The message was clear: do not use The DAO. The term stands for Decentralized Autonomous Organization, what was then a new crypto-based governance structure, pioneered in large part by the Ethereum community. In this case, the DAO offered a crowdfunding mechanism, designed to let anyone contribute to a pool of capital, and share in a new pool of tokens. 

At the time, many saw the Ethereum DAO as an inspiring alternative to venture capital. It was in theory. But in practice, it struck me as a system with enormous potential for failure. 

By the time some of the vulnerabilities we warned about were exploited, 5% of all ether was in a wallet controlled by the attacker, with another 10% at continued risk.

How it unfolded

At the time of the DAO attack, I was a computer science professor at Cornell University, teaching a cryptocurrency course during the industry’s infancy. The entire bitcoin market value was only about $10 billion, compared to $1.4 trillion today.

One evening, in the spring of 2016, I found myself at dinner with Ethereum researcher Vlad Zamfir, in a small French restaurant in downtown Ithaca, New York. Vlad told me about something new: a radical experiment in raising capital.

The first red flag I spotted wasn’t technical. It had to do with governance.

For starters, DAO participants couldn’t just withdraw their funds. You had to create something called a “child DAO,” go through multiple waiting periods and voting rounds, and then attempt to extract your funds. Such a convoluted voting system I feared, would lead to distorted incentives and catastrophic outcomes.

As far back as August 2014, two years before the DAO went live, my colleague Andrew Miller warned of so-called reentrant contracts in the code used to build it, which could allow attackers to drain funds. We decided the risks were too serious to keep private. 

So that May we began writing the document, A Call for a Temporary Moratorium on The DAO, highlighting the vulnerabilities. Three weeks later, the attack changed crypto history.

What happened

Imagine an ATM that checks your balance, dispenses the money, then deducts the amount from your account. In a normal ATM, this wouldn’t cause problems. But in the case of the DAO, the hacker discovered a way to make repeated withdrawals before the balance was updated. A bug in the smart contract led the blockchain to believe that, even after numerous withdrawals, the user still had funds available. 

Roughly $60 million worth of ether was drained from The DAO. 

Following the attack, a PhD student at Cornell and a member of the Initiative for CryptoCurrencies and Contracts (IC3), Phil Daian, published a detailed account of what happened. The market value of Ethereum was only about $1.5 billion, giving the attacker enough crypto to destabilize the entire ecosystem. 

The Ethereum community decided to reverse the transaction.

On July 20, 2016, a hard fork was implemented during the first IC3 boot camp. I joined Ethereum creator Vitalik Buterin, my fellow IC3 co-founders, Professor Ari Juels and Professor Elaine Shi, to celebrate the occasion by popping a bottle of champagne. 

But the hard fork also split the network in two. The chain that returned the funds retained the name “Ethereum” and most of the blockchain developers. The unchanged chain was dubbed Ethereum Classic, which still limps along today, though its market cap is far less than 1% of actual Ethereum.

The episode was chaotic, painful, and deeply divisive. But it also forced the industry to mature.

Lessons learned…and not

Before The DAO, most blockchain development operated under what I call “YOLO engineering.” Developers trusted their intuition, wrote code quickly, and assumed things would work out. That approach collapsed overnight. 

Audits became standard practice. Entire companies emerged to analyze smart contracts. Formal verification used in aerospace and military systems began appearing in blockchain research.

For me personally, the experience shaped my later work: when software controls large amounts of value, correctness is not optional. Years later, I built new blockchain systems with the lessons firmly in mind. 

And yet, many users still trust with very little information.

The rise of artificial intelligence introduces new risks. AI systems will discover vulnerabilities faster than humans can fix them, making exploits far more common. 

Yet I believe the world is ready to build another DAO on the Ethereum blockchain. Not the naive version attempted in 2016. A better one. The blockchain’s creator, Vitalik Buterin, seems to agree.

Today we have better security practices, more rigorous engineering standards, and a decade of academic research. Institutions now demand more reliable infrastructure.

Most importantly, we learned that popularity is not a substitute for technical correctness. Technology doesn’t care about social consensus. It only cares about code.

Ten years after the collapse, I believe we finally have the experience to build it properly. And this time, we have to get it right.