Michael Flaxman Interview

Q: Today I am with Michael Flaxman, a local expert on Bitcoin. Michael, you are a great proponent of Bitcoin. You have a background in economics from Brown University and experience of three venture-backed companies. You have built your own Bitcoin wallet and block explorer. Talk to us about innovation in digital currency. Digital currency has captured the media’s attention in the past year. Let us roll back the clock. To your knowledge, what was the environment that created this massive technological transformation you say is as big as the Internet?

A: It surprises me Bitcoin wasn’t invented sooner, since it utilizes old technology in a way that is extremely useful. The property gained with Bitcoin is censorship resistance; this removes a trusted third party from the equation. You can transact without the hassle to go to a bank or government and saying, “Hey, you update your database. I trust you will do that for both of us.” That is the magic of Bitcoin. Everyone can have a copy of the database and validate all the transactions. Participants choose to behave honestly because the incentive compatible system increases profit across the board.

Q: Most companies today start with a founding team, an entrepreneur, and maybe some software developers. Strangely, Bitcoin started with an anonymous individual. Do you think digital currency needed to emerge through a mysterious person?

A: The way it happened makes sense. Bitcoiners are known for being privacy-focused. If you go to a Bitcoin meetup, it is optional to give your name, and sometimes photos and videos are prohibited. I like the expression, “Bitcoin is the currency of enemies.” If you trust somebody, you do not need Bitcoin. However, if you do not trust him or her, it is extremely valuable. It makes sense for Bitcoin to come from this bizarre group of cypher punks who stray from the mainstream. These are people who use PGP, a program designed to encrypt emails to ensure the message is sent to the intended sender while it remains invisible to anyone else. Additionally, these individuals are most ideologically aligned with digital currency. They love open-source software and the idea to verify things themselves instead of trusting others. Did digital currency have to come about that way? Maybe not strictly, but it makes sense that the early adopters would be of that mindset and therefore shape the community.

Q: What about digital currency’s innovation since Bitcoin? We have seen many of these Altcoins come up in the last year. On a scale of 1 to 10, can you quantify the level of innovation with these new coins?

A: In terms of quantifying the variety out there, it is hard to keep track of. Sites like CoinMarketCap list their coins by value. They track hundreds of them and many are still excluded. However, in terms of signal to noise, the vast majority of these coins are simple forks. An open source software project is freely available on the Internet. Anyone can copy it and make a few changes. This is called forking software, and it has been happening since the beginning of open source software. In the land of cryptocurrency, participants have an incentive to tell everyone, “Buy my fork because I have made some improvements.” Most of these forks do trivial things. Litecoin is probably the most popular one, where they have two-and-a-half-minute blocks instead of a 10-minute block time. Instead of using the SHA-256 hash function, they use one called Scrypt. Charlie Lee created Litecoin over the weekend, and it is now worth billions of dollars. Would you call that innovation? I wouldn’t.

I don’t mean to pick on Litecoin, but it is the most well-known. There are other coins whose code bases were written from scratch. I think of a project like Monero that uses a different protocol for everything. Even their Base58 encoding is unique. You can judge their tradeoffs on the merit of their accomplishments. However, in a sense, everything is a fork of Bitcoin. Bitcoin made these other inventions possible. They borrow ideas from Bitcoin, and the only question is whether they took the code base.

Now, we are seeing direct forks of Bitcoin. An example is Bitcoin Cash, where they took what is called the UTXO set (essentially an address snapshot) and said, “Okay, we are going to emulate Bitcoin and give our coins to its holders in hopes that they use them.” To me, Bitcoin is where I see the focus, the best developers, the smartest people in the space, the most second layer applications, and work in areas I am personally excited about. I tend to tune out much of the non-Bitcoin news because for many years, there have been promises of something 1% different that will change the world. The ideas are often incentive-driven. The person who made the first version owns a ton of it. They want you to believe it will change the world because they would become fantastically wealthy. In turn, their followers subscribe to the same scheme (although they may not become as wealthy), and the people under this second generation would get rich as well. And on the way down we go, until someone is left holding the bag. I caution people to be extremely skeptical with anyone who says they have something new. To be 10% better and competing for currency is not enough incentive to build a better mouse trap.

Q: Can you tell us about Bitcoin’s current innovations?

A: Bitcoin has substantial active development. People often don’t realize that, or do not understand Bitcoin because it’s so complicated. I think the trickiest part about Bitcoin is explaining something to the masses that involves game theory, computer science, and economics. I have read endless comments on Twitter from people who lack a basic understanding of protocol. Although it is not expected of them to possess this understanding, their remarks lead to loads of misinformation. Bitcoin is open source and the most decentralized Blockchain by far. Any five people could get hit by a bus tomorrow, and Bitcoin will be totally fine. You cannot say that of any other cryptocurrency. As far as innovations go, the most exciting changes are made around the hardest problems: privacy and scale. Bitcoin has numerous impressive accomplishments. We started signing and verifying signatures with OpenSSL, and the program (while free and open source) is not optimized for this problem. As a result, Bitcoin created a new faster library called libsecp256k1. The secp256k1 curve Bitcoin uses served as inspiration for the name. libsecp256k1 speeds up signature verification by a factor of about 6.

Bitcoin’s mantra is “do not trust, verify.” The bitcoin blockchain is nearly 200 GBs (not all of it is signature) and you want signature validation to be fast. This is a classic example of something that nobody really sees. Most of the low hanging fruit has now been captured, so it is not easy to keep duplicating these gains, but a lot of work has gone into that. We have switched from uncompressed to compressed keys, in version 0.6 of the Bitcoin software. Having segregated witnesses is incredibly powerful. It allows so many things, but it is thought primarily of as a scaling solution because it effectively doubles the block size. However, it fixes transaction malleability and allows script versioning, which will enable things like Schnorr signatures on the network. It is useful for Lightning Network transactions because instead of having one-way payment channels, you can now have multi-hop payment channels. There are a host of other benefits with SegWit. For example, you can manage the UTXO set and provide better incentives for people to spend multi inputs and pay fewer outputs. In the past there was always an incentive to expand the UTXO set. Segwit is a very complicated technology, and the way it is done in soft fork is incredibly clever. It uses the same mechanism as BIP16, which is how we got multisig or pay to script hash (another very cool technology that is for security). There are a ton of active developments in Bitcoin, and I expect we will only see more. The protocol rules are fairly solidified though, which is a good thing. It is important to that know that the thing you are buying will not change next year.

Other cryptocurrencies are not decentralized. They have planned schedules. Every six months or so, depending on the currency, they change the rules. This can be a minor change, or it can be a major change, but somebody has the decision-making authority to do that, which is terrifying. You do not want to buy something where the inflation schedule is not fixed because you might find that you now own a smaller percentage of that currency. There are many cryptocurrencies that have not figured out the inflation schedule yet, whereas Bitcoin has known from day one. So, if you are running node software, you can always verify the rules yourself.

Q: In most start-ups that hire developers, they pay them in some mix of cash and equity. With Bitcoin, what incentives do developers have to continue to innovate?

A: The biggest trend in this space is that it’s very hard to hire somebody with at least five years of experience programming in Bitcoin. This is because if they have been working with Bitcoin for five years, and they have any financial resources as a software engineer, they should have been making good money. They probably bought Bitcoin, so they might not need money anymore. There are a lot of companies with software engineers who are retiring, which is very weird. Also, there is not much overlap between software engineering and cryptography, so the demand has skyrocketed.

If you look at the people who are involved, you can tell they would be involved regardless of the price. They were involved when the price was in a long trough from early 2014, when we were at 1200 a coin, all the way down to 200 a coin over the following two years. You have guys like Adam Back, who is cited in the original Tor whitepaper and the original Bitcoin whitepaper. He is a cypher punk at his core. I don’t think he is waking up every day asking how he can profit off Bitcoin. He believes this is world changing software, and it is something he should be a part of. I don’t want to speak for him, but I imagine that’s what he would say. There are a lot of people like that in the community.

Another point is that Bitcoin is one of the most impressive software project out there. It is a giant honey pot for bugs. If you can find a bug in Bitcoin, and you can exploit it, you can make a ton of money. This is also why Bitcoin is considered more secure, because it’s battle tested. It’s worth the most, and has been around the longest. If there was a huge vulnerability, it would have been discovered by now. If you can contribute to that you can get a lot of credit in the community. You can also command incredibly high consulting rates. This is a strange model where it pays to contribute. However, everyone is motivated for different reasons. There are contributors who are completely anonymous, so nobody has any idea who they are. There are contributors who are pseudonymous: they reuse an identity, and they are known online by that identity. Then there are contributors who are publicly known. There are also some people that fund the development. There is Chain Code Labs in New York where a few wealthy individuals pay the salaries of Bitcoin core developers just because they want to. They probably also hold a lot of Bitcoin.

It’s a bizarre ecosystem. There is a strong incentive, if you have the skill-set to go launch an ICO, because in this funding environment you can raise so much money. Bitcoiners who have been in the space for a long time are passionate about it as a technology that can change the world, and the ICO crowd usually just wants to make quick money and move onto another project. There is some overlap, but the groups have separated further over time.

Q: Is the code all open source?

A: Yes, it always has been.

Q: Is this a new paradigm where there’s a lot of innovation in open source software – instead of locking up the software inside a vault, like Google, Facebook, and Amazon?

A: I think it feels new to a lot of people. However, even inside of Google and Facebook, they use a ton of open source software. Facebook is super famous for React. Python is a language that was made popular by Google, as is Golang. This model of proprietary software languages has been on the decline for decades. One misconception outsiders have is they think open source software means that someone can see your data, which isn’t true at all. They just know your programming language. It is the equivalent of them knowing that you are running Windows, but it does not mean they can read your files. Open source software always out-performs closed source software because it has far more eyeballs on it. Contributors are also very passionate about it. They want it to accomplish things, so it’s very bottom up. It’s a great thing to see in cryptocurrency, and we are seeing improvements in a lot of cryptography libraries because of this. The boom in Bitcoin has led to a boom in applied cryptography interest.

Q: Let’s talk about ICO’s as an alternative funding mechanism. Is this something that you think is going to change the way innovation happens?

A: It’s making changes, and there are some good aspects to it. Overall, the industry is still immature. You won’t see basic protections like liquidation preference or proper vesting. It’s set up to bring out the worst right now, but there are some potential problems that can be solved – such as liquidity. There isn’t a great reason why, if you’re an early employer or investor in a start-up that becomes successful, five years later you cannot realize some gains. There are a lot of reasons why that doesn’t happen, one of which is the factor of negotiation. The company generally negotiates for something called a ROFR (right of first refusal) which makes it very hard to sell your stock. There are also specific lockup rules here; you cannot sell right before an IPO. If you’re an outsider, which someone in that position is generally considered, then they would have a lockup after the IPO. Some of these things exist for good reasons.

The ability to access liquidity is something good the ICO craze has brought on. However, a lot of those issues are actually more regulatory in nature. You don’t need a censorship resistant, decentralized platform to do this because the companies themselves are very centralized. They could just use a SQL database. There is no reason why they need this insane level of complexity, and there are all kinds of negatives in this. Roughly one percent of all private keys are lost every year as private key management is incredibly complicated. This process differs from that of your traditional bank, where if you lose your password you can just go to the branch and set up a new one; instead, this is a cryptographic signature. You can think of a password at the bank as an authentication, but a cryptographic signature is more akin to encryption. If you are looking at encryption versus authentication, the first is bound by math and the other is bound by a permission system. Thus, you can go to a bank and say, “Hey, update your permissions to give me access.” They may check your I.D or may need some physical proof that you are really an officer of the company, which is a straightforward process. If you die, it’s not like your bank balance dissapears with you. Your children will have access to it. In the land of cryptocurrencies is all about the cryptographic signature. The signature is everything because you can spend funds with it, but the funds are impossible to access without it.

Imagine you’re an early shareholder in a company like Uber and you get hit by a bus. How does your family get access to your shares? Or, you are really worried about such a scenario, so you take your shares and make copies of the private keys with control to those shares or tokens. Then, you hide them all over the place and some malicious third party says, “That is worth a lot of money,” and they break into your safe at home and steal it. We’re not used to having the custody of those kinds of assets ourselves, especially when the dollar amount gets very large. I think people are generally comfortable with the idea that they might have $1,000-$10,000 of gold at home. Obviously, they would have a safe. They wouldn’t want to go around town bragging that they have all this value, but it’s within the natural human paradigm that you can sleep with it under your mattress. However, we’re talking about seriouswealth in your brokerage account; your life’s work and your retirement. In that light,  this becomes a pretty scary concept.

There are lots of negativities when you introduce a Blockchain. The joke is, “If you use a Blockchain to solve a problem, you’ll end up with two problems.” Blockchains are very good at getting censorship resistance, which is the reason for its existence. But, you give up scale and user experience, so you have a ton of issues with a company. Why you do comply then? Because, maybe you’re not allowed to sell to somebody who is on a government list or resides in a banned country. Well, Blockchain has no idea of any of those things. Maybe a judge wants to reverse something and says, “Give that back.”  Well, how do you give it back if you can’t generate the cryptographic signature the Blockchain recognizes? Tokens use blockchains designed to solve the problem of censorship resistance to do something totally different, which feels off to me. A better solution would be more sane regulations, which make it easier to sell shares to non-accredited investors. Then, we wouldn’t use this complicated rube goldberg machine process to accomplish some other goal.

I hope that, in the long-term, the ICO brings about better access to liquidity and perhaps less regulations about who can buy stuff. Obviously, you do not want widows and orphans investing their life savings into scummy ventures to lose their money. You also want freedom for people to make those choices. However, the current way of getting around the rules seems a little bizarre. It could be very negative for the token industry if we were to see more SEC enforcement.

Q: So back to the technology. What are your thoughts on criticisms of Bitcoin, such as the cost of electricity usage, the problems with Proof of Work, or the accusation that it’s a socially wasteful math problem? Can you comment on Proof of Stake vs. Proof of Work?

A: Let’s start with efficiency and proof of work, then I will move on to proof of stake. As far as efficiency is concerned, I think it depends what you compare it to. The expression is, “hard money is hard because it is difficult to create.” If you look at how gold solves this problem, we spend an exorbitant amount of money. Further, digging up gold from the ground and then putting it back in the ground is very harmful to the environment; you can think of Fort Knox as the best example. Gold extraction in some ways is a silly exercise and terrible for the environment. We do it simply because gold is perceived to have a lot of value. This is due to many of the same reasons for Bitcoin’s perceived value: it’s portable, divisible, durable, scarce, hard to counterfeit and fungible. Bitcoin is very much trying to be the digital version of gold. I like the expression “nerd-gold.” You have to make it hard to create, which means to receive the marginal revenue available for creating a Bitcoin, you need to be willing to spend that marginal cost. Electricity is a very even, clear way of measurement.

What is so cool about Proof of Work is that you do the work, but you also prove that you did the work. Everyone can measure how much work is being done. The best attack on Proof of Stake is something called “stake grinding” where you obfuscate the work that you do. Stake grinding is running a complicated simulation where you try different transaction histories that might grant you the right to mint future coins. You might have to run billions and billions of computations to try to come up with a version of history that works out more favorable to you. If you succeed at doing this, you collect more cryptocurrency and more coins. You have the same game where you have your marginal revenue equal to your marginal cost. As long as your marginal revenue is higher than your marginal cost, that attack is worthwhile.

With Proof of Work, I know exactly how many resources you spent to attack the network because you have proved you have done the work to do it. With Proof of Stake, you can spend an obscene amount of resources to present a version of the Blockchain history that’s favorable to you. Nobody can proove whether that just happens to be the calculation, which happens to be favorable to you. Furthermore, it’s even harder to tell the attack is specifically favorable to you, if you did not do much work, or if you got lucky. With Proof of Work, those details are transparent.

One more thing to keep in mind: the amount of value being secured is massive; measurement is not on a per transaction basis but rather on a per value basis. Thus, sometimes people will say, “Bitcoin is only doing X transactions per second Y work. So, each transaction’s cost is the amount of electricity for a small family,” or something like that. This argument assumes that it’s the only transaction. However, one of the amazing things about Bitcoin transactions is there are all kinds of ways one kind of transaction could represent a much larger number. The simplest example is if you think of it as a settlement system. Maybe two Bitcoin banks, which are like a Coinbase and a Bitpay, are settling up at the end of the day, the end of the hour, or at the end of whatever period of counter party exposure duration they want. Therefore, they say, “Okay, I am sending you 50 Bitcoins,” which is “one” transaction. But, behind the scenes it represents the next settlement of thousands of transactions. Thus, we would say, “Well, that transaction uses a lot of electricity, but it was actually the net settlement of ten thousand transactions.” You could see a world with a Lightning Network, which is the new thing coming online now, where these numbers get really large. Therefore, you might do fewer transactions on the main chain but would be settling far more than were off chain. Net settlement is very common in other technologies (think of a bar tab!), there’s no reason to assume it will be different here.

Q: Most of the new ICOs are more based on Blockchains not Bitcoin. Is that right?

A: Yes. Although, I think when people say “Blockchain, not Bitcoin,” what they mean is a private permissioned ledger. Mostly, ICOs are just using a different Blockchain or a different ledger for their Blockchain. However, they are still Bitcoin-like in that there are miners, there is inflation, and it’s supposed to be trustless. Although, in the beginning there is a centralized team. The idea is they will shepherd the project and get it to a point where it becomes decentralized. However, this is very hard because the world tends to skew towards centralization. Thus, it is very hard to go from centralized to decentralized. There is a lot of incentives that push you from decentralized to centralized, but their pitch is usually that we are going to be decentralized like Bitcoin one day if we succeed.

Q: Is that synonymous with saying the best use case for Blockchain is digital gold or store of value, instead of identity management, land titling, or having 911 all over the world?

A: Yes. For example, if I had a ledger that anyone in the world could read and verify, and it was impossible to forget, what would be the most valuable use for that Ledger? The answer would clearly be value itself. In other words, the killer app of Blockchain is Bitcoin. However, as far as those specific examples, there are reasons why they might not make sense for ordinary users. For example, if Grandpa loses the private key for the title to your house, is the house no longer his? That seems bizarre. If a judge says, “You may have sold this house, but the person you sold it to is a criminal or paying with ill-gotten gains how do you reconcile that? In the Bitcoin community, the expression you will often hear is that the code is law. The idea is the Blockchain is always correct.

Bitcoin itself has a value that is totally new. Gold is a classic example of a bearer instrument from the past. They used to call it coupon payments from bonds because you could literally clip the coupon to redeem the payment. We used to have stock certificates, where you would give me the money and I would give you the stock certificate. Obviously, they were forgeable and unsecured on both sides of the transaction, but they were used like dollar bills. If we look at these system, theyit don’t really make sense for censorship resistance. Land itself is highly centralized and censorable; if your Blockchain disagrees with the court they are going to have no problem sending a sheriff to evict the new “owner”

However, identity is unique and cool, so there could be an opportunity there. One thing that makes me cringe though is watching the older generation log into their Amazon account and seeing that it’s such a difficult process. I wonder if they should be holding the private keys to their identity because bad things happen. If you lose your private key, there is no one to call to do a password reset. I know the Blockstack protocol has talked about doing something where you can have multiple people vouch for you, which would be nice to have in a protocol. Fundamentally, you are pulling identity away from big companies like Google, Twitter, Facebook, and LinkedIn – where my identity is right now. It’s not ideal that those are my identity providers. Now, there are more decentralized ways to do that. I could publish a fingerprint of my PGP key, which is a lot harder to cheat. I could also use the the Web of Trust, but for the average user, Facebook is who they are, Twitter is who they are. However, those are privately controlled companies that could decide to block people whose values they disagree with. So, there is value in the idea of decentralizing your identity.

However, I‘m not sure if I’m optimistic that end users will manage those keys. If your hard drive fails and you don’t have a proper backup, then you would lose your identity, which seems bizarre. The worst-case scenario is that you lose everything. You can get a new driver’s license, but the government wouldn’t  say, “Oh, you forgot your passport? You can never have a passport or driver’s license again.” Those are all recoverable things. The idea of starting out with a new identity ten years into your career because you lost your password just does not seem like the right solution to the problem.

Q: Thank you Michael. Those are very interesting comments. Do you have any other thoughts on Innovation in the digital currency space?

A: Yes. I think the biggest problems have always been privacy and scale, and we will continue to see privacy and scale. We also see it on altcoins like Monero, ZCash, and MimbleWimble protocol, which are all trying to implement better privacy. I think that’s the billion-dollar problem. This audience hopefully includes students here, such as computer science or math students, and I would say this is an incredible opportunity to focus on solving that problem.

Q: How would they get started on that?

A: The material is quite complex, and the learning curve in Bitcoin is steep. Fortunately, there are many free resources. There are books on programming and mastering Bitcoin that are really good. There’s also Jimmy Song’s, “Programming Blockchain,” which I’m excited to help teach in a couple of cities. Both of those have free versions online in some limited form. Jameson Lopp also publishes a list of free Bitcoin resources that’s good. Strangely enough, one of the best sources for info about what’s happening in the cryptocurrency community is Twitter. All the code is open source, so you can jump in anywhere.

Q: Thank you for visiting us at the Mays Innovation Research Center.