July 31, 2023

Browser Wars: Brave vs. Goliath

When looking at today's issues of privacy, social media, and AI, we can draw a lot from the battles of the browsers over the past 30 years. To explore this, we welcome a close friend onto the show, the father of JavaScript and a Silicon Valley legend, Brendan Eich. Brendan left the CTO role of Mozilla firefox years ago and embarked on a journey to scale up a privacy-first browser and search business that would reward the sites you visit with a token reflecting your attention and value.


Hosted on Acast. See acast.com/privacy for more information.

Transcript

Richard Kramer:

Welcome back to Bubble Trouble, conversations between two humans with a pulse as opposed to computer generated voices. It's the real life double act of independent analyst Richard Kramer, that's me, and the economist and author Will Page. This is what we do, we lay out inconvenient truths about how business and financial markets really work. We're now deep into our inquisition into AI and whether it's a bubble or just spells trouble. We welcome a close friend of the show and... A close friend, personal friend, the father of JavaScript and a Silicon Valley legend, Brendan Eich. Brendan left the CTO role of Mozilla Firefox years ago and embarked on a journey to scale up a privacy-first browser and search business that would reward the sites you visit with a token reflecting your attention and the value you see there. He's a true nerd, a tech visionary, and a good friend and no better guy to the way big tech is seeking to grab resources and control the AI revolution. Back in a moment.

Brendan, and full disclosure before we start in the fact that I'm a tiny early investor in Brave, but I always believed we needed alternatives to the chocolate, vanilla and strawberry basic flavors big tech were offering. Can you give us a bit of your background about your journey from writing JavaScript in a frenzy of all-night programming to deciding to fight the heavyweight champions of the tech market?

Brendan Eich:

Sure, and thanks for having me. I went to Netscape after a startup you haven't heard of that became a patent zombie called MicroUnity. I went to Netscape in April 1995 after turning them down the year before when I should have gone, because that was the entry level and the company was just starting. But I realized at that point that Netscape was about to take off the IPO in August 1995. Speaking of bubbles, Jim Clark knew that the internet was going to bubble fast and he wanted to get ahead of it or goose it so he pushed for an early IPO. I'd come from a company, Silicon Graphics, where I joined in 1985 out of grad school when I was used to IPO after three consecutive quarters of profitability, no EBITDA fake profitability, and so I was kind of shocked by Netscape going public so soon, but it worked.

I think it worked so well that we had pets.com and the whole debacle, but it wind up with what the Fed was doing, what Greenspan was doing, and what technology had been doing. I wrote JavaScript because they tried to get me to come and do a language called Steam in the browser, and Steam, if you are a nerd, is a list language, very beautiful language. When I got there, there was no Steam in the browser, it was going to be either Java only or I could do this sidekick language to Java, hence the name JavaScript. That's what they wanted in the marketing department, so we finally got a trademark license from [inaudible 00:02:46] in December of that year. Meanwhile, JavaScript I'd hacked out in 10 days in May. When you're doing software, you realize it gets frozen early, you can't change it, that happens. All you kids out there, be careful what prototypes you deliver for rapid productization because they will ship with all their bugs and more.

Richard Kramer:

Before I get onto Brave, one of the phrases we hear bandied about all the time is surveillance capitalism. Another one that we spent a good bit of time in Bubble Trouble dissecting is the Creator Economy, and we did a lot of podcasts about what a bubble that proved to be for all but the lucky few. You're kind of trying to solve both issues with Brave; give people their privacy back and reward creators. Is the first, protecting privacy, going a little bit better than the second, rewarding creators? Or, are there just too many creators out there to reward, which is something that Will's looked at in the music industry with all the tens of thousands of songs uploaded every second to the streaming platforms?

Brendan Eich:

Yes, first of all, I think you can't serve two masters, so we pick users over creators. All creators are users, not all users are creators. We put users first when we have to. We try to reward creators too because what I saw, doing Netscape, then Mozilla, and Firefox is the problems Will's observed everybody sees, network effects or power laws or double power laws that multiply create these very uneven distributions, and distribution power is what matters. JavaScript itself as a programming language benefits from this, right? But with Google, especially in the Web 2.0 era and, to a lesser extent, Bing, which chase Google on search, so Microsoft was trying to come back, there was a lopsidedness to the internet that most of us early web people didn't like. We thought the web should be much more level a playing field, many more creators who are doing okay, and we saw everybody get captured by the big platforms.

Firefox was a party to this because we had a search deal with Google starting in 2004 that was very rewarding, so rewarding that we kind of got captured at Mozilla, I'll say this because I was there at the time, by Google, and then they did Chrome. I realized in doing all this that the browser is this sort of underappreciated agent that should represent the user, but instead of being the user's agent, it's usually big tech's agent. In fact, all the big tech companies have their own browsers. With Brave, we're trying to disintermediate all those big platforms, give a browser to users that puts the user first and lets those users or creators get paid directly, so this should lower fees, it gets rid of a lot of fraud vectors, it makes it a much thinner stack of revenue sharing, eliminates counterparty risk. It's just a better approach if you can do it.

That led us not only to do Brave as a user agent, but also to look at blockchain because once you block all the ad tech payments that go through conventional Net 30 invoicing or conventional finance rails, you have to re-plumb the finance to include the user, to deal the user in. There's hardly any way to do this except crypto, so that's why Brave is into crypto as well.

Will Page:

Brendan, great answer. You've covered so much ground in that first question, I want to dig deep on all of it. A couple of quick points. Firstly, our previous guest, the AI music company, Boomy, B-O-O-M-Y, they had an interesting take on that creator consumer point that you refer to. The conversation with Boomy started like this. They said to me, "Will, how many creators does Spotify got?" I said, "8 million." Then they said, "How many people are on the planet?" I said, "8 billion." Then they said, "It's not a lot, is it?" Their intention to turn 8 billion people into creators, it was quite a jaw-dropping opener. But you talked about privacy, I wanted to come in on this privacy lane real quickly and ask a kind of policy question around privacy. Now, here in the UK we have our online harms bill going through Parliament, it's been going through for years now, similar bills around the world.

I call it the Christmas tree bill, the more twists and turns it takes, the more politicians put stuff on it, the more unworkable the bill actually becomes. But I made this point in my book, which is privacy campaigners want a fully encrypted messaging service whilst at the same time demanding red flags when someone is sharing indecent images. The point I was trying to make is, for these politicians and these lobbyists, pick one or pick the other, but don't try and pick both. Do you see, in the privacy debate from where you're coming at it, this trade off between people wanting their cake and eating it, which is privacy is a much more complex subject and saying, "Ban indecent images, but make sure my service is encrypted?" How can you comment on that?

Brendan Eich:

Yeah, privacy, I'll use the UK pronunciation, is not a unitary good. This is something that has caused people to become neoists, privacy neoists, or to turn from where Zuckerberg was 13 years ago where, "Oh, privacy, nobody cares about that," to I think 2019, Zuckerberg said, "Facebook keeps you private," so he changed the word. That meant Facebook has your data, but everybody else doesn't so when you get a monopoly, you stop sharing. But privacy is a complex thing, it's a set of security properties in a technical standpoint and those are never solved, it's an eternal struggle. That means you can't really market it easily. But people have become more conscious over time from [inaudible 00:07:57], especially with big tech, with ads, and the cost of ads and tracking. People now using Brave know that it's a lot faster, they don't get those YouTube ads, their battery lasts longer. These are tangible wins from practical privacy.

I want to subordinate privacy to our user-first agenda. If you put the user first with computing, it's a serious obligation because a lot of computers now you don't really own, you buy them from Apple every two years and they're very expensive, and inside the hardware there's a magic inner ring of trusted computing that may contain DRM and who knows what, that's not your computer. But if you ignore that, if you try to at least get the browser and the web to operate in your interests, then you have to put the user first in a way that wants privacy, it wants dealing the user in on finance, it wants other goods that are not just privacy as some kind of overarching single principle because it's not. That means we have to be open to economics, because we're trying to replace a lot of the economics that we block with our shields.

Brave Shields blocks almost all tracking in ads, we block those YouTube ads. I think privacy, you'll hear people say there's a privacy utility trade-off, and certainly that's true for somebody else's utility like YouTube's because we're blocking their ads. But for the creator, that should not be the case because what I've seen rise, especially on YouTube, is the creator fan networks that supervene on top of YouTube and they often rely on crypto, Patreon, the super chats on YouTube, but also other payment methods. That economy is where I think we should move. That's what the web early guys like me wanted, and gals. We wanted to see the creator fan networks to be free of these gatekeepers and these onerous taxes.

Will Page:

There's so much in this to cover here. I mean very quickly, one of the best acronyms that Rich has come up with in the history of this podcast was the DDA, he called it the Data Donation Agreement, which is what you sign when you join these platforms. I thought that was one of his favorite ones. Also, you just mentioned there as well platforms on top of platforms. I did a piece of work with Twitch during lockdown to help artists learn how to livestream, and it's just interesting to see Patreon could sit on top of Twitch, Stream [inaudible 00:09:59] could sit on top of Twitch. That's something in a platform economy to be aware of is when you get these platforms sitting on top of platforms, it muddies the water a little bit.

But I had a second quick foundational economic question for you before we hand it back to Richard, which is you mentioned network effects in your first answer, and platform economics and two-sided market theories. For the listeners, I get a bit baffled with this subject because I studied economics, I get it, but sometimes economics is common sense made complicated. When you read literature on platform economics, on two-sided market theory, do you think this is groundbreaking stuff or is it no different from a bouncer at a nightclub who lets women in free and charges men a premium to get in? I mean, sometimes I think, "This is not that revolutionary after all, is it?"

Brendan Eich:

No, you're definitely right. Economics has become a bunch of obfuscated nonsense to excuse the depredations of the very rich.

Richard Kramer:

I think we can clear this podcast right now, that just wraps it up. Thank you, Brendan, for making everything so clear.

Will Page:

Here's Tom with the weather.

Brendan Eich:

But I think there is an obvious network problem when you have people on one side who want to get something and people on the other side who are putting it out there, how do you find each other? How do you market? How do you reach people? That's where these gatekeepers impose their taxes and choke a lot of creators out. I think if you look at the underlying web, it's more of a peer-to-peer network, the IP protocol was originally. There's a lot of centralization through peering arrangements and big routers around nation state boundaries, but it's still the best we've got and it's a lot better than just having everything on Google or everything on Spotify. I think that one of the goals with Brave is to make the browser be the sort of distribution agent instead of making these websites you go to or even a single blockchain be that. Can the browser be a more muscular client that represents creators and users interests above those of these intermediaries who've just gotten so much power through being on first or second?

Will Page:

I hear it.

Richard Kramer:

Brendan, I think it was well over a decade ago when you were at Mozilla that you heard me give talks about the vampire squids of the web and how that you would have these large, big tech companies that were phishing with a P-H-I, phishing to try to hoover up as much of the data that they could and for them, my other phrase that I love to throw out there was, open is a four letter word. I mean, it was basically a curse word, the idea that you would have open anything, you wanted to close things off and keep it all to yourselves. But can you dig in a little bit to your own effort to build an alternative to that world, especially now in search, but first in the browser? How does the ant move the elephant? How do you unseat the power of what in the case of Apple, Microsoft, Google, and Facebook, these are top 10 brands in the world?

How do you see, before we get into talking in detail about it, AI being able to catch up to the vast scale advantages and dollar value advantages these companies have because of the sheer money they can throw at cloud computing resource?

Brendan Eich:

Sure. It's a great question because I've done this arguably several times. Netscape took over from Mosaic, but you could say that was easy. Firefox took over from Internet Explorer. Some people say that was easy because Microsoft had de-invested in the browser after losing the USB Microsoft antitrust case and feeling punished in Europe by regulators, but it wasn't easy, believe me, we were there and we had to restart that market. The way you do it is with lead users. Lead users shape future markets. This is sometimes called user innovation networks, Eric von Hippel at MIT's written about this extensively. You have whole categories or industries are invented by basically rank and file plumbers that created a lot of the tooling that got standardized in the 20th century or US fiberglass surfboards, things like that. That's true of the web.

The web has a ton of this going on with web developers. They create frameworks, libraries, servers, and tools and those take off. Microsoft comes in and sweeps up with things like TypeScript and VS.code and that's great, it's a sign of the web maturing, but it's still early days. There's a lot of this lead user effect that Brave can capitalize on to move the browser market. We're growing. Growth is challenging because partly you have to use the big platforms like Google to grow, we have to pay for ASO and SEO. But again, if you can get these lead users, they'll promote you to friends and families, you'll get growth whereas other browsers have come and gone. We've seen this even recently, other search engines too, you saw Neeva exit to Snowflake.

Search is worth talking about because what we learned with the Google deal at Firefox is search is the other side of a single coin where the face is the browser. The user sees the browser and a lot of them think that's the search engine, or sometimes they think the search engine is the browser, and that's a reasonable conflation because they're so tied at the hip. That's why Google pays Apple so much to be default search in Safari. That's why we have these browsers captured by OS vendors. Android has Chrome as default and Google's been foisting it and still is probably doing this even though [inaudible 00:14:56] I think smoked them or tried to. You can't wait for the regulators.

As I recall, what they did with DG COMP when she was running things was they said, "Oh Google, you can't force Chrome through a dark contract on these operators or OEMs that says they won't get Google Mobile services if they don't make Chrome the default browser. You must have a browser choice panel," or the same thing with search, you must have a search choice panel. Well, okay, regulator, how do we do that? The regulator says, "I don't know, you design it for me." Google says, "We'll let you pay ASO, app store optimization costs, to hack your search engine, your browser up and rank in your country, and then you'll get into the search panel." Google made money on the remedy, which they then controlled the ball on through the Play Store, it's still captured. But in spite of all this, we're persevering and lead users are the way, I think they determine the shape of future markets and they will lead to a better web. The timing is unclear, but I think it will happen.

Will Page:

Brendan, you're going to need a rubber desk for this one, but I wanted a calculation that stated that Google had paid more in fines to the European Commission than they had paid creators in copyright.

Brendan Eich:

I don't know the numbers, but I can believe it.

Richard Kramer:

I think it's about 7 billion to the EU that they paid thus far.

Brendan Eich:

Yeah, still not enough. The thing about AI...

Will Page:

Still not enough to fill their pension black hole as well, but they'll keep fighting until that's filled.

Brendan Eich:

There's another thing you can do beside lead user effect, it relates to it, and that's if you're not a technological optimist, you should get out of software because there's just too much terrifying going on and you wouldn't get out of bed, you'd hide in your covers. I am an optimist and I do see technological innovation helping things like the LLMs that led to a lot of silly chatbot hallucinations, that's promising, it isn't connected to physics or mathematics yet, but it will be. You still have a lot of hallucinating, even GPT-4, but this has changed a lot of costs for us.

When we do Brave, it's not just a browser and it's not just a search engine, which we're doing too, and it's not just all this crypto economics that deal the user in and the creator, it's something that comes by default with OSS and browsers that are captured by the OSS or owned by the OSS and that's, oh, you get text-to-speech automatically for free. Oh, you get page translation among many natural language pairs. Well, those used to be super expensive and only Google could really mount the machine learning teams to train them and really to run the infrastructure for them. That's gotten cheaper and LLMs are making it a lot cheaper to do all sorts of natural language processing like that.

Will Page:

Wow.

Brendan Eich:

I think that's going to make it... Personal models that are finely tuned, smaller, more precise and even private are the future. That's something Brave's investing in.

Will Page:

Brendan, I got a couple of questions for you; one on history of browser wars and two on the regulation of the browser wars. I just want to make sure my history is accurate, and we'll get into the rabbit hole of attention economics in part two, but is it true that when Google was battling with Yahoo, a KPI of the Yahoo browser was time spent on the browser, so they flooded it with Yahoo Finance, Yahoo Mail, it's a cluttered homepage, but the key to Google's success was KPI was least time spent on the browser, that it was just a white space of the search bar, they want you to get in and get off as quickly as possible? Is that the origins of how Google won and Yahoo lost?

Brendan Eich:

Well, so Yahoo didn't have their own browser, they had a toolbar for IE, they had various web portals and properties. They never made their own browser, which was interesting because we talked to Jerry when he and Dave Filo were still running the place, and Filo is the tech guy so we got on great, but we couldn't get them to adopt Firefox. At the time we learned, and I learned this through the standards body with JavaScript, Yahoo was trying to sell to Microsoft and they didn't take the offer when they should have, but they never had their own browser. But you're right, Google said, "Let's declutter." Partly this was just I think Larry and Sergey being idealistic grad students and minimalists and whoever they had on design liked it. Fake Steve Jobs called it their Montessori kindergarten approach to their office decor and their homepage, right?

It was very simple, the four colors and not much there to distract you. I had friends who went there and became Google Fellows even and they would optimize Google economics through how they adjusted the placement of things on that page and on the results page, especially the search result page, let's face it. But keeping it simple is better and Yahoo was very complicated, all that web one portal stuff got too complicated I think. But browser wars did benefit from moving towards slimmer browsers. Firefox did it by factoring out a lot of the [inaudible 00:19:26] that were in late nineties browsers and suites of browser mail, calendar, news agents into extensions. We've made extensions popular with Firefox, and Chrome copied it and now set the pace for extensions with the Chrome Web Store. Chrome itself became more mineral because Google said, "Don't you just want this postmodern super thin order around our beautiful painting of our services on the web? You just want to see our stuff, our web content, we'll get the browser out of the way."

When I first saw it over a friend's shoulder, I thought, "Wait, is that a notepad with tabs?" I couldn't tell what it was, it didn't look like a browser. But thinner browsers are better in general if you have this extension mechanism that lets expert users, especially these developers and lead users I mentioned, get them extensions and plug them in.

Will Page:

Got it. Brendan, just real quick there in the regulatory question. I always take a lot of fun and pointing out the idiotic irony when people say, "We're surrounded by all these tech monopolies," and I point out that you're using the plural to describe monopolies, like how can there be more than one? How many tech monopolies do you need to worry about before you know you've got a problem with competition? But, how do we define the browser market that you're now battling in? For example, Amazon dominates online retail, but it doesn't dominate retail. Google could dominate online search, but we search when we're not necessarily online as well.

I remember I went to this eminent professors and regulatory conference in Brussels, all the great and good of the DG COMP are there. I'm telling you, Brendan, this was 2016, 2017 and I didn't hear the word app mentioned once, they just talked about the web without even referring to apps. I'm thinking about app metrics like downloads, active users, session counts. You have to triangulate those figures to understand app economics, they weren't even aware that apps existed. If you were a regulator, how would you define the market in the browser wars?

Brendan Eich:

The regulators are always a beater or 10 behind and they're always easily tricked into doing something that freezes incumbents and their power relationships. We've dealt with, again, these sort of Pareto or worse distributions where it isn't a monopoly, but if Google or NetScape at its prime had 80%, we could do things quickly and get things done. In some ways, you see that in nature, you see that in biology. If you're an evolving system where there are certain players that got on first, certain genes are successful in helping your immune system cope with novel diseases, they're going to sweep the population. JavaScript did that. I'm going to say it's a good virus or a good gene. But it needed to be standardized and needed good stewardship, it needed to be evolved over time. That was where JavaScript, I think, and a lot of the web standards got stagnant and security holes and other problems, and just stagnation hurt the web in the late nineties into the noughties and killed Firebox and then Chrome.

But we don't deal with pure monopolies, and monopolies, what are they doing now? It's not the textbook that weighed loss through price power, everything's free anyway, it's all loss-leaders. What they're really doing is they're just trying to corner the future markets. They have this big advertising exchange where they make all their money, everything else is free, and they want to make sure that any future thing like AI they can dominate. You may have seen Google, some leaks claiming somebody inside Google said, "We have no moat on AI," and that's probably true. I'm hopeful that through technology and user effects, we innovate around these things. But we've always battled them through being the David versus Goliath, we haven't waited for the regulators. We do work with regulators especially on privacy, also on areas of regulation that I think aren't so much privacy as antitrust. I'll just say it.

I mean, DOJ was calling everybody in the valley, they called me, I talked to them. They're not going to be quick, but sometimes they do wake up every 20 years and do something, and it's important to keep them informed otherwise they will freeze the world around the old incumbents.

Richard Kramer:

Okay, Brendan, we've got to go to the break. There's so much to unpack we're going to get into in the second half. I think it's fascinating now that, as I tried to point out at one of those conferences in Brussels in the past two years that I've attended it, this great idea of the DMA and opening up all this Google search data to all the companies, and there are basically four companies in the world; Amazon, Apple, Microsoft, and Meta that would have the cloud computing capacity to ingest billions of search queries a second and maybe to be able to make use of it. What I think we're ending up talking about is monopolies or quasi-oligopolies trying to compete with each other. It's nice to know that there's a few other folks nibbling around the edges and not simply throwing their hands up in the air. With that, we'll be back in a moment in part two. We want to go deep down the rabbit hole with Brendan's comments about AI and how it could unseat some of the structures of power. We'll be back in a moment with more Bubble Trouble. Thanks.

Welcome back to part two of Bubble Trouble with my dear friend, CEO of the Brave browser, Brendan Eich, and Will Page. We're about to go down the rabbit hole, talking about the impact of AI and whether it could undermine those big tech companies that seem to be dominating everything these days. Brendan, are LLMs just hitting the starting line now with the idea of grabbing as much data as they can get their paws on? We had a discussion in our last podcast about copyright. Can someone stand against their stuff being stolen by all of these crawler programs and AIs that decide that anything out there on the web is fair game? Will AI be won by those who can get the largest training sets? How can we fight this reality of the bigger getting bigger and biggest being even more gargantuan with the crawling they're doing in their AI programs?

Brendan Eich:

That's a great question because Brave has a search engine now, as I said, browser and search are flip sides of a single coin, and that means we have a query log and a click log. This is very valuable. It means we know what people are querying without knowing who they are, we don't link this to any profile or identity, and therefore we can do things like answer queries, zeitgeist sort of queries, things that you want to use to train AI. When you looked at what OpenAI released with ChatGPT, it was up through middle of 2021 and everyone noticed this and is grumbling about it still, but having a good data feed is hard. There's a [inaudible 00:25:44] called MetaCraft where you have all sorts of layers of adversarial gaming, and the bigger the system the harder it is to defeat this.

That's why I think small models finally tuned are better. You're seeing the really big dumb models are becoming commodified. If they are trained from some huge feed like Google could have easily, it's going to be not just dumbed down to a very low common denominator, it's going to be subject to all sorts of gaming from nation states and commercial interests, it's going to be full of junk. I think what we all want is a better handle on our own, curate a high quality personal set of experts, friends, and colleagues, and that's where these smaller models can shine.

Richard Kramer:

That's fascinating because what we're all afraid of is from, whether it's professional publishers who all of a sudden are waking up and realizing that ChatGPT had taken a subscription to the Wall Street Journal and sucked down every article they've written in the past 50 years or creators who realize that their blog is being used to train an alternative AI generated blog which has launched a sub-stack and said, "Why don't you pay me instead of this creator because I'll be half the price?" I do agree that a lot of the training that's happening for individuals is superfluous because it doesn't fit what our needs are. Will, I know you want to pick up on a couple of points here, but I'll toss the mic over to you.

Will Page:

Sure. In part one, I tried to lay a sort of economic baseline for Richard's guitar solos, and we talked about privacy and how you can't really have both encryption and content moderation, you have to pick one or the other. We talked about two-sided markets. The simple model of you think about to how McMaster launched the Diners Club in Manhattan, 7% of the merchant, zero to the consumer, more people will dine at your restaurant, you'll regroup your fees. We also talked about monopolies with the plural, with the S on the end, or what I like to call [inaudible 00:27:39]. Brendan, I'm sure you'll love... We had this hilarious politician called Screaming Lord Sutch from the Monster Raving Loony Party, lost 46 by-elections in his career in politics and played with Jimmy Page in a rock band, but he used to always campaign for two competition authorities.

I think that's something that you could campaign for in your work as well. You can't have one monopoly upholding competition, you need two. That gives us kind of a foundation of where the economics of these browser wars are. What I want to turn to is a subject in my book which I call the strategy tax, and people agree to disagree on what this word means, it's still in its formative years here. But what I like to think about it is when a tech company foregoes competition today to win competition tomorrow... You actually alluded to this at the closeout of part one. My favorite example is during the lockdown everybody had to do video conferencing, and Google made it easier to use Zoom than their own Google Hangout product. I always thought that was interesting, large monopoly, big company, but they're going to offer the competition over their own product to make Calendar more attractive, so they're foregoing competition today in video conferencing tools to win competition tomorrow in terms of who's got the best calendar.

Now, Rich's debates with me on the run when he can keep pace with me many times, but I just wanted to toss this idea of strategy tax to you. Do you think that when we look at tech and how it competes for convenience, like you said in part one, monopolies reduce output and increase costs, these monopolies are expanding output and for the consumer eliminating costs? Do you think there's such a thing as strategy tax where tech companies forego competition in order to serve convenience?

Brendan Eich:

Well, I do, and your Zoom example is great, but I think it's rare. I wish it happened more often. I think Zoom just had the right timing to take advantage of lockdowns and Google couldn't catch it. Google hate products, I mean what else can I say? Except for search and their own infrastructure internal products, they've killed some... You've seen the Google graveyard, whether it's [inaudible 00:29:35].

Will Page:

There's a company every 18 days.

Brendan Eich:

I guess it's a meet now, I don't even know what they call it, but they just don't love it and they were never going to be able to make it to overtake Zoom, so they just roll with that to help their other products like you said. But that's more the exception in the rule I think with Google. On YouTube or you see this with other companies like Facebook or Reddit now, Twitter, once you get enough power you should close down those third party apps and clients and APIs, you just shut everything down. The strategy tax may come help you in a case like Zoom in lockdowns, but it's a rare thing.

Richard Kramer:

But I think we have a great example of that with Microsoft in the sense that they have, time and again, made a good enough product, and I think Teams is a great example. They have the enterprise base, they saw that Slack was starting to eat into enterprise messaging and Zoom was starting to eat into enterprise video conferencing and they put them both into a product called Teams and the minute they got everybody to adopt Teams because it was bundled into their office subscription, they stopped developing for it and it is shockingly bad, but it is good enough. I guess the question is, both with these AI models and with all the products from these tech companies, how do you keep them on their toes?

I guess you would say a company like Brave is keeping them on their toes so that they don't just rest on their laurels once they've solved that convenience question for consumers that you are locked into, something I was going to talk about in the first set section, the behavior of defaults. I mean, we're all ultimately lazy. We all ultimately have the power of brand and the power of default. Will loves to talk about Barry Schwartz and the paradox of choice and how it paralyzes all of us to have too many choices, but we just say, "Oh, well it's easy to buy on Amazon, let's go do it," without realizing that we may not be getting the best deal because there is no price comparison on a platform like Amazon.

Brendan Eich:

By the way, I love Schwartz's work and I think we saw an example of that with Brave. We had Brave Search using the Bing API for image search. We couldn't do image search ourselves, we were starting small, we did web search, we do our own indexing, we don't crawl the whole web like Google does, but we couldn't do images so we used Bing API. People said, "Hey, Brave Search does images. Great, there's an image tab, I can search for images." Some of them noticed in 2021 spring when TechMan stopped returning the [inaudible 00:31:49] protestor in front of the tank. That happened on Bing with their image search, it happened on Brave Image Search and people suddenly said, "Oh wait a minute, you're using Bing's API," but a lot of people didn't notice.

Recently in preparation for what I called the Bing-ectomy, getting rid of our dependence on this Bing API for image search and cutting over to our own image search, which is coming out next week, we gave people instead a choice, we said, "Oh sorry, no more [inaudible 00:32:12] Village Brave Image Search, it's really just Bing API. You pick Bing or you pick Google," and people hated it. They hated the interrupt, they hated having to choose, they hated the fact that we weren't pretending like we were doing it. I mean, you'd think we'd get credits for being honest, but it didn't happen so I think there's something deep there. That ties into the power of defaults. If the defaults are good enough, people will stick with them and they will stick with them way too long, like with Internet Explorer on Windows. Firefox is a new browser that got attention or Brave, I'm not saying we're at the same level that Firefox got to because frankly it was easier against IE, it was so bad.

Chrome is a harder competitor for Brave to crack, but we can do it because of ads and ad blocking. You can give people enough of a choice that some people will... Lead users will adopt it, they'll tell their friends and family and that will start a rising wave of consciousness which goes through regulators, it goes through standards bodies, through web developers first and that can move markets.

Richard Kramer:

Your point is basically that outside of these lead users, it is bloody hard to get your average person to care because they will just...

Brendan Eich:

I think what you said about Teams is just right on because that happened to IE, that was the same story as Internet Explorer. [inaudible 00:33:18] Sorry.

Richard Kramer:

How many people are using the Edge browser today because, I don't know, that's what came with my Microsoft package even though it is shockingly bad?

Brendan Eich:

Yes.

Will Page:

Just on defaults, I think it's interesting to think that through, which is from defaults to inertia we have, and every single member of our audience has a great example of that in front of us, which is the QWERTY keyboard, Q-W-E-R-T-Y, specifically designed to slow secretaries down because [inaudible 00:33:45] inefficiently out of a keyboard. Yet every time we've tried to migrate away from QWERTY it's failed. Do you see a sort of QWERTY dilemma facing the browser wars in terms of we always do what we've always done?

Brendan Eich:

Yes, this happens in all mature categories and I think with browsers, it's kind of happened. There's the tabs on top, the address bar Chrome kind of set. Everyone's used to typing into that bar keywords, which I think is good though some people still type google.com and then put their search into the middle of the page, a surprising number of people do that. But yes, there's a stagnation in UX you have to fight over time. On the other hand, the browser is continuing to dominate as the universal app, especially the bigger the screen, not just on desktop but on iPads. You get other apps on iPad but you need fewer of them on my phone and you don't get them foisted so successfully on users. A lot of our Brave loyalists will take the mobile site and put it a home screen icon, a shortcut that launches a Brave runtime around that site and it's a better app than the app. This is true of YouTube especially.

Will Page:

Just a quick one there. At the very start, you introduced what you're doing in terms of rewarding attention essentially, and that's a fascinating topic for me. There's a chapter in my book called Paying Attention and right there I always like to point out the languages currency is paying attention. Other languages, Swedish, French, Spanish, it's offer, share, or give attention, but in English we use a currency of pay attention. Talk to me a little bit about that because I've often had this bug in my basement, which is the value of attention is not necessarily what you receive now. Let's develop the Barry Schwartz paradox of choice, we have too much choice out there in front of us. It's the ability to shut out distractions like what's the value of focus? Do you have any thoughts on that, which is I don't value your attention, I value the ability to screen out the distractions?

Brendan Eich:

Sure. If you use Brave and then go back to Chrome and go through the top 10,000 publishers, it's really unpleasant, there's just a lot of clutter from the ad tech and the first party promotions, and the third party smuggled into the first party because they tricked the publisher into putting something [inaudible 00:35:45].

Will Page:

Wow, it's an inefficient market.

Brendan Eich:

Yeah. Using brave, it's a lot cleaner and so your attention can be focused where you want it and that is more valuable, that increases your productivity in raw terms. Attention is very tricky though, and Richard mentioned the DMA trying to get search data shared after a three to six month delay I think is what they're talking about, that's too late, that's way too late. You need to get stuff fresh. This is the problem for these AI systems, they need to be trained on fresh feeds because the human feedback reinforcement learning is very quick to adapt and if you're out of date it's awfully stale. You need attention to be sort of discounted. You need to get fresh data in and you need to get the fresh results in front of the users, and that's a challenge for our attention economics.

Richard Kramer:

Now, Brendan, as someone who is not in the business of serving massive vats of Kool-Aid or kombucha or whatever is served in big tech cafeterias, I've got to sort of paraphrase one of my very favorite authors, Philip Roth and one of my very favorite books, The Human Stain, where he talks about the sanctimonious just being so crushing. Having listened on Monday morning in the UK on Radio four to Nick Clegg, the former deputy prime minister for Meta, begging for regulation of AI, I have to wonder there is this just the fox saying, "Could we please have the slowest rooster in the chicken coop watch benignly while I routinely nip in and steal some chickens?" Clearly there's been a huge amount of fear-mongering around AI and all of the tech companies are beating their breasts going to the White House promising to be, you love this word, responsible in their using of AI while they continue to advantage themselves commercially. How do you see this whole debate over regulating AI with companies themselves for once, who would've thought it, begging to be regulated?

Brendan Eich:

Yeah, I mostly ignore it because I would be really mad if I paid too much attention. Of course, it's what you say, it's these frauds and charlatans speaking up for big tech saying, "Let's regulate it." Some of these charlatans are like six cult founders in Berkeley, these weirdos, you know how I mean? Then they're saying, "Oh, it's going to be [inaudible 00:37:59] and it's going to destroy us. We have to pause it." I think that's all garbage. These LLMs are powerful and they're changing things like I mentioned text to speech and page translation, natural language translation, and they're going to create very powerful agents, especially if you can then tie them in through multiple models into physics and math. Steve Wolfram's stuff does physics and math, right? It does calculus, so you should be able to do that. That's going to be extremely powerful. It should not be towards a missile launch codes. But I don't think it's going to destroy us, I think idiocrasy is more likely to be a problem.

I'm just trying to ignore this to the extent I don't mock it, and we're out to innovate around it because if we move quickly and we get these lead users... Consumer is where computing evolved to from enterprise initially government scale, giant room-sized computers, it's now in your pocket and what consumers choose still matters even though the gatekeepers are there. Enterprise will still be stuck with Teams because it's bundled with Office, even though it's terrible and it's harder to dislodge that, but if you go against the browser there's more choice. The consumer is king, I hope. We even saw this with lockdowns in the COVID era because people were working from home and the IT department couldn't restrict the apps you use so much. I think that helped Zoom.

Will Page:

Brendan, quick one for myself though, it's a very selfish question I guess, but we do like to probe which professions are going to get upended by LLMs and AI. My sister Annie who listens to the show is our professional interpreter, French, Russian, and Norwegian, a strange set of languages. But you've mentioned translation a few times, do you think the translation profession and the interpretation profession is going to get upended by this technology?

Brendan Eich:

I think that they'll change things like all other automation has like the lever and simple machines did in prehistory, it's going to make it important that you have new people who build wheels or carts whereas before people just carried things, and so...

Will Page:

That's a good way...

Brendan Eich:

The best translators will still be needed because these systems are kind of blind, to use a religious term, Sola scriptura systems, they're just pertaining the text is the source of truth or maybe it's the new criticism, the text interprets itself, right? This is not true. It was superseded by the Post structuralist Stanley Fish who said, "I interpret it for you," but these systems are not grounded in physics and that's a problem. That's why you get such bogus answers. People ask them to write... Code interpreters, you'll pay for it, and it can still write code that looks very pretty and it's absolutely wrong, it's ridden with bugs. There's room for human intelligence still because we haven't got, in my view, a compelling complete model of the mind mapped onto the brain, so we're not going to build AGI right now. We're going to build these amazing agents that are really good with language.

Will Page:

Just a real quick one before Richard takes you to the smoke signals conclusion of our podcast. As a three straight veteran of the browser wars, I just wanted to throw this curve ball at you. When did you first start seeing bands change their name to gain Google search, like the famous one, The Weeknd drops the E off the weekend, or there's a Glaswegian band called Chvrches which spelled their name with a V.

Brendan Eich:

Yes.

Will Page:

My favorite one was when the Glasgow band started... This is around 2012, we started coming up with band names to play with Google search, there's a band that called themselves Casual Sex. I don't recommend you look for photographs of this band on Google search because you ain't going to get what you're after. But when did you first start seeing people play around with words to gain the search engine?

Brendan Eich:

I think it was around 20 years ago, I think you're right. Google really came in, to my notice, 1998 and I was still using things like Alta Vista and Google was really... It was really good then.

Richard Kramer:

Web crawler.

Brendan Eich:

We used [inaudible 00:41:30] deal in 2004, but around that time, around the time of user innovation colonizing the sort of user-generated content platforms, which I think was one of the big Web 2.0 innovation, the ReadWrite web, you could have a lot of people contributing to folksonomy's or to blogging sites or commenting systems. You started seeing people adopting pigeon spellings to beat others in search.

Will Page:

It's been at it for a while.

Richard Kramer:

By the way, as a great example of the kind of evil symbiosis between band names and Google search, one thing I just heard today was that TikTok as a platform has, as its third-largest advertiser, an ADHD medication that promotes self-diagnosis called Cerebral. The platform that does more than it can possibly imagine to divert, dissect, and chop up your attention into the littlest bits it can IE few second videos is also the leading platform for advertising a self-diagnosed ADHD medicine, which like TikTok itself is extremely addictive.

Brendan Eich:

God Damn, Kilogram. God Damn.

Richard Kramer:

And very addictive.

Will Page:

This reminds me of, I once went out on a date to a place called Atlantic City, won't be going back there again, and the date went pear-shaped because I kept on arguing with my date, lovely Haitian woman, and she said, "Look at all the economic development in Atlantic City. This is all Donald Trump," this is 2005 by the way, "Look at all the good that he's done." I kept saying, "But is this the right type of economic development? These are all casinos, is this what we should be building?" We argued, argued, argued. On the bus back up to New York, I had the winning case, which is all these billboards as you left Atlantic City promoting counseling services, "Buy two, get a third three," for gambling addiction.

Brendan Eich:

Yeah.

Richard Kramer:

As you got back to New York, you could see the greatest extractive industries of all of the banks in their tall towers at Lower Manhattan. Let me move on, Brendan, to our final portion of Bubble Trouble. As someone who's listened to the show many times, we have a section called Smoke Signals where we ask our guests to give us a couple of those "Uh-uh" moments when you hear something you just know is a misdirection play, is terminology or metaphors that just make you cringe. When you're listening to people talk about big tech or AI, what are the kind of things that you're hearing, what are the couple of things you'd warn people to be cautious of, show that we're in bubbles, kind of like $6 trillion for the metaverse that just make you facepalm and roll your eyes?

Brendan Eich:

Sure. This has happened... Perhaps it's happened even more often now, you're seeing this with all the gloom and doom around AI and need to pause to help the incumbents really. You're seeing it with what I think is not so much talk because there's plenty of hype talk, it's the VC investments that are now already going bad for things like Jasper. Was it a hundred million or over 500 million or a 10 figure valuation? It's essentially a thin wraparound OpenAI. AI is not going to be this easy first mover on the user interface wins, it's not just going to be a chatbot that becomes your NPC friend for life, your BFF that you can talk to while you order your ADHD medication for TikTok overdosing. People are going to find I think the real uses for these LLMs in systems that have higher quality and that's why I'm in favor of these smaller models finely tuned.

I see lots of hype. I look at VC and debt, which PE loves, I think you had a guest talk about this recently as warning signs when you see debt being abused and then you see interest rates went as things get rolled over, things are going to blow up. They're going to detonate. I'm probably going to piss off my few VC friends left, but there's just too little accountability. A lot of them were front running crypto projects and it wasn't a good idea except that they got out early and then now they're avoiding crypto, so thanks a lot guys, we'll see you on the next hype cycle, and those who went into things like Jasper are going to suffer. Debt is obviously now costly, and as it rolls over that's going to hurt a lot of these debt finance companies with these private equity super companies. Where did SPACs go for instance? Remember the SPAC hype cycle?

Richard Kramer:

Is there another one in the technology field where... I mean you must see so much obfuscation, misdirection kind of just rank nonsense masquerading as the most intelligent person in the room. I mean, where is your biggest alarm bells in terms of bullshit generator ringing right now with respect to this whole AI hype and especially to some of the stuff coming out of big tech?

Brendan Eich:

Yeah, I mean the AGI panic pushers that I shouldn't discount it because maybe they will get a regulator to do something really dumb, that would be bad. Will mentioned content moderation. Obviously, there's bad content out there and it's often illegal and you have to worry about it, we have to worry about that in Brave Search. But if you have technology that can handle it in a way that preserves privacy, you should use that. There are techniques for doing this. I won't get into all the mumbo jumbo of the mathematics, but crypto algorithms and protocols can do a lot for us. We rely on them for our credit card numbers to be safe when we do payments on the web, and that was just a big innovation from the late nineties that had to be patched up, it's fixed and made secure over time. We should have more like that.

It can handle things like CSAM and other problems that need to be detected. Apple's trying to do it with image recognition and that's got a big false positive problem. You can, again, see Cory Doctorow on this. I don't think that's the last word on it. I don't think we should drop our hands and say, "Everything has to be unencrypted so that big brother can just check for the naughty bits," because that has too much downside risk. But I do see panic porn and AGI concern as warning signs, both people trying to walk in incumbent power and somebody else is trying to sell something, they're trying to push their new VC funded super AI startup.

Richard Kramer:

Brendan, we could go on and on because I think we've only just scratched the surface of a bunch of super interesting topics. We'd love to have you back in another year and see if Brave got to 50 million users at this point, if you can get to 100 million and equally how you're going to develop your own fight the power version of AI tools that maybe show some of the big tech guys that bigger isn't always better and biggest isn't always best. Thank you so much for joining us. On behalf of my co-host Will Page, this has been a terrific episode of Bubble Trouble with Brendan Eich, CEO of Brave.

Brendan Eich:

It's been a 10.

Richard Kramer:

And we'll be talking to you next week on another topic about AI hype. Back soon.

Will Page:

If you're new to Bubble Trouble, we hope you'll follow the show wherever you listen to podcasts. Please share it on your socials. Bubble Trouble is produced by Eric Newsom, Jesse Baker, and Julia Net at Magnificent Noise. You can learn more @bubbletroublepodcast.com. Until next time, from my co-host Richard Kramer, I'm Will Page.