Lessons From Self-driving Applied to Web3 - Asta Li

Hello everyone. Today I will talk about the boom and bust cycles in autonomous driving and Web3, as well as the lessons learned from these cycles. To introduce myself, I'm a reformed autonomous driving engineer. I have worked on various autonomous robots over the past decade. What started my journey into autonomous driving when I was at Uber in 2016 and then became one of the first engineers at Aurora, starting in 2017?

I spent a lot of time thinking about data infrastructure while working on autonomous driving and left over a year ago to create data infrastructure for Web3. Today I'm building Privy, a developer tool that manages off-chain user data , and I thought it would be fun to share some parallels between my previous life and what I see today in Web3. To give structure to the talk Web3. To give structure to the talk , I will briefly review the past and talk about the shared history between autonomous driving and Web3 in terms of boom and bust cycles. And then I'll talk about a couple of conclusions.

Regarding expansion and consolidation. History doesn't repeat itself, but it rhymes, according to that famous quote attributed to Mark Twain. I think an interesting perspective from which to view history is Gartner's hype cycle, where we see this technological trend over and over again with an innovation trigger. There is an expansion of companies and projects that emerge with a peak of inflated expectations, followed by a phase of decline and consolidation as they enter the abyss of disillusionment. And then, a further slope of illumination beyond that.

Let's talk about that spark. In the case of autonomous driving and Web3, it is not a single spark, but a combination of technological innovation and social coordination that causes this phase of expansion. Regarding autonomous driving, I believe the greatest innovation revolves around deep learning and the research that has driven unprecedented advances in computer vision. And for Web3, much of this revolves around the launch of Ethereum and the ERC20 standard. Besides technological innovation, I believe the least discussed spark for expansion is actually human coordination.

So, in the case of autonomous driving, it's like the DARPA Grand Challenge, a government-funded initiative for various projects in the academic and industrial fields, with the aim of trying to create an autonomous car in a competitive environment. These are regulations that make autonomous driving legal on the streets for the first time. And as for Web 3, it's about standards that people agree upon and coordinate to enable the promise of Web 3 interoperability. In fact Web 3 interoperability. In fact , I think there's a parallel: the lack of regulation is one of the reasons we've been able to conduct interesting new experiments in the Web 3 space.

Beyond that, the enthusiasm, funding, media coverage, and public interest in autonomous driving and Web 3 are key points of coordination that drive innovation and expansion drive innovation and expansion . Therefore, expansion is not only driven by technology, but also by human coordination. So, let's move on to the expansion phase. Regarding autonomous driving, I started in the sector in 2016 autonomous driving, I started in the sector in 2016 , and between 2016 and 2017 there was a massive explosion of autonomous vehicle companies. I was going to CVPR, a computer vision conference, and it seemed like every company there was focused on autonomous driving.

There were about 40 or 50 companies trying to launch level 4 or 5 autonomous vehicles. There was a huge influx of capital, a lot of enthusiasm, and headlines saying, "By the end of 2018, we'll have fully autonomous cars on the road," with Elon Musk tweeting nonstop about coast-to-coast self-driving in just a couple of years. This is how I felt at the end of 2017, at the height of enthusiasm for autonomous driving, looking forward to my glorious future with autonomous robots. Meanwhile, interesting things were happening on Web 3: 2017 saw the first mass public awareness of Web 3. It was impossible to pay attention to the news and not hear about cryptocurrencies.

This led to the rise of ICOs. In particular, there was a Block One ICO of $4 billion in a single year, and they didn't have any product, which I think is crazy. And during 2016, there were about 800 ICOs that raised more than $20 billion in capital, with Bitcoin surpassing $10,000 for the first time. So , once again, this was me in late 2017, at the height of Web 3, looking forward to my glorious decentralized future. But at the same time, you know, I think that by the end of 2017 we also saw a kind of absence of consumer applications beyond access to layer 1 blockchains.

And I still hadn't even been in a self-driving car that wasn't operated by a person in the driver's seat. So that brings us to the consolidation phase that occurred after this peak in 2017. Where it turns out that autonomous driving requires more capital and time than expected, and throughout the years following 2017 we see a huge consolidation. This includes Amazon buying Zoox, Apple buying Drive.ai, GM buying Cruise, Nuro acquiring Ike, Aurora and Uber's autonomous division, and so on Uber's autonomous division, and so on .

And this ranges from acquiring talent for pennies on the dollar to strategic partnerships. But regardless, the consolidation was very real. Meanwhile, the largest self-driving car companies continued to grow, and now Waymo and Cruise, for example, each employ thousands of people . Similarly, in Web 3, after the rise of ICOs, there were many unfulfilled promises, many tokens without underlying contractual utility, and companies that took years to fulfill their promises or never launched anything. So we are in a kind of abyss of disillusionment.

But alongside consolidation, we're also seeing massive successes emerge after 2017. These are some of our first consumer applications and enduringly valuable protocols that are still around today and are fundamental parts of the Web 3 ecosystem. And interestingly, much more so than in autonomous driving, there's a natural path to consolidation in Web 3 related to network effects, where value increases with each additional participant and the incentives to consolidate are quite explicit. And to be fair, the cycles of expansion and consolidation in Web 3 are not just one, but multiple. Web 3 is broad enough to have several overlapping hype cycles across DeFi, crypto gaming, stablecoins, DAOs, NFTs, and more.

Therefore, it is not just a single curve. So this leads us to the question: at what point in this process are we within Gartner's hype cycle? In other words, where on the illumination slope are we right now? And there's a lot of doubt about whether we've already arrived, both for autonomous driving and for Web 3. Well, and I think that regarding autonomous driving, there's a valid criticism, and it's a slide that I could never have shown if I were still working in that field.

However, the vast majority of autonomous driving kilometers are still covered with one or two operators. There have been accidents involving autonomous driving. And, you know, I couldn't say if we've already saved lives couldn't say if we've already saved lives . Similarly, there are criticisms of Web 3, such as "I'm still waiting for my glorious decentralized future," but in reality there is a lot of centralization, whether in asset ownership or throughout the Web 3 infrastructure, as Moxy recently discussed. But, in general, I take it as a sign that, in all areas, we are still in the early stages and that there is much more to come.

And so, from this shared history of rise and fall, expansion and consolidation, I believe we have much to learn from both autonomous driving and Web 3. The first lesson is really a conclusion I draw from my experience with Web 3, and that is that in autonomous driving we should dedicate more time to breaking down silos. Much of this reflection stems from why Web 3 became so popular. To begin with, I believe that the rise of ICOs occurred due to open access to capital, which allowed companies to raise funds without having to be part of the funds without having to be part of the Silicon Valley "boys' club," and being able to raise from investors who were not only private institutions, but also from the public and people who actively participated in those protocols and networks. I believe that Web 3 has delivered on its promise of interoperability, and that by building openly, overlapping projects, and working with open source code, Web 3 has been able to advance incredibly fast and launch products in a way that autonomous driving has much to learn from.

An open secret in autonomous driving is that all the technology stacks are more or less the same, since the same algorithms run on systems made up of almost identical building blocks . And I have personally written the same geometry libraries several times. I would love for autonomous driving to adopt the open-source, collaborative spirit we have in Web 3, and I believe that would lead to faster adoption and progress in the sector. In particular, I think that, for example, we could greatly improve autonomous driving in terms of data sharing, i.e.

, sharing a disconnection data set where disconnection data set where , whenever an autonomous vehicle makes a mistake on the road, an operator should take control. I would love to see those scenarios shared between companies and used to create simulations that we can all benefit from. But even though autonomous driving systems all look the same, I don't think we should make them interoperable. So I'm not advocating for a proliferation of autonomous driving standards that nobody uses. And this is something that has been a bit frustrating for me in Web3, where everywhere I look there are 18 standards that everyone wants me to use.

That's why I think there are nuances in this exhortation to break with the silos. And on the subject of nuances, I understand why autonomous driving companies want to own everything, and it's because the interfaces are complex and it's easy to over-design interfaces prematurely. So I think that, when it comes to breaking down silos and sharing, there are things that make a lot of sense to share more, such as research, algorithms, infrastructure, disconnection data, and so on. And there are places where breaking down those silos is less useful. For example, it is less useful to share raw sensor data or the internal code of the modules.

Similarly, I believe there is room for nuance regarding coordination and decentralization in Web3. The other open secret regarding Web3 is that it is much more centralized than we want to admit and than we say it is. And I think the nuance here is that centralization, to begin with, is not always inherently bad . I believe that coordination in decentralized structures is very important and can be a starting mechanism. So this is sort of a revision of my original lesson, which is to break down the silos, but to do it in a more thoughtful and nuanced way in a more thoughtful and nuanced way .

And the second lesson is to invest in tools, because from autonomous driving I have learned the profound importance of tools and infrastructure. In autonomous driving, we have discovered that in the history of expansion and consolidation history of expansion and consolidation , the best companies, the ones that are long-lasting and still relevant, are those that, instead of trying to launch an autonomous driving system in just a year or two, are the ones that built for the decade. They are the ones who invested heavily in infrastructure and tools. They were the ones who invested in creating labeling tools or data infrastructure, onboard execution or simulation pipelines that have proven their value as the foundation for this decades-long autonomous driving project. And I think the same lesson can be learned for Web 3, which is to remember that we are building for the long term.

And I think there's a perspective on the long-term value of Web 3 that shouldn't be underestimated. I think there's an opportunity here to fundamentally change how people think about data ownership. I believe there is an opportunity to invest in tools and infrastructure for the long term. And I also believe that part of building long-term on Web 3 is focusing on usability and creating better tools for it. Which brings me to this second point, regarding lessons about tools, is that also, in the space of autonomous driving, many of the companies that most boosted the sector were those that created great tools.

Whether it was Scale AI, which created labeling tools , or the various sensor and LIDAR companies, I believe the autonomous driving companies that had a lasting impact were those that also built their own tools. And in Web 3, I think there's an incredible opportunity to work on great tools that will power the next generation of Web 3 businesses. So, personally Web 3 businesses. So, personally , I'm very excited to build on the Web 3 infrastructure layer. Taking these lessons back, I think the takeaway from this shared history of boom and bust, expansion and consolidation in these two industries, is to break down silos with nuance and reflection, and to invest deeply in infrastructure and tools.

That's what we do at Privy: we create developer tools that eliminate data silos. We make it easier for Web 3 companies to handle confidential user data and understand it beyond just being an address in the chain. We make it easy for users to control and share their data across applications and chains. So we're just getting started. In autonomous driving, I saw billions of dollars of investment that led to the training of an entire generation of robotics experts who otherwise would not have existed.

And I see Web 3 attracting incredible talent in the same way, creating a generation of builders who think about decentralization, distributed systems, security, privacy, and more. That's why I'm so excited to be part of this space. I think we are still in a very early stage and there is much more to build. Thank you.