether it was another page on the first party, the publisher or it was the ad you clicked and you wanted to get to some full page ad unit, and that that loophole, once it was known, got exploited. So my friends at Apple came up with intelligent tracking prevention few years back, I forget when they started it, I think it was after we started Brave. But Brave also had benefit of, you know, some really great work in other extensions, I’ll call out you block origin, which is another one of these

Really tracking an ad blocking tracking prevention and ad blocking extensions. And so we’d seen that if someone was diligent, they could keep up with the adversary who was trying to track users. And also try to unbreak sites that had these sort of accidental malfunctions when you did block tracking or ads. So that was informative to Brave. And since I was starting a new browser, I thought, this is the time to do something for privacy. I’m annoyed by these things. I don’t like the feeling of tracking. Just retargeting which is trying to not only compile a dossier on you, but decide whether you’re in the mood to get shoe ads. Maybe you just bought shoes the other day at the physical store, but they’re going to show you shoe ads until you know for 30 days, and that’s annoying, it can actually be worse than annoying can be abusive. It can be used by bad actors. It can certainly spoil a birthday surprise for somebody who walks in, if you’re being retargeted for the gift you bought the previous night especially if they’re children, and you know users have rights to protect themselves. But the really nefarious stuff that we noted when we launched Brave in early 2016 was front page news on all the sort of internet news sites. You had, I think in early March malvert, icing ransomware hidden inside ad JavaScript that was being loaded as if it were an ad was being placed through an Ad Exchange, it was getting into the New York Times, I guarantee you, The New York Times did not contract directly or do direct sales of its ad space with a malware vendor. The way this happened was through the what’s called the loom escape, or the ad tech ecosystem that evolved from those fundamental building blocks of the image cookie in JavaScript. And this is an amazing ecosystem. It mediates between the publishers who have ad space, what’s sometimes called inventory and more important they have audience which are the people that attract not only when they’re on that publisher site, but across all other sites, we’ll get back to that mediates between the publishers ad inventory. And the brands and the marketing agencies that help the brands put ads into those ad slots. That can be done with direct sales and the biggest sites do it. And that gives them a better margin and higher quality control and other things. So they have to have a Salesforce, they have to have their own tech team that can do the some of the ad serving integration. Turns out a lot of them use Google. Double click, which I mentioned already, even for direct sold ads. But if you don’t have scale, and there’s, you know, millions and millions of websites, and you end up not being able to afford direct sales, you can’t you don’t have the scale to attract the brands anyway. You can’t afford a sales team. What do you do? Well, you’re told to go. I’ll use Google as an example. There are many go to the Google, I guess it’s called Google Tag Manager, Google publisher tags now, I used to call it double click all the time or double for publishers DFP. They’ll go to that site and they’ll say, Hey, here’s a bit of JavaScript copy this and paste it into your head of your HTML document. And, you know, Bob’s your uncle, now you start getting paid. And it is kind of magical. Because it’s not a lot of work, though publishers often get it wrong or they aren’t super technical. So it’s even pasting a little JavaScript thing can be a challenge. But assuming that they have people visiting their site, and Google sees that, Google then can arrange. And again, I’m picking on Google, but there are many companies that do this school can then arrange to make a match between the brands trying to put ads for goods and services in front of people and the ad space that the publisher made available, either, you know, by configuring this little script to paste it in or doing something in its dashboard or doing some extra work in the HTML there lots of ways to do it. And this evolved into such a complex system that you have layers of intermediation, you don’t just have the ad server, or the Ad Exchange, let’s call it that. Google is the premier example of in the middle you have things like supply side platforms, which are these entities that help publishers or supposedly help publishers get better yield get better ad revenue from sort of middle or bottom tier slots. You have innovative companies like header bitters, which may be still around, they certainly had a heyday a few years back, where, even though you’re using, let’s say, Google for your main ad, revenue as a publisher, ahead or better a company like Sanofi, which I advised, can come in and say, put a little extra JavaScript in the head of the document. And we’ll take a first look, as soon as the page starts loading at the user. And if we through our tracking cookie, and so on, can can get a better ad deal than your main ad partner will, will take over the ad delivery will even reprogram in the page, the Google ad server that you’ve set up to do your main deals with, and will preempt those deals and do our own deals and will give you a better yield. And so you had sort of this mutualism, or parasitism on top of the main ad server by a head or bitter. The other side of the Ad Exchange, you have sort of data management platforms. Some people call the brand or advertiser facing side, the demand side platform. Google really is kind of the combination of the Ad Exchange and the demand side platform. And they kind of tie it all up into a nice package. But there are companies that do individual pieces of these functions individual three letter acronyms like DMP DSP, SSP supply side platform, and it becomes this confusing alphabet soup of intermediaries. When you’ve got too many intermediaries. You’ve got problems, the fees are high. There’s a lack of transparency like you don’t even know necessarily if you’re getting paid properly this this has happened supply side platforms were essentially cheating publishers by I would say lying about how much they were making and they would take a bigger feed and they promised publishers could find this out by by buying some of their own ad space. I think the Guardian famously just bought out its own ad inventory over a month and found they were making 30 pence on the pound. So 70% of the gross adspend, which comes from the marketers, again, they’re paying for the audience or the ad space, really modern way of talking about it is the audience 70% of that gross ad spend was being taken out by intermediaries. And so you know, that’s too high a fee. There’s not transparency, you could be being cheated. Other bad things happen. I mentioned malware we want spray that helped us grow because people started realising there’s a real threat in the Ad Exchange. The malware vendor never had a direct relationship with New York Times New York Times was using Google, I’m pretty sure you may have read they just announced they’re going to get rid of almost all third party or all third party ads. Because they have the scale and the high quality first party data on their subscribers and readers. They’re still going to use the Google ad server I believe. It’s just hard to replace that. But at the time 2016 New York Times, definitely was not just using the Google ad server. They were using third party ads. What is the third party.

 

Ad it could be somebody who’s trying to get an ad into a middle or bottom of the page slot, not a great ad and make some extra money for the New York Times. And it goes through the Google Ad Exchange. So that advertiser has to sign up with Google, but not necessarily with New York Times, generally not. And that that company is amazing. This has been diagrammed by some of the security researchers could be a fake, it could have fake LinkedIn profiles for executives, it could have fake address. It could be an address in I don’t know Ukraine or Russia or somewhere.

 

It could be pretending to be in respectable company when it’s really trying to distribute ransomware which takes over your PC and holds it hostage for Bitcoin, and encrypts the the file systems this solid state disk drive and when this happens, especially to somebody’s grandmother and they want their PC back with all their grandchildren’s pictures, the laughing skull tells them on their screen, how to buy Bitcoin and pay the ransom. And you’ve probably heard about hospitals that have had ransomware take over the entire set of computers in the hospital. This is a serious criminal problem. It’s a criminal enterprise. And it uses the Ad Exchange. And the kicker is, those intermediaries, including the main exchange operator, take their fee. And the criminals pay to put the app through its 40 cents, cost per thousand impression ad. It’s a low cost ad. But they’ll pay for it because they make so much profit on the ransom. And this happened to the New York Times BBC Online, AOL, I believe, and a few other notable sites. So I think continues to happen. It’s it’s not a solved problem. The New York Times going away from third party ads is one sign of the times that when you’re big enough, like New York Times, the Washington Post, you want to solve many problems like revenue problems, getting rid of intermediaries, getting rid of mouth advertising, so called malware and ads. You will get rid of third party advertisers. But if you’re using the Ad Exchange, and it’s matchmaking, so called advertisers against Real publisher space, then it can make a match without anybody checking that the payload doesn’t contain an exploit kit loader that’s going to try to bang on all these exploitable bugs in your browser or your operating system. Take over your PC encrypted disk and demand Bitcoin. There’s a flip side to malvert icing or a dual a mirror twin, which is ad fraud. So that’s when real advertisers know malware, are getting their ads put in fake ad slots in fake publisher pages that are being viewed by no human they’re being viewed in the matrix, often in cloud infrastructure by a bot a headless browser instance. And it looks to the JavaScript that loads with the ad including anti fraud JavaScript, it’s sold by vendors, like I don’t know double verify or moat which Oracle bought these vendors say our JavaScript can tell whether it’s fraud. Well, you can, I’m here to tell you I created JavaScript, I made it very mutable and extensible and easy to mock up. Standard objects are built in library objects and replace them even. And so ad fraud involves doing that some extent to put the double verify anti fraud JavaScript in the matrix. And, you know, if you’re in the matrix, sometimes you see the black cat twice, it means that the agents are reprogramming it that tells you to get ready for the attack. But if that was Neo, you have to be kind of the one you see that most people don’t. Most advertisers and double verify script writers don’t, they just can’t. There’s really no way in this sort of arms race for the JavaScript that loads to be sure it hasn’t been put into a fake environment like a fraud bot environment, put into the matrix. The Matrix is telling its brain that the steak is juicy and delicious. And so the average Heiser pays, thinking that it got a real impression, you know, it’s a fraction of a cent, but these add up, and it just blindly pays through this Ad Exchange to settle with the fraudulent, fake publisher, the big publisher, maybe even impersonating through domain name, claim or even spoofing that it is the new york times when it is not. Now, the fraud actor here has to sign up with the exchange, let’s say Google, as a publisher, so you know, Google police’s this but it’s hard. And the brand that’s running the real ad and ultimately paying the real revenue through the intermediary who gives the remaining part to the fake publisher is not able to check for discrepancies like, Hey, I have an ad on the New York Times through some direct deal. I know it’s its publisher identifier in the Google Ad Exchange on the publisher side. And now I think I’m paying you at times but it’s got a different ID This must be a fraud actor. I’m not paying it. You could do that. If you were diligent enough, you can get the logs that have all these IDs in them.

Most people don’t I think sometimes, you know, you can’t do it. And you know, there have been attempts to innovate around this by letting publishers declare exactly which third party ad vendors are allowed to place ads on their on their site, but that that got wrong foot in several ways. One again, you can sort of fake your way into the domain namespace. Now there is some of these so called trusted third parties were playing both sides against the middle, and the publishers didn’t take them out of this permission list so it went bad. So I hope I haven’t made your eyes glaze over. But this is really a baffling ordeal for anybody trying to make money and yet, it’s the main way the web is funded. And with Brave we wanted to get a better way going not just protect users from tracking which is their right, but let our users choose individually to give back. Something in lieu of the ad revenue that was being lost through through blocking tracking. Brave mainly blocks tracking, we don’t block ads, like on the Google search page. We don’t block Facebook feed ads by default, some of our users want this and we’re adding options for it, or Twitter feed ads, let’s say which generally are missing me. But they are part of the publishers content. And we neutralise or block their tracking, so we don’t take them out. We view the more like part of the first party content, we do block out so much tracking that almost all third party ads are blocked. And that’s why Brave gets such great speed wins battery savings, data plan savings.

And it is proceeds as an ad blocker even though we’re not intentionally trying to block those first party ads. But, you know, we want the user to be able to give back so we did something on top of what we call Brave shield, the baseline mode of Brave which is shielded from tracking and therefore from most ads and shielded from fingerprinting, which I haven’t talked about, which is a devious sort of alternative to cookie based tracking. It’s maybe less precise but it can be used also to sort of surveil users at least try to put users into categories of interest. So we block all the tracking, fingerprinting. And then we add the basic attention token system we did in 2017, based on a prototype we built using Bitcoin. And that lets the users who choose to give back to their creators even lets them since last year, I think lets users earn revenue for anonymous ads, non tracking ads. And this sounds kind of impossible, except you realise the browser was always in the tracking game as a passive servant of the JavaScript that was being loaded from the tracking companies or, you know, originally the pixel that was being loaded in my earliest rendition. So browsers should have some, you know, choice here. It’s not just a blind servant of ad tech. Maybe that happened because it’s owned by an ad company, or a company that has an advertising division, or a search ads business. But if users can choose browsers, which they still count on, even on the operating systems that deny you the choice of the default browser like iOS, from Apple, then users can choose browsers that have tracking prevention built in. And so Brave was ahead of the curve here. But a lot of other browsers have now grown tracking prevention features that relieve the user from having to get an extension or so called add on.

But, you know, Brave, I think it’s the only one that lets you earn from private ads to give back because instead of Brave being a blind runtime for the ad tech superpowers, we want it to be the user’s agent. And if the user chooses, then the browser can study your data locally, all browsers see your data, they see your navigation history, that’s why you can clear it. You see, whatever cookies you do allow to be set and Brave blocks, third party cookies by default. They see in a browser, see your your clicks, they see your bookmarks, if you choose to bookmark, you see where you are scrolling on a page and what’s in view, and they see absolutely what’s in view with ads. There’s been a lot of cheating over the years about what’s that ad really visible or what covered by something, was it halfway up from the bottom of the page? You know, so called below the fold was somebody claiming that the ad was seen when it wasn’t? Was there a viewability or visibility fraud or mis measurement going on? Well, the browser ultimately has precise rendering model, maybe even pixel accurate depending on how it’s done. And so it knows exactly what’s going where and what’s in front of the capacity stack, semi capacity stack and sea otter stack. And so realising this with Brave we thought the option for us if we don’t turn it on by default should allow a local agent local machine learning nothing too heavyweight, that is all open source so you can speak to our experts can inspect the trust. To learn study from classify interest from your data feeds your history mainly and the clusters of words in documents you browse that have to be loaded by the browser and rendered as you know, in a font. So the browser is doing all this text handling. It’s fairly lightweight on the side to do some natural language processing and machine learning. And that’s what we do if you enable Brave rewards, which is the system we have for giving the user 70% of this private ad revenue, and it is private, because we don’t track you, there’s no server tracking your history. Instead, your browser is studying things, to derive a set of interests that might be hot or topical for you. And then match those, let’s call them keywords, or segment identifiers against the keywords in a catalogue of ads and offers and everybody in the same region. Everybody in the UK for instance, gets the same catalogue. If we get big enough, it will be everybody in you know, the Greater London area, or we can even get finer grain but it’s a large anonymity set and by taking that catalogue as a download, that can be downloaded, compressed and efficiently updated incrementally. You don’t identify yourself, everybody’s getting the same catalogue in the UK, and there’s no signal back to us who you are in taking it. It’s like downloading a Safe Browsing list, which is something we also offer With some extra anonymity VCP Google, and Google provides a Safe Browsing service to many browsers to prevent you from being fished or being deceived into going to a malware site through like a link in an email. So it’s similar to that anti phishing Safe Browsing list. And it’s, it’s a catalogue of ads and offers, it’s not the ads themselves, it’s links to various edge cached ads, we don’t run the edge caches the advertisers generally do. And we don’t consider the edge caches to be an adversary in this model threat model. So we’re not too worried about that we can’t really defend against network level attacks anyway. Though I mentioned something about that in a minute. So when you are opted into this Brave ads part of the reward system, the catalogues downloaded you know that if I buy that the machine learning studying your interest, you’ve been researching cars at a slow burn, it’s escalated, you’re now on certain car sites you’re honing in on certain European sports cars. Let’s say the local machine learning can pick the best offer from the catalogue about maybe an alternative brand you were considering, but you cooled on or maybe it’s the lead brand and gives you an extra choice offer you didn’t know about. And it floats on notification. We can put this not in any page, but in your attention space, your ad space, the user has promo rights to control notifications. In fact, some people don’t like them. So we’re looking at alternatives or in app versions. But if you opt into the system, the first call to action comes in a notification to you that’s not associated with any page you’re on. And it may be at the right time to you it’s not going to bug you if you’re typing away or mousing away. Wait, still, there’s signup, idleness. it respects you know, other apps and screensaver. But when you are ready, and you see that notification, you can click on it and just seeing it and the call to action, small amount of text and little image is worth something. But if you click on it, you’ll get a full tab with Brave shields up so it’s your again, protecting from tracking, even from the advertiser and that’s the landing page. You could think of these Brave ads, these user ads as detached search ads, it’s like there’s no search engine. You don’t have to go to Google. But you’re getting these little text ads like you find on the search engine result page from Google or Bing. But they’re floating separately from your pages in notification. And if you click on them, you get tabs. And the tabs can go to full, full page content, promoting the, whatever the brand is promoting, offer you a discount, you’re still not identified, you’re still shielded, but you can engage with the advertiser, this could become a long term engagement channel with the advertiser, if you like it, it could be a place to sign up and get a test drive and get a discount could be a cost per lead ad model, which is very high margin. And we will give the user 70% of the revenue using our own cryptographic token on aetherium.

The basic attention token. And that’s important because we block all the conventional ad tech, we’re not the blind slave of the ad tech companies as a browser. So by letting users opt in, we need something to replace all the payments traditionally done through, you know, banking relationships or invoicing, credit cards that the ad tech companies use to pass the revenue along from the gross revenue, the marketer puts in chipping away the fees as the intermediaries, the middlemen take their cuts, till whatever 3030 pence in the pound is left for the port Guardian, right? We’re giving it to the user because he’s ads are in the user space. But we’ve always talked about another option, we’re still working with key partners to do this. It’s not something we’re rushing into, because it requires great scale and a great partner. But we could do ads in Pages two, if the user and the publisher both consent. And in that model, we give the publisher 70%, much better than the 30% regarding God, and we give 15% to the user that model, even though it’s not launched, we fixed the revenue splits because we want this sort of transparency. We want the blockchain auditability at the end of the day, and we want the user to get a cut so the user gets at least what we get 1515 in this perspective publisher ad model, and so it’s always at least what we get going to the user 7030 on the user ads, and 70% to the whoever owns the ad slot. Those are the two principles 70% of the ad slot owner, and the user gets at least what we get. We tried to do this to align our interests with the users and tie a knot that prevents us from being you know, mistrusted or from ever going wrong without blowing ourselves up if the user realised that they were being underpaid, for instance, and having things settle on the blockchain helps with auditing, there are anonymity problems I won’t get into here, but blockchains typically are public so you can see everything. So that’s not you don’t want to do micro transactions on the blockchain that also get costly. But again, having a browser helps you can buffer and you can use zero knowledge proofs or blind signatures various cryptographic protocols to ensure anonymity and prevent likability of events and user IDs.

So again, great doesn’t see you. But we know from the catalogue matching through the browser software running on your device that we have a good match. And then when you look at the ad, we can, we can send one of these cryptographic proofs to a server that doesn’t have any idea of the user identifier, the Internet Protocol address gets dropped by our edge cache partner fastly, and they can’t see inside the encrypted payload. And then we see the payload without the IP address, we can do the cryptographic protocol, verify the proof and count a view as from an actual user. But we don’t know which user we just know it was a real user and the user can get in return sort of in a receipt form, get the 70% of the revenue. This is what our white paper laid out is somewhat idealised form. We built it and as far as I know, it’s unique and we have I said, over 1.6 million monthly active wallets I think it might be higher now.

I think if people like brave users, for maximalists down to custodial account, if these users insist on their rights, then we will be in a better state.

Jamie: Brendan really appreciate your time, keep doing all the great work you are doing with Brave, I think it is the killer app at the moment. Looking forward to having you again at some point.