13 Sep Brendan Eich Interviewed: “We HODL”
Brendan Eich Interviewed: “We HODL”
Published: 13 September 2018
London Letter: Joining us today is Brendan Eich, the co-founder and CEO of Basic Attention Token and Brave Software. We’re at the headquarters of Brave Software in downtown San Francisco. Hello Brendan, thanks for having us.
Brendan Eich: Hey! It’s a pleasure having you guys.
LL: After our report on BAT in January our audience grew significantly. There was also a noticeable bullrun on BAT taking place shortly after. Nevertheless not all of our readers are so familiar with your project. Can you shortly introduce your project in a few sentences to our readers?
BE: Sure. The Basic Attention Token is an ERC20 token on Ethereum designed to be a unit of account of human attention in a reformed and private system of internet advertising and anonymous contributions. This system could replace the current web digital ad system which has a number of problems.
We see a mature but troubled ecosystem of digital advertising, ad-hoc donations, Patreon, subscriptions, the list goes on… But by far the most funding for the visible web comes from ads. It’s estimated that it was $200 billion globally last year and $80+ billion in the US.
Unfortunately that system is an inefficient market. It is a system of arbitrage based on intermediation between advertisers and publishers. That’s because the web has such scale that you can’t be like a newspaper 100 years ago and go to your local department store to do a direct ad sales deal.
You have to use ad exchanges: demand side platforms as an advertiser and supply side platforms as a publisher…
LL: Tons of middlemen along the way…
Brendan Eich: Correct. This creates excessive fees. The ad-exchanges are open to all sorts of abuse and they have no economic incentive to reject abusive players because they make a fee on every transaction.
So you get abuse from the so-called “buy side” – the advertising side – where malware distributors buy cheap ad spaces and put cuddly disguised exploit-kit loaders and fake ads. You also get the counterpart abuse which is ad-fraud where fake publishers, fake humans played by bots, take ad-revenue.
LL: Can you give an example of a fake ad? Do you mean these pop-ups claiming you’ve won the latest iPhone?
Brendan Eich: Worse – ransomware. That’s a big criminal business globally, some people at Google have been studying it recently. It first broke in 2016 when a new malvertising campaign was on the front page of the New York Times, AOL, BBC Online and other reputable web publishers.
The reason it got there was not because these publishers made direct sales to a malware vendor or ransomware vendor (laughs).
It’s because they were using ad exchanges.
They were using programmatic advertising. “Programmatic” is just a jargon word meaning automated ad exchanges.
The Basic Attention Token is designed to express human attention in a way that bypasses exchanges and all those intermediaries and therefore realises savings by cutting out the fees. BAT also brings back users who’ve left the system via ad blocking. Adblocking is a negative externality.
You have a system where something bad is not expressed by the price being lowered to make up for the impact of the bad thing. Instead the bad thing, like pollution, just hurts everybody or it involves someone accidental and negative externality such as an AdBlock user.
So, with the Basic Attention Token we help to bring people who object to ads either on principle or because of these abuses back into the system.
There’s no easy way for anyone but the largest publishers to do direct ad sales – the smaller the publisher the less they do direct. Everybody who’s below a certain size has to rely on programmatic, even some of the bigger publishers have to use it. That means they’re all prey to these parties who intermediate.
If we can get users who opt out of the tracking that’s required by this intermediation back into the system without tracking, without hitting them with malware or subjecting the advertiser who’s buying the ad inventory to fraud, then we should be able to bring people into this new ecosystem and then we’re blocking the old ecosystem.
So what we do is an environmental rescue operation on the web.
You start small with this kind of project. There is a precedence – take Firefox as an example.
It was not doing AdBlocking but it was bringing back browsing when browsing had been reduced to a very poor state by Microsoft which got in trouble as an abuser of its monopoly.
In the US law you can have a monopoly (this is also true in Europe despite it being a lot more aggressive).
Monopolies happen in nature, they happen in biology, they happen in markets. Microsoft had a monopoly on the operating system on the personal computer. Windows was the operating system for the PC in the 90s. We all got PCs, bottle monitors, we played games, we played Quake.
That was okay until Microsoft then said: “Oh, there’s Netscape. Netscape is building a browser and doing a deal with Sun Microsystems to distribute Java and the Java Runtime Environment – that could be a threat to us.”
“Let’s use our market power with Windows to make a copy of Netscape, Internet Explorer, bundle it with Windows and use the monopoly at tying Internet Explorer into Windows where you can only have that browser or it was the default browser.” If you used a different browser they would reset to IE.
They thought “Let’s use that monopoly power to take over the browser market”
And they did that. I was at Netscape at that time, we all went through it.
But it was the users who brought Firefox back. It wasn’t government regulators.
Firefox wasn’t a traditional venture funded company. It was an open-source project that Netscape spun out. It took us a long time and it wasn’t a sure thing. One of the key elements that helped Firefox succeed was that we did build a better product.
Now, to take on this advertising problem. You cannot please everybody at once – you cannot please the publisher and the advertiser and the user. Especially the user who blocks that, who rejected the system and left. That’s why at Brave and BAT we start with the user.
LL: One of the things that helped Firefox’s success is that you gave the users choice unlike Microsoft who tried to enforce the usage of IE.
Brendan Eich: There were other browsers but Firefox had tabs, it had pop-up blocking, it had speed. Now the competition among browser has levelled out because the big browsers- except for Apple’s Safari – are somewhat or even entirely captured by advertising businesses.
Google is an advertising business.
Mozilla has a dependency on a big search deal with Google they renewed last year. They can’t really risk it too much. I think it may even have a hidden clause which says “no AdBlocking, no tracking blocking”.
I know that Microsoft’s Bing does search deals with browsers but they say “if you block ads and trackers by default we won’t do a deal with you”.
There’s been a capture effect even with secondary search engines like Bing. Google is arguably not a monopoly on search in the US – it’s somewhere around 80%, maybe more of the search queries by volume in the server logs. In Europe, Google is a monopoly – the search queries are more like 95% – and they’re in anticompetitive law trouble there. But in the US not so much.
Even so, Bing is unhappy with adblockers, you can tell: a search engine is a combination of a publisher and an advertiser if it’s selling space for text search ads.
If you’re an encumbered system you don’t like the disruptive changes that Basic Attention Token brings. But like ad blocking – it’s been rising globally for many years, it’s rising on mobile in Asia.
The first thing that we realised while building Brave was that we needed to protect users from malware, protect from bad experiences, from tracking as well as ads.
But our users said “What about that site? I like that site. I never click on ads but even the little ads that I see provide a fraction of a penny, the so-called cost per thousand impressions. I know if I’m blocking that I’m hurting my site a little bit. How can I help them?”
We said “Okay, Brave is protecting you, let’s add a cryptocurrency so you can then give back anonymously.” So we created Basic Attention Token upon the Ethereum blockchain.
LL: Was it after this malware attack which affected the publishers like NYT?
Brendan Eich: Yes, that was in March of 2016. So in 2016 we saw malware on the NYT. At that time we were building Brave. Then by the fall because we’d already started the Bitcoin Brave-payments prototype (we didn’t have the Basic Attention Token yet) we realised: we just want to give back.
We could actually do anonymous ads in the browser and pay them a revenue share. Everything we’ve now tied to the BAT was originally developed with Bitcoin in mind but Bitcoin got expensive.
Even in 2015 when I started Brave I wanted to have a system of social credit that allowed us to give users tokens of some sorts. In 2015 I called it “Brave coin”, just an idea – from nothing, for nothing.
So with the Basic Attention Token we’ve precreated a pool (a user growth pool) and then we did the sale of a billion on top of the 500 million we’ve precreated.
So that gives us over 366 million of user growth tokens plus reserves to give to the users. And that’s what we’ve been giving our grants from. We couldn’t do that with Bitcoin.
We’ve shut down the Bitcoin prototype and replaced it with the Basic Attention Token. Its idea, like with the Brave Coin idea from 2015, was mainly about giving users tokens to get them started from this pool. Now, the pool would eventually get exhausted but maybe we’ll have grown so much that it’ll stretch the entire ecosystem so that the coin becomes more dear – that’s just a mathematical property – and we can make it last.
We’re not worried about the long run because we think in the long run crypto should be easy to buy, there should be lower volatility and greater stability of the price, at least the second derivative should be a constant. We shouldn’t have these wild swings – for instance Bitcoin had an amazing spike today.
Anyway, we had this idea that in the long run with a large number of users, with other apps than Brave brought into the ecosystem, the price will stabilize. This is actually mathematically supported in the section 7.3 of our white paper, it’s in the appendix. We make an econometric argument that as the system grows the price becomes more stable. It doesn’t mean it’s constant, it just won’t swing wildly.
LL: Yeah, just like DAI swings a little.
Brendan Eich: Yes. Even USD Tether swings a little, if you can believe it.
Willy Woo, WooNomic on Twitter, did a chart on volatility of major tokens and coins last December 27th and believe it or not, BAT was less volatile than Bitcoin. And it was only number two, the least volatile was USD Tether in his basket of tokens.
So, we think we succeeded in our goals, we made the user growth pool, of course we’ve raised funds to fund the project. If we grow we will have a more stable unit of account over time.
There are various arguments for why you should or shouldn’t want a token to be an app coin. We don’t view this is as an app coin. But people who then say “I’m just doing a protocol coin”, they don’t have any leverage over their future.
0x is interesting. But I’ve seen people then say “we’re using 0x protocol but not ZRX” – they’re not using the token so they’re bypassing the payment.
LL: 0x is used to settle the transaction fees on decentralized exchanges. Basically it ties the use. On top of that, ZRX will be used as an governance token.
BE: Yes. Fully decentralised exchanges are challenging to govern. I mean it’s crypto2crypto but it also involves atomic swaps so it’s collateralized. There’s no margin or any risk like that but there are fees. It is slow enough on the current Ethereum blockchain that there is front running risk, there are other problems.
So, we’re pragmatists. I think you’ll notice that also. We start semi-decentralised. We can’t decentralise everything at once, it’s just not gonna work.
LL: What if Ethereum, the base for your currency, is going to fail? If it’s not going to work in the long run, can you switch to other blockchain?
Brendan Eich: You can, I’ve seen a number of projects do that. You can generally do it – commit a protocol between few chains to migrate token from one chain to another.
That would be a big mess if we did it. I mean, it could be done, it would be like a wartime rescue operation. And I don’t see Ethereum under that kind of wartime threat right now.
There’s a lot of disgruntlement with the pace of innovation on Ethereum. Plasma is coming in to make it a lot faster.
So, there are other people doing new chains and I’m sure in 10 years there’ll be a fantastic blockchain that is high throughput, low fee and probably anonymous.
To get to 10 years from now we can’t bet on that and then just take time out from reality. We have to be in the market. That’s why we’re using Ethereum right now.
LL: That’s why we are asking, because we would like to give our readers a sense of backup plan of the projects that we feature in our cryptocurrency newsletter.
Brendan Eich: Yeah, we thought about it. It doesn’t have to be a detailed plan right now because there is no imminent threat, but it’s possible.
You may see this – people will migrate chains. Now, I know projects that went from Bitcoin to BitcoinCash for instance.
LL: Your ecosystem is already being put to use. What problems did you face while creating it, while developing BAT & Brave?
Brendan Eich: The problem that we’ve also faced is the user growth pool – neither Bitcoin nor BitcoinCash solve that. And then a problem, I think everyone struggles within crypto is making it usable by average users, so-called “normies”.
Are they going to want something where when they lose the key they lose the money?
Are they going to want to dispute a charge like they do with their credit cards?
They say “that wasn’t my charge” or something wasn’t delivered, I want a refund, I want a chargeback. You can’t do that with crypto. You’d have to use some higher level provider.
I think that’s what we’ve done with the 30 days of private analytics on the browser that leads to the settlement through this anonymous protocol. It is more usable because it allows users to say “Hey, I don’t like that site, I’m gonna turn them off, I’m not gonna support them” or “I want to pin half of my monthly budget to Wikipedia.”
After the settlement and the anonymity protocol does its thing, there is no linkback to the user identifier, there’s no way to get a refund.We created a system that has different properties from subscription. For instance, I myself subscribe to the Financial Times. It’s one of those things I forget about but when I see a piece on Twitter and it’s behind the paywall, I can login and get it.
But I can also say “I’m turning this off now” or “I turned it off a little late in the month. Can you give me back that month’s charge, I didn’t read anything?”
That ability doesn’t apply to anonymity protocols. If you’re using the browser you don’t want to put everything on the blockchain because you’ll identify yourself right away and the gas fees will be really high. Those are the two problems but we dodge those with the 30 day buffer.
We also work on the user interface – it’s more like a frequent flyer system but not one you forget about. We give you the BATs from the user growth pool and that’s something you can do now, every month. If you go to your payments panel, you can click on “Claim my tokens” and then contribute back to your favourite YouTubers, creators and so on…
LL: For now that’s the only way to get BAT tokens…
Brendan Eich: From us!
LL: Yes, from you. Although recently you’ve started the ads user tests. How is it going so far?
Brendan Eich: It’s early days yet, so it’s a small, hand-picked group of users. I just was playing with the build today and it’s working. I think we now need to start exposing it more.
We are doing private ads to users – it’s kind of like a search ad. You get a little text notification that tells you there is an ad to look at. You can get rid of it or if you can click on it and it opens a new tab and you can get rewarded by viewing it.
Basically we’re lifting ads out of the page. This is a “user-as-publisher” model – the user gets 70% of the revenue.
Our model has fixed app-store-like revenue shares – 70% to the inventory owner. That’s the publisher if it were on a page – we haven’t done those yet. It’s the user if it’s in the tab. The other thing about the tab is that you don’t identify the user right away but the user gets to interact with the brand who’s doing the ad.
If you like it you can start to identify with it. It could become an engagement channel with that brand, it could become an eCommerce channel.
LL: So how is it going to look like really? A user visits a website, say, LondonLetter.org and we run BraveAds. Is there going to be like a pop-up?
Brendan Eich: No, not really. The idea is for publishers who want to work with this, we can do a more traditional model where there’s an ad in the page – (but that’s going to be phase two where we do a trial with the publishers. Right now we’re in phase one doing the trial with the users.)
That’s something where we’d need publishers as a partner and we haven’t yet pushed on but some publishers want to do that. So, in this model it’s Brave putting the ad into this slot with the publishers cooperation. The publisher would get 70% of revenue.
But this model we’re now, doing trials on, is a user model, it doesn’t require any publisher. It’s a two body problem – it’s the advertiser and the user.
There used to be something called “All Advantage”. It was a PC app in the late 90s. It did really well in the US. It was like a browser but it didn’t have ads in the main news, articles (it was a news reader). It had ads on the strip at the bottom and it gave users a revenue share.
We’re trying to lift ads out of the page where they have brand safety and other problems and put long formats into a user private channel.
There’s no ad-server, there is no exchange, there is no tracking partner, there is no data management platform.
LL: Without tracking those advertisements are going to be very broad..
Brendan Eich: No, they can be targeted and that’s an interesting idea.
The targeting can be based on your entire browser signalling history that’s blocked. You know, we’re blocking third parties from tracking you but, if you’re opting in this ad program, we’re not ignoring all the signals that would be tracked.
So, by privately running machine learning in your browser it can see you’re reading the Wall Street Journal and you have an affinity for European cars – maybe you’ve been searching for a new car and you haven’t found one yet. Now you get an opportunity to look at an Audi ad.
We don’t like the word ‘targeted’ because it suggests somebody remote with a rifle, right?
The idea with these user private ads is not just to give the user a share of the revenue. That’s just the standard, 70% for whoever owns the ad space. So, we’re following a fixed fee structure. But also to let the user’s wallet fill up again. So users who opt into this do not get tracked, they do not get identified to the advertiser unless they choose to be later.
But they get 70% of the fee, the advertisement gross payment and that fills their wallet for giving back to their favourite sites through our User Interface. So, when you opt into the ads, this system of Brave Payments is enabled as well. I think we’re gonna call it ‘Brave Rewards’.
That means – this is the really tricky part – we don’t make the users identify themselves to us since we’re not giving them tokens to take to the bank. We’re giving them tokens in the wallet in the browser in a controlled flow back to their favourite creators.
We want to let users who earn that revenue take it out if they choose to – it’s their tokens. But they’re going to have to go through a bank or an exchange to do that and those all involve what’s called ‘KYC’ – Know Your Customer.
So, we’re working with Civic on making that decentralised KYC solution usable so that people can take their BAT directly out. That’s not available yet but we will make it available this year. We want to let users take it but we think that most users will give it back, by default to their favourite sites and creators.
We think the donations are important for the ecosystem. Most users are going to be just letting their own attention tokens circulate. And most of the publishers, though not all of the YouTubers, some of them want BAT, but most of the publishers especially the big ones they don’t want crypto.
They don’t understand it, they don’t want the risk. So we have a partner which is like Coinbase – Uphold. They are a bank in part and they can convert to fiat, they can convert to other crypto.
So, if our platform becomes the dominant use of BAT, the money comes in for the user at the beginning of that 30-day period. On the 30th day the money goes through that zero-knowledge proof protocol in a single transaction to Uphold and then it goes into settlement.
Settlement takes 15 to 30 days and then it goes to the publishers. That means that the whole time is somewhat fixed because the publisher exchanges BAT back to dollars or pounds and that goes back to the advertising site.
So we do see a circular flow, we see the user as a more efficient intermediary than all the intermediaries in today’s ad tech.
There could be speculators, there could be other uses of BAT but we stabilise the hold time which helps us keep the price stable for BAT. That’s part of the argument in the white paper if you look at that section.
LL: Where can the publishers sell their Basic Attention Tokens for US dollars?
Brendan Eich: Uphold will do that for them. In fact, that’s already happening.
Uphold uses Bittrex for the BAT-to-Bitcoin probably because Bitcoin has the biggest depth and then Bitcoin-to-USD is on another crypto-fiat exchange. So those are two steps to get out.
LL: Let’s assume a scenario that in a couple of years Brave is going to gain such popularity as Firefox with hundreds of thousands of publishers that get paid in Basic Attention Token once a month…
Brendan Eich: Yes, they have volatility risk.
LL: The publishers like to have cash in hand for their operations, so once they receive BAT from their contributors or in other words their paycheck comes in, we can expect them to sell these tokens immediately on the market. With thousands of publishers doing that at the same time, we can expect the price of BAT go massively down on paycheck day.
Brendan Eich: It could. Though what we also see is with the ad system is people buying Basic Attention Token. Again, not directly, they can do it through our pool. So an advertiser buying ad space can bring US dollars as they’re used to doing.
LL: But that specific day would be a great opportunity to short Basic Attention Token.
Brendan Eich: If we settle on the same day. We’re small now so we’ve settled them on the same day. But as we scale we can make settlement happen on an ongoing basis so it wouldn’t be one day.
When I’ve mentioned the 30 days of private, on-device analytics – that’s an individual clock. My thirty days are not yours. It doesn’t count when I go on vacation, it starts on a different day for you because I joined the system on a different day. So, over time we can make each of these clocks asynchronous so we can have settlements over time. While we’re small we’re doing settlements on a single day.
But it’s a good point, it’s something that could happen at scale.
Since we’re giving our grants, it won’t shock you to hear some people tried to make fake browsing to take grants and steer it into a YouTube account. But those are easy to find – relatively easy.
There is always going to be that kind of a fraud threat but it’s much less than the current ad tech system where there are so many intermediaries and so many open exchanges that you really can’t stop fraud.
As we scale up we wanted to decentralize as much as we can. We’ve been talking to various zero-knowledge-proof experts about how to do this. We’re looking at Z-Key’s ‘starks’ or ‘snarks’ – Z-Key’s snarks and state channels was what the white paper alluded to.
None of this is quite ripe yet and Ethereum isn’t ready for it so we’re still using this semi-decentralized approach but as we go… and maybe as you said, in extreme conditions switching to a different blockchain we can make the system more decentralised. We can certainly run settlements more often and avoid affecting the price.
LL: So to sum up this part: which part is centralised?
Brendan Eich: The accounting server with the zero-knowledge-proofs is a centralised service we run. That is hard to put on chain, because that’s where a series of zero-knowledge-proofs that are not linkable neither to each other nor to the user ID even though it only happens once every 30 days.
Like the way people said, we could try putting in a state channel and advertise the Gas costs. That may happen. But even if you have a state channel you often have a centralised counterparty at the other end of that channel. Putting it on chain in some sort of 0x style-fashion has the same problems that fully decentralized crypto exchanges have. They can front-run, they’re slow, the fees are high because of the multiple transactions.
People like 0x but if there was a great write-up on Medium about the state of decentralised exchanges recently by someone who came from the real financial world, HFT world… and I think you correctly said – best examples are hybrid. They also have a centralised order book and they use the blockchain to arrange the matchmaking. But what they actually do to fulfill the order is using off-chain order book.
You can think of our accounting server as an anonymous order book for getting BATs to the right publishers without disclosing the identity of any user.
LL: The user donating that BAT…
Brendan Eich: Yes, right.
So we’ll see how that decentralisation goes. I have friends who are jumping onboard, like “I’m ready for the decentralised future, decentralise all the things”
And I’m like ‘Are you really going to decentralise everything, like ice-cream?’ No, it’s not all going to be decentralised.
LL: Since you are talking about 0x, maybe you can comment some more on it. Our recent report was about 0x and we think that just like the anonymity of Bitcoin transactions sparked the adoption to the wider public, now with 0x when you just connect your wallet without annoying KYC, whatsoever, no footprints, you can also trade completely anonymously. Crypto2crypto, no tax, no government regulations…
Brendan Eich: It’s a powerful idea. I think blockchains will get to the point where the front running risk will be immaterial, the fees will be low. It just seems like with Ethereum it’s currently not quite the case.
This HFT expert who wrote that Medium post – he said, he wasn’t freaking out about it but he said that better exchanges, like IDEX, are using a centralised order book. He’s an experienced HFT guy. He’s going to look at this purely decentralised system and he won’t have stars in his eyes about the ability to do an anonymous trade of crypto with somebody you don’t know. He’ll look at it in terms of fees, can it be front-run, etc.
Decentralisation is a journey we’re all on and it’s going to take a while to get fully decentralised.
As we scale it up we can connect our ecosystem with other apps like Opera, Firefox, almost everybody except Chrome (laughs) because Google doesn’t need it. It is the third phase.
That’s why the roadmap takes the BAT platform beyond Brave. We have to be the first, we have to prove the concept.
LL: So you pretty much plan to expand beyond your famous advertisers-users-publishers triangle.
Do you also consider third party services where people can create advertisements just like Adespresso for Facebook?
Brendan Eich: Yes, I think that if the ecosystem is big enough it will draw clean, value added vendors in it. You’ll have specialization but you won’t have layers of pollution you get with the current system.
LL: You‘ve mentioned the fraud… Google still struggles to fight against click bots that publishers run on their own websites to earn money on ad sales.
Brendan Eich: Or fraud ads, we face that too with Google.
LL: So once the wider public is going to realise that they can make some money by just viewing ads you can expect the same thing happen to your ecosystem.
Brendan Eich: Yes, there are two things we can do to help this.
One is – we don’t have the exchanges that are happy to take their fee whether it’s a fraud bot pretending to be the ad or not – we don’t have exchanges. We are putting the exchange on each device.
But the other thing we can do is: when you have code in some app we want to prove that the code hasn’t been tampered with. That’s integrity, that’s something that you can use features of the hardware for now.
That helps but you still might have robots pressing on touch screen now.
LL: So do you have any ideas as to how to prevent those Chinese arm robots clicking on ads?
Brendan Eich: Yes, that’s where we are in some ways in the same situation as clean advertisers – like Google which cares about that too. It’s hard for them.
We got on Google and they’re working on this problem now. But it’s clear that they have a problem they haven’t solved here.
LL: Do you imagine some sort of cooperation on that?
Brendan Eich: Yes because we’re using their advertising platform to promote Brave on Android. But we’re fighting fire with fire to that extent… to grow a small browser you have to get the word out. In the old days of Firefox it was mostly word-of-mouth, the web was smaller and Mozilla never made a paid distribution deal until maybe very recently. We’ve talked about it while I was still there but it wasn’t necessary, it wasn’t the way it was done.
But now to compete against Chrome in particular you have to get the word out through ads. So we’re using search ads, we’re using PlayStore ads and then we added in-app ads; as we found out: it wasn’t robots it was fraud apps that actual humans have somehow installed and they were taking advantage of those humans, they were running the ad too often. And we were paying the fraudsters, this is the annoying part. So it was in Google’s interest to help us suppress this problem.
But fraud is inevitable in any large scale system with humans in the loop. Anybody in finance, credit cards or advertisement will tell you this. Anybody who says they will eliminate fraud is telling you a fairy story.
LL: It is also important to point out that viewing ads on Brave browser is off by default.
Brendan Eich: Yes, that’s true. A part of the brand value is that we don’t want to impose anything on anybody. We want people to consent for a fair revenue share which by default goes back to their favourite sites
LL: BBC.com is a verified Brave publisher. Have they already implemented your ads?
Brendan Eich: No. They don’t have to do anything because those are user private ads. So this is the developer build that has the user private ads. No publisher…
LL: Oh, so the publisher doesn’t even know…
Brendan Eich: No, they’re not involved because the ad isn’t going into their space. If it was going into their space we would give them 70% and the user 15%, keep 15% for ourselves and we would do it with them as a partner – we would not impose it without their consent.
So we’ll always get publisher consent if they’re involved but our first ad model will not involve publishers – it involves users and their private tabs. We also have Tor private tabs, I don’t know if you saw that..
LL: Yeah, we’ve seen it.
Brendan Eich: So we’re increasing the options for users to be private (was that the missing word here?). One of the things that our current build does is trying not to disturb the user with ad notification.
These are all tunable parameters, but we’re trying to do something that isn’t annoying you. I think one of the reasons people use AdBlockers is not just the ad that jumps in front of you or covers the page but just that it comes at the wrong time.
With a local machine learning agent we can study your habits. If you’re looking for a camera at lunch and you didn’t find the right price or the right model we won’t bother you to go back to look at work, but later that night if you’re goofing around on YouTube and then go idle for a minute – that’s a chance to say ‘here’s the camera you want’.
We think this could work a lot better than the shock down approach that programmatic advertising in particular takes where someone goes… Facebook does this pretty well so for the advertiser it’s a nice dashboard, Google does this well.
You say ‘I want to buy an audience. I’m Ford and I have a new truck model that I want to promote. Give me all your 30 to 50 year old men who are truck buyers’.
That targeting at scale we can do in a differentially private way. Not on device – we have to let the dashboard see a perspective forecast of who the Brave users are who might be truck buyers.
That doesn’t mean we take their data out individually – it means there is a private algorithm for getting very coarse buckets of truck buyers, people who like certain luxury goods, people who are in certain demographic ranges.
We haven’t built that dashboard yet, that’s ahead of us.
But when you asked about our ads being all ‘one size fits all’, they’re not personalised. They can be personalised using this matching against the catalogue but the ad buyer wants to also see an idea of what are the users like.
LL: So all that data about the user is stored in the browser.
Brendan Eich: Yes, not on a server.
LL: Storing this data locally makes a first layer of protection, sort of.
Brendan Eich: Yeah, even when we do cross-device attribution which is attractive to advertisers, the data stays encrypted on any server that stores the sync back. We use Amazon S3.
So, most browsers allow you to synchronise your bookmarks, passwords, protection. Maybe your browser history though you sometimes don’t want the full history. But you can’t synchronise your laptop and your mobile.
We do it, but we encrypt on the clients so there is no plain text on the server – we can’t see your history. That means all the machine learning can see your interesting cross-device history. So if you’re shopping here and there and then you really start to get serious, this is where we can, using only the local machine learning, look at your entire contour of your browsing of the last, say, 10 days.
There’s often a look-back window in advertising where you can look back mostly 10 days and see if somebody was shopping for a car and they started being serious, they did some heavy research on their laptop but they didn’t quite buy it – we can give them an ad based on that entire history.
LL: How is it any different for every day users, for normies, say a 40 year old who likes trucks? Whether their sensitive data is stored within the browser or in some data centres in California?
Brendan Eich: So the segments have to be pretty coarse otherwise you could be identified. For instance, if I’m a Bitcoin billionaire and I’ve just bought a Swedish supercar…
Brendan Eich: Exactly. I’ve pretty much identified myself. So we won’t have that a category – that category of cars is too fine grain for us. But if you’re talking about a million Brave users and you’re talking about trucks – that’s a broad enough category, it doesn’t identify anybody.
It’s also helpful that it’s a consent-based system. We also want users to participate in the configuration of the dashboard. Now, it has to be configured well by default. You can’t have users obligated to tune it. The machine learning has to work pretty well without any hands on.
But users who care could say ‘I actually want to say I’m looking for a more rare item than your segmentation allows or I want to turn off this segment – I don’t want to be identified or associated with that at all.’ So they’ll be in control.
Right now in ad tech, if you’ve noticed how the ads sometimes have an opt-out button in the corner – that doesn’t work. I know people who’ve clicked on it. It leads them through a maze of 300 vendors most of whom have broken opt-out forms. You try to opt-out and they actually opt you back in because it just doesn’t work.
LL: Yes, that’s confusing. But with every user having his own dashboard, do you see people giving more detailed data away?
Brendan Eich: People understand that this is my personal data and it doesn’t go anywhere – it stays on my device. People understand this and they can let it just go by itself and they can tune it a little if they like.
When it comes to curating your data that’s shared with advertisers: there is a minority cohort of data nerds who like to do that. They even do it on Facebook. Facebook has a fairly open dashboard for its users. So if you’re on Facebook you can train your ads a little bit. Say what you want about Facebook – they are a tracker, they do obviously surveillance capitalism. They got in trouble in UK with Cambridge Analytica.
But they do have a really open box ad approach where you can tune it. We asked users whether they need those knobs… There’s a computing idea which is users selling their data. Now, it turns out you don’t get paid much if people disclose their data because it becomes stale. They say that a cookie that’s 30 days old is worth not as much, and if it’s 90 days old it’s trash, so just forget about it.
People’s habits do change, there is certain seasonality, there is certain regularity but people’s interest in e-commerce habits do shift. So if you’re selling your data you don’t want to give out the seasonal stable data because you’re just giving away the store. If you give away the data that’s going to shift over time you should probably not be underpaid for it. You should make them come to you and that’s our model.
People say ‘Google sells your data’ or ‘Facebook sells your data’ – they don’t do that because they’d be giving away all the data for someone else to mine.
Instead they bring advertisers into the transactional model with these dashboards. Then many small fractions of a penny add up and that is how their ad budget gets spent. Then the questions arise: did it really work? Was it fraud? Did it drive sales? It’s hard to know – it’s not a science.
We think users generally want something that’s automatic. If they like the ads and they don’t tune them – that’s better. So we think the average user, most users won’t want to get into this data curation business. If you really curated your data very carefully it might be super valuable.
The other fact about advertising is that there is, like I said, at least $80 billion in the US last year across probably 250 million humans (that’s optimised for adults and young adults online). That’s $320 per person if you do the math. So you end up with a gross spend of $320 and the user gets nothing.
It turns out some users are much more valuable. So it’s not all flat $320 per person. There are some that are indeed buying these fancy cars or medical devices. This is true for Google Search, I heard this directly from Sergey Brin in 2004. And when I asked him why Firefox was making so much money off the new search deal – on the first month it performed very well, and we only had a few million users at the time – I think it was Firefox 0.9. He said ‘Oh, but your users do more shopping, searching and eCommerce’.
We’ve optimised our search ads for the long tail of high margin items where they can charge more for the ad space or for the keyword. So it’s not a uniform distribution, it’s not a flat distribution. There are whales, there are whole species out there who are worth more.
That’s where I think Brave also pulls early adopters from. We get more valuable users that way – just as Firefox did.
LL: So how does the current bear market market affect your operations?
Brendan Eich: The market might be turning around. We’re not speculators. We’re building a utility token so we pretty much ignore the price. We hope that as we build utility the price stabilizes – as per 7.3 in our white paper. We take the view of all the smart investors – we HODL.
I think that’s what people should do with crypto if they have conviction. It isn’t for momentum play although I know people do day trading crypto. That’s a full time job and it’ll make you crazy, it will overwrite your brain. I think it’s important to try to fulfill the vision of the utility token that we articulated and others have articulated.
So we’re pretty much ignoring the market conditions.
LL: We’ll be wrapping this up with our last question: how about the numbers of Brave downloads or publishers? Are they rising?
Brendan Eich: Yes they indeed are. Brave currently has 4 million monthly active users and over 21,000 verified Publishers.
LL: Thank you for your time Brendan, it was an interesting experience talking to you.
Brendan Eich: Likewise, it was a pleasure meeting you.
If you’ve enjoyed this interview and would like to learn more about Brendan and his ventures, be sure to check out Brave Browser, BAT and his Twitter. Don’t forget to read other interviews we’ve prepared.
Want More Badass Content Like This?