Written by: Joe Zhou , Foresight News

"It seems that Vitalik has undergone some subtle changes in the past year."

This was the first thought that popped into my head when I finished my second interview with him in Chiang Mai.

Let's rewind to the end of 2024. Our first conversation took place in a quiet, enclosed room on Nimman Road in Chiang Mai. At that time, he was full of excitement about innovation in the Web3 application layer, and we talked for a full 90 minutes, from Farcaster to Polymarket to Solana and Base.

This time, the timeline shifts to a weekend in Chiang Mai at the end of January 2026, and the scene changes to a completely open space.

That afternoon, Vitalik strolled from "Sihai Community" all the way to "706 Community Co-living Space". On the second-floor balcony, he was sitting alone on a swing, swinging leisurely and contentedly, his relaxed demeanor just like any ordinary community member here. I sat down next to him, and as the swing swayed gently, I bombarded him with a series of questions.

We were surrounded by our own busy community members. There was no heavy security, nor any deliberate pomp and circumstance. Before long, seeing us chatting, a few curious members of the 706 community naturally gathered around. We sat down on the ground, just like a group of students chatting on a university lawn.

In the ensuing conversation, I was surprised to find that his thought process had undergone a substantial iteration over the past year.

Whether it's Web3 social networking, prediction markets, or AI, his perspective has become sharper and more specific. He dissected his observations on leading projects such as Polymarket, Farcaster, UMA, Chainlink, MetaDAO, and Base, and shared his latest assessments on Ethereum's role in the AI ​​era, decentralized stablecoins, and RWA without reservation.

Of course, the things that haven't changed are the most important.

He still has no fixed abode and never stays in one city for more than two months; he still has no bodyguards and will join our large group of people queuing up for a buffet in the cafeteria; he still has a great passion for decentralized communities and tirelessly travels between various locations in Chiang Mai.

When the conversation ended and night fell, the swing stopped. As usual, he dashed out onto the street before it got completely dark, hailed a ride-hailing car, and left alone.

Outside that decentralized kingdom worth hundreds of billions of dollars, he has always been defending his freedom as an "ordinary person".

Below is my latest conversation with Vitalik. The final section also includes selected questions from other 706 community members present.

 Photo: Vitalik at the 706 community in Chiang Mai

Vitalik's Chiang Mai Reflections: Technological Mastery, Why Have Applications Gone Astray?

Joe Zhou: A year ago, on the eve of Devcon, I invited you to do an exclusive interview in Chiang Mai titled "Vitalik, Forty-Two Days in Chiang Mai." Today, a year later, we meet again here. This return to Chiang Mai, what new feelings has this city brought to your personal outlook?

Vitalik : I've seen some communities thrive, like the Four Seas community, which has undergone many different changes. There are many activities, many people, and most importantly—they haven't become boring.

Joe Zhou: Environment and time often reshape thinking. A year later, I'm curious, how has your thinking paradigm regarding the core issues of Crypto changed? Where is your focus now?

Vitalik : The biggest change is that I see a huge disconnect between technology and applications.

Over the past year, Ethereum has made tremendous progress in scaling technology. Our gas capacity has increased from 30 million to 60 million, and this year's goal is to reach 300 million. Including the successful implementation of zkEVM and significant improvements in the user experience of infrastructure such as wallets—it's fair to say that the technological developments have been very successful.

However, I see many hidden concerns at the application level. Looking back five or ten years ago, the community had a very diverse and grand vision for the entire ecosystem. At that time, everyone was full of hope, thinking about creating a DAO, creating various decentralized applications that would truly change the way society collaborates, such as creating a "decentralized Uber." However, I feel that too many people seem to have forgotten these original intentions.

Crypto has succeeded financially, but it has lost its way in governance. For example, the current "token voting" mechanism of DAOs is flawed. The most typical example of the explosive popularity of Memecoin in recent years is that in early 2025, even Trump personally launched Memecoin. But I think that when he greedily issued his second token, MELANIA, his first token, Trump, was already dead.

Joe Zhou: Last year we had an in-depth discussion about SocialFi applications like Farcaster. A year has passed, and from today's perspective, how would you evaluate their development?

Vitalik : SocialFi is currently in a rather awkward phase. The biggest structural dilemma of SocialFi is that if you tie social and financial elements too tightly, financial incentives often backfire and overwhelm social incentives.

When users are no longer there to acquire quality content but to make money, they begin to generate a large amount of spam in order to maximize their profits. This is a dangerous sign—because financial attributes are destroying the essence of social interaction.

I like Substack's model. If you look at Substack's top ten authors, they are all very thoughtful and produce high-quality content. But if you look at the top ten authors on some Crypto SocialFi platforms, they are often just inflating their numbers or engaging in hype. The difference is that Substack does curating and community building. They work hard to find authors they deem high-quality and diligently help them integrate their work onto their platform, rather than simply providing a tool for issuing tokens. This is something Crypto entrepreneurs need to learn.

Joe Zhou: This seems to explain Farcaster's recent transformation—why they stopped focusing on pure social networking and instead turned to making wallets?

Vitalik : Yes. They haven't found a way to make it bigger. They're not content with making a "small but beautiful" product; they aspire to have tens of millions or even hundreds of millions of users. Based on their current trajectory, they believe the wallet sector is more likely to achieve this scale of mass adoption than pure social networking.

Joe Zhou: A few years ago, there was a general consensus in the industry that the application layer was about to experience a "major explosion," but that didn't happen. Looking back four years ago, did you also have the same optimistic expectations?

Vitalik : Yes, I did think about it. At the time, my thought was that the core bottleneck for the application's failure lay in the limitations of the underlying technology—such as insufficient scalability, slow speed, and poor user experience.

By 2025, at least at the L2 (Layer 2) level, these fundamental technical hurdles will have been largely overcome. However, embarrassingly, we still haven't seen a large-scale emergence of good applications. The only sector that could be considered to have seen explosive growth in 2025 is prediction markets, but frankly, even these have exposed significant problems.

Joe Zhou: What exactly do you mean by the "problem" you mentioned?

Vitalik : If you look at the discussions on Twitter, Polymarket's most frequent tweets are things like "Which team will win next week?" or "Will Bitcoin's price go up or down in an hour?" I think these short-term bets don't have much social significance in the long run. In theory, the prediction market is a successful tool (because it works), but we need more meaningful applications.

I think mechanisms with long-term incentives would be better. For example, Robin Hanson's Futarchy (prediction market governance) is very interesting. In traditional governance, people usually vote for people (presidents, members of Congress) or for methods (such as "Should we build this road?"). Robin Hanson's governance philosophy is that people only vote to decide the "goals" (such as GDP growth or reducing unemployment), and then use prediction markets to determine the "methods." Traders in the market, in order to make money, will "buy" the most accurate data with real money. Currently, MetaDAO is experimenting with this.

Behind the $70,000 windfall: Vitalik's "anti-madness" strategy and the hidden dangers of oracles

Joe Zhou: Do you still use Polymarket? I remember you used it quite frequently last year.

Vitalik : Yes, I made $70,000 on Polymarket last year.

Joe Zhou: How much principal?

Vitalik : $440,000.

Joe Zhou: Many people lose money, how did you make it?

Vitalik : My approach is simple: I look for markets that are in "crazy mode" and then bet that "crazy things won't happen." For example, there's a market betting on whether Trump will win the Nobel Peace Prize. Or some markets predict the dollar will go to zero next year during periods of extreme panic. When market sentiment enters this irrational "crazy mode," I bet on the opposite, and that usually makes money.

Joe Zhou: What specific sectors do you usually focus on on Polymarket? Crypto? Politics? Entertainment? Economics?

Vitalik : It involves politics and technology. If you want to make money, you need to go into those markets where people are caught up in crazy and irrational predictions; that's where you can make money.

Joe Zhou: You're the founder of Ethereum, so do you have inside information? With the war in Venezuela, netizens have discovered that some people seem to have advance knowledge. Have you encountered a similar situation?

Vitalik : Here I'd like to discuss a significant case regarding a vulnerability in oracles. There was a prediction market about the Ukrainian conflict, betting on whether the Russian army would control a certain city. The contract defined "control" as whether they controlled the city's most important train station. The data source (oracle) was anchored to Twitter and maps from the ISW (Institute for War Studies).

Then something happened: ISW employees, perhaps by mistake or intentionally, hacked their own system, and their maps suddenly updated to show that Russian troops controlled the train station. This caused something that was initially considered to have only a 5% probability (almost impossible) to instantly become a 100% certainty in the prediction market. Although ISW retracted the update the next day, they may have already paid out the money.

This reveals a huge problem: current Oracle data sources (such as Web2 news websites and Twitter) have far too low security standards. They never imagined that a single message they posted would determine the ownership of $1 million on the blockchain.

Joe Zhou: This does sound like a really crazy thing. You just pointed out some problems with Oracle. How should we solve it?

Vitalik : There are currently two main approaches to solving the oracle problem.

The first path is the centralized model, which simply means trusting a company, such as Bloomberg, to provide you with accurate information.

The second approach is Token Voting, also known as the decentralized model. Its logic is to allow those holding governance tokens to decide "what is the truth" through voting. UMA is a representative of this model. (Note: UMA is a decentralized oracle protocol on Ethereum that relies on token holders' votes to determine the authenticity of data.)

However, public trust in UMA has been declining recently. This is because many believe it suffers from a game-theoretic flaw: if a whale decides to manipulate the voting results, ordinary people have little chance of resisting. Because in this mechanism, even if you vote for the truth, if you oppose the majority, the system will judge you as the loser and you will lose money. This forces people to vote with the whales instead of following the truth.

I think a reliable oracle is very important because any DeFi project now requires one. If you want to develop Real World-related applications (such as putting real estate on-chain or predicting real-world elections), you need an oracle. Currently, the DeFi industry generally trusts Chainlink. However, Chainlink's mechanism is quite complex and relatively centralized.

I have always hoped that we can find a better solution in the future.

 Photo: Vitalik sharing his book at the Si Hai Community in Chiang Mai.

Ethereum's way of survival in the AI ​​era

Joe Zhou: I'd like to discuss one of the hottest topics right now: AI. Over the past year, the market has had incredibly high hopes for the combination of AI and cryptocurrencies, but the current atmosphere seems to have shifted to a collective sense of uncertainty. In this new era of AI, what role do you think Ethereum should play?

Vitalik : At its core, Ethereum is a decentralized world computer. Its core attribute is "permissionless"—humans, companies, and AI agents all have equal access rights. This means that AI can hold assets, conduct transactions, and even participate in DAO governance on Ethereum. From this perspective, Ethereum is ready.

Joe Zhou: But the point of confusion for everyone is: where exactly do we connect them? How do we bring these two grand concepts to fruition?

Vitalik : First, we must be wary of a mental trap: don't combine things just for the sake of combining.

To give an example, even in the AI ​​era, our underlying TCP/IP protocol (Internet Protocol) does not need to be restructured because of the emergence of AI. The same applies to blockchain—as an underlying trust protocol, it may not need to undergo drastic changes.

However, if we look for the intersection between AI and crypto at the application layer, I think there are indeed several directions worth paying attention to.

1. AI's Bank Account . AI cannot open a traditional bank account. If an AI agent needs funds to perform tasks, crypto is its only option.

Second: Market prediction . AI can participate in prediction as a trader, providing more accurate information.

Third: Content authenticity : Use blockchain to prove whether the content was created by humans or generated by AI.

Joe Zhou: There's a popular term these days called "Vibe Coding" (meaning writing code easily with AI assistance, without focusing on details). Do you still write code by hand in your daily life?

Vitalik : Sometimes. I still maintain the habit of writing code by hand. My coding work mainly falls into two categories:

The first type is practical scripts, which are small programs I write for my own use, mainly to improve my personal productivity.

The second category is research and verification. When I'm researching some complex cryptographic algorithms, I'll write an implementation myself (usually in Python) to verify my mathematical ideas through code.

Joe Zhou: Do you use any of the popular AI programming tools on the market right now? For example, Claude, Gemini, or Manus? Which one do you personally prefer?

Vitalik : Actually, I'm not tied to any particular tool. I mainly use OpenRouter. It's an aggregation platform through which I can access all models. For coding, I still use some mainstream tools, such as ChatGPT, DeepSeek, and Gemini.

Vitalik in conversation with the 706 community: On motivation, initial aspirations, and ideals

(The following content is compiled from a joint interview between Joe Zhou and members of the 706 community with Vitalik.)

706 Community: What is your motivation for doing things right now?

Vitalik : My motivation mainly comes from three levels, or rather, three kinds of urgency.

First, it's to avoid the "doomsday scenario" of cryptocurrencies. My biggest fear right now is that the entire industry will eventually degenerate into a place for 100% cryptocurrency speculation, with only conjecture and no real applications. If that happens, the industry will die out of boredom as people get tired of it. To avoid this outcome, we must build real value—create better DAOs, create decentralized applications that truly penetrate all industries, and build more open DeFi.

Second, we need to improve Ethereum's technology. Frankly, Ethereum's current technology isn't good enough. While Layer 2 (L2) networks have solved scalability issues, they are still largely highly centralized. We need to make L2 more efficient and decentralized, allowing application experiences to truly catch up with Web2.

Third, if we fail with Crypto, the future technological world could very well be completely dominated by Centralized AI, which would be a very dangerous future. Crypto is our defense against this trend of digital totalitarianism and for maintaining diversity and freedom in the technological world.

706 Community: This is a hypothetical question: If you were to discard all of Ethereum's historical baggage and redesign Ethereum from scratch, what would you do?

Vitalik: Frankly, the technology roadmap won't change much. Because no matter how many times we start over, Ethereum's core goal remains the same: to be a decentralized application platform.

Before the advent of blockchain, the most successful decentralized network was BitTorrent. It was great, but it lacked a crucial component: a globally shared database (Global Shared State). So on BitTorrent, you could share files, but you couldn't create a currency, establish asset ownership, or run a DAO. Because without "state," it was impossible to record "who owns what."

Therefore, if I were to redesign it to fill this gap, the platform must possess two core characteristics: First, scalability. Without high scalability, on-chain costs are too high, limiting it to high-net-worth DeFi transactions and making it unsuitable for mass-market applications. Second, speed. Only with sufficient speed can a usable user experience be achieved.

706 Community: As an ETH holder, I would like to ask a pointed question: What do you think is the biggest risk that ETH is currently facing, but which is most overlooked by the outside world?

Vitalik: To be honest, I'm not too worried about the technical risks anymore.

What truly worries me is the application layer. The so-called "failure scenario" isn't network outages or hacking, but rather—that while thousands of applications are developed, upon closer inspection, none possess genuine social significance. The biggest risk is having the most powerful decentralized technology but using it to create a bunch of toys or casinos.

706 Community: Looking ahead to the next 5 to 10 years, what role do you think Ethereum will play?

Vitalik: I hope it will act as a central hub for all decentralized applications, serving not only finance but also penetrating all industries.

Its core value lies in "true ownership." Here, if you own something, it truly belongs to you. In the traditional world, you are always subject to a large company—they have the power to block you, the power to change the rules, and the power to charge exorbitant fees. We need to change this situation. This change is not limited to finance; it is also extremely important in areas such as social networking and identity verification.

Therefore, the primary prerequisite for Ethereum's success is that our technology (scalability, user experience) must be powerful enough to truly support these applications and become their solid underlying layer.

706 Community: Will the development of AI and quantum computers pose a 51% attack threat to Ethereum?

Vitalik: I don't think so. We need to clarify the concepts: the essence of a 51% attack is to attack the consensus and coordination mechanism of a PoS (Proof-of-Stake) system—this requires controlling 51% of the network's funds, not its computing power.

Quantum computing primarily threatens cryptographic signatures, not consensus mechanisms; as for AI, I believe it is not only not a threat, but rather something that can help. For example, AI can help us perform formal verification and discover code vulnerabilities, thereby making Ethereum more secure.

706 Community Member: Are you following Hyperliquid?

Vitalik : Actually, I haven't been paying much attention to it.

706 Community: You just mentioned that you hope to see more applications emerge. What type of applications do you most want developers to build?

Vitalik: The first is decentralized social networking (DeSoc). Current social media has huge problems. While most people dislike Twitter (X), the awkward thing is, we lack a sufficiently good alternative. Users are locked into the platform and have no freedom to "move" to another. We need to build a truly user-owned, portable social network.

The second is a "smarter" DAO. DAOs remain a very valuable concept, but we need to make them smarter than they are now. It can't just be about issuing a token and having a vote. Developers need to think more deeply: What are the specific goals of this organization? What kind of governance structure best matches these goals? We need to conduct more experiments and do things differently than before.

Third, it would be better if we had more decentralized stablecoins.

706 Community: Does "decentralized stablecoin" refer to the kind pegged to fiat currency?

Vitalik: The most interesting part is this—if we can achieve "de-fiat currency," that would be true innovation.

Joe Zhou: If it's not pegged to fiat currency, then what is it pegged to?

Vitalik: Anchor to real-world values. For example, anchor to the CPI (inflation indicator), ensuring your money is always worth the same amount of bread; or anchor to energy prices. This is true "stability."

Joe Zhou: This logic sounds very similar to Facebook's Libra project (a basket of currencies) back then.

Vitalik: Indeed. I think Libra's concept is good, but the execution went astray. They turned a vision of decentralized currency into a corporate version privately controlled by Zuckerberg. Because of Facebook's terrible privacy record, people instinctively feel fear. So we wanted to create something similar, but a decentralized version.

Joe Zhou: Have you been doing any interesting "personal experiments" in your life or in your use of technology lately?

Vitalik: (After a moment of thought) I'm trying to completely break away from X's official client. Now I mainly publish and browse content through decentralized aggregation protocols like Firefly.

706 Community: What is your ideal Web3 social product?

Vitalik: It doesn't necessarily have to have features that Twitter doesn't have; nor does it necessarily have to have new financial-related features, but it needs to be of higher quality than Twitter. The most important issue is not what features it has, but who the users on this platform are.

706 Community: Currently, most ordinary users don't really care about "data sovereignty," which may be why Web3 social networking hasn't taken off. At this stage, where do you think the solution lies?

Vitalik: To be honest, I don't know either.

Author's note: Thanks to Qiuqiu and other 706 community members for their help with this article.