Sometimes you get lucky. Researching a story on bitcoin, I went to a meeting in Roppongi to meet enthusiasts. Vitalik Buterin, who is central to the Ethereum project, happened to be there. After his speech at the meeting, he then happened to come and sit with me. He also allowed me to record the 45-minute conversation we had about Ethereum, cryptocurrencies and the direction in which the Internet is heading.

Others were present while I spoke to Buterin, and one person in particular changed the subject of the conversation quite dramatically. That part of the conversation is marked with a “*.”

Here are three things to know about Buterin:

  1. He is not the most outgoing person. Read the words on his T-shirt in this picture. He was wearing this to the meeting, where he delivered a speech.
  2. He is much less of an evangelist for cryptocurrencies than most of his peers in the community.
  3. He was, at the time, very much involved with Chinese developers. He had been staying in China for around 50 days at the time of the bitcoin meeting, which took place April 30.

Here is the conversation I had with Buterin unedited and in full:

How would you personally describe Ethereum?

It’s a blockchain for anything.

In general, you can think of a blockchain as being a computer. In bitcoin’s case, it’s a very specific computer, with Ethereum, it’s a fully decentralized computer.

The simplest practical use is crowdfunding, Kickstarter style. You launch a campaign and set a goal, and if you reach your target, you get the money, if not, everybody gets their money back. Theoretically, that could all be decentralized in Ethereum.

Even with things like Kickstarter and so fourth, there’s a lot of human effort goes into those applications, particularly if it is financial and you care about security. You need a lot of people to be involved, because lots of stuff has to be done over and over again. Here, you just have one platform and it’s already done. You can assume that everything on the computer is just going to keep on going. There is a lot of labor that can be saved and particularly a lot of monopoly profit that can be taken out.

How is cryptocurrency technology going to change the world? I’m not asking for a doom and gloom scenario, but more an agnostic assessment of the possible disruptions.

Up until now, technology has replaced menial jobs with high skilled jobs; its replaced jobs with unemployment; and its replaced jobs with jobs that involve interacting with people. The reason restaurants hire people is working in them requires a skill for dealing with people and you cannot replace that with robots. You can think of blockchain technology as part of that larger context, but it is somewhat different. Instead of just replacing factory jobs, this technology happens to also replace management to a substantial degree. It is also unemploying the rich people, instead of just the poor people.

What’s stopping Ethereum from doing that today?

Ethereum has to be released, get to the point where it is stable and get to the point where there are lots of interesting applications for it.

How about bitcoins and cryptocurrencies in general? We see a lot of evangelizing but I’m not sure I see much to suggest that they are about to go mainstream.

For blockchains, there are still technical improvements that still need to be made for them to become competitive. The specific ones that I am concerned about are first of all scalability. The way centralized systems work are one person reads a thousand books, with bitcoin, a thousand people read a thousand books. And once a thousand people turn into a million, it just becomes inefficient. Transactions in bitcoin cost 3 cents right now. People in the U.S. venture capital industry are in a bit of a dreamland right now, because they think bitcoin is better because PayPal charges 30 cents. Well, guess what? In China, you have AliPay that charges zero cents. The fact is, each payment transaction costs zero cents to process. At some point, if bitcoin gets anywhere, Mastercard and Visa are just going to drop their fees by 80-90% and they are going to squash it. Fundamentally, every node processing every transaction is just not the sort of paradigm that can go up to tens of millions of people.

Bitcoin right now pays $300 million a year for secure mining. At some point, that subsidy is going to run out. So either transactions are going to go very high or network security is going to drop off.

And transactions take 10 minutes to confirm. There is this concept of instant confirmation for bitcoin transactions right now. But when you buy something off a website now, there are ways to revert those transactions with maybe 5-10% effectiveness. Theoretically, you can even bribe miners to revert, and there’s no security against that at all. The reason why instant confirmations exist is because they are not actually confirmed. The reason merchants are willing to accept them is that even if we cheat them within 10 minutes, they can just cancel all your products.

There’s a chance for large-scale fraud then*?

The fact is, if you are in a restaurant, it is easy to run out without paying. It is very easy to cheat. Even still, people don’t cheat. It’s the same with bitcoin, people generally don’t cheat.

So what’s the problem?

Once you move away from just payments — to transactions in retail stores, to financial transactions, to running lotteries — people just expect a response time that’s faster than 10 minutes.

Anonymity is a big draw with bitcoin.

The state of bitcoin right now is basically an arms race, where on the one hand you have people coming up with clever algorithms that deanonymize transactions, on the other hand, you have people creating things like mixers that anonymize better. Ethereum 1.0 will be slightly less anonymous than bitcoin… fine, that’s a trade off. The one interesting project I’m looking at is Zerocash, which uses zero-knowledge proofs where you can agree to transactions but there is no linkage at all to where it came from.

When we look at the Snowden revelations, we can see that anonymity is becoming more difficult. Can cryptocurrencies deal with the sort of surveillance we live with today?

I want to see the zero-proof technology to let the big data people see the results of aggregation without ever actually seeing any of the intermediate data. It turns out that is possible in a lot of cases.

How did you end up where you are today? Was it a political thing that drew you to bitcoin or did you just like the technology?

I was initially interested in part in the politics and in part because it is cool technology. I was particularly interested in combining cryptography and information to get as close as possible to the perfect information economy, even though that is obviously as impossible as zero friction. But just like zero friction, people will try and get as close as they can.

What do you think happened at Mt. Gox*?

The theory that seems most credible to me is that Mt. Gox at some point in 2011 lost 200,000 or 300,000 bitcoins and Mark Karpeles decided this was really embarrassing and he was just going to try and make it back in fees. When the bitcoin price went up, it started running away from them. They may have started to speculate internally to make the money back; they got into very serious trouble at the end of 2013, and that’s when they started trying to manipulate the markets…

What evidence do you have that they manipulated the markets*?

Well, technically, I don’t. But they delayed withdrawals from Mt. Gox for four to six weeks in mid-2013. What that did was it incentivized people to throw bitcoin into the exchange because there was a price difference between Mt. Gox and the other exchanges. And at some point in February, the amount of bitcoin they had in storage at some point hit zero and they decided to blame transaction malleability.

What do you think of Mark Karpeles*?

As a guy who runs a company he is a complete failure. If my theory is true, that implies he acted extremely unethically. He wrote Mt. Gox in .php, which generally has a very bad reputation. It’s a language that is designed to make it easy to pop together a lab application in 10 minutes. By itself, that’s a good thing, but it also makes it [susceptible to problems].

There was one point after the big spike in April 2013, where the price dropped by about 70%. During that whole time there were a large amount of transactions and Mt. Gox basically just shut down… there technical infrastructure just broke. To some degree, this was extremely unprofessional because they should be able to handle 10 or 100 times what their current volume is, that’s just standard practice. It just seems that it didn’t really hire people to improve the code base. It could easily have been a one man job right up to 2013, when he probably hired a couple of developers.

So what are you doing with Ethereum to make sure similar problems don’t occur?

Our security policies inside Ethereum are incredibly paranoid. We have a three out of four multisig wallet, where two of the keys are in cold storage. One is inside a laptop inside a locked safe inside a locked house. To be honest, one of the things we learned is we optimized way too heavily for security against theft and not enough for security against losses. We had one semi-serious incident where if a couple more things had broke we would have lost access. Fortunately, we had a lot of redundancy and the redundancy did its job.

So you cannot say you are 100% secure? Is it even possible to be 100% secure?

It’s always a trade off between security against loss and security against theft and convenience. You can improve the trade-off curve with better technology and better interfaces. With Ethereum, we are going to try to launch with a really good multisig wallet and make it default. But still…

If this tech is to really take off, surely it needs to be like Facebook, where these days you have grandmas and grandads using it. Will that happen?

We’re trying to design Ethereum to be maximally grandma friendly. The point is that grandma, she isn’t using Ethereum, she is using some wallet or chat application or whatever else. And each of those applications, it’s their responsibility to make themselves user-friendly.

We are going to try and make as easy to use as possible is our security policy. What you have now is a system where you download applications and giving them root permission to run software. We are trying to get as far away from that as possible.

How about surveillance? Will you be NSA-proof?

We are trying as hard as we can to be NSA-proof. Our peer-to-peer protocol already has wire protocol-level encryption.

* Questions asked by someone else