In many ways, the crypto currency Bitcoin had a very bad year in 2014. It has fallen from a value of over $1,000 per Bitcoin last January to just over $300 today. This volatility and the attendant speculation has led some critics, notably Nobel Prize-winning economist Paul Krugman, to consider the cash alternative to be a form of "the long con." In positive economic terms, he questions whether Bitcoin can be a stable store of value.
I often agree with Krugman, but I sense that his deeper unease is with what he terms the "normative economics:" is Bitcoin a good idea for society? He identifies his unease with the observation that, "Bitcoin fever was and is intimately tied up with libertarian anti-government fantasies." It is hard to ignore the libertarian agenda behind Bitcoin and yet it is also hard for me not to see programmable currency as a necessary building block of the 21st century global economy.
My use of the word "block" is intended to evoke the foundational technology that underpins Bitcoin, the block chain. For the uninitiated (that's most of us), there is a public ledger of Bitcoin transactions that are published to a distributed network of participating computers. Once every minute or so a batch of transactions are released in a "block" to be validated by the network and added to the "block chain." A unique cryptographic "hash" identifies each block and transaction and permanently fixes them in chronological order.
To confuse matters slightly, block chain (two words) is different from Blockchain (one word capitalized) which is a software and wallet company that hosts a real-time browser of Bitcoin transactions. Blockchain, the company, just secured the biggest capital raise ever in the annals of Bitcoin. Indeed, as Biz Carson writes in GigaOM, "investment in bitcoin is up as the price goes down."
Block chain, the concept, is even more important than Blockchain, the company. One way to understand why is to consider one of the most ambitious projects related to Bitcoin, Ethereum. Ethereum is an open source platform for smart contracts built on top of block chain technology. It recently completed a $15-18 million round of crowd funding using its own value token, poetically called "ether." Stephan Tual, CCO of Ethereum, describes block chain technology as "a method by which trust can be reached on an open network." Bitcoin is the first and best know use case, but Ethereum, says Tual, "makes this technology applicable to just about anything else." Tual describes Etherum in the video below:
At the center of Ethereum are contracts that Tual says are "programs that follow a series of steps every time they receive a message called a transaction." Ethereum itself is a "programmable distributed network," in which all manner of transactions and peer-to-peer agreements can be automated. "The fact that Ethereum is by its very design fraud and tampering resistant," Tual continues, "means that it offers a new range of solutions to everyday problems that are currently solved at great expense." According to Tual, these include voting machines, domain name registration, registration of legal documents, medical software and transfer of goods and services.
Networking innovator Paul Baran
If we are getting too far out into the ether here, let's take a step back in time to the 1960s and one of the most important figures in the history of computer networking. Paul Baran was an engineer who joined the RAND Corporation in 1959. According to his Wikipedia entry, "Baran took on the task of designing a 'survivable' communications system that could maintain communication between end points in the face of damage from nuclear weapons." He went on to become one of the inventors of the packet switching network, one of the key technologies that underpin the Internet. He also articulated the key differences between centralized, decentralized and distributed networks. In terms of "survivability," centralized networks expose a vulnerable "single point of failure" not present in less centralized networks.
I had been following bits of news about Ethereum on HackerNews for months, but Toronto tech advisor William Mougayar put the whole picture into perspective for me with a recent post on his blog. In The Blockchain is the New Database, Get Ready to Rewrite Everything, Mougayar describes five critical concepts of this new distributed application space: the blockchain, decentralized consensus, trusted computing, smart contracts and proof of work (or proof of stake.) These components comprise a platform that makes all manner of automated transactions possible.
This way of thinking about automation is where I think that both Krugman and the libertarian fantasists miss the boat on the implications of Bitcoin. Yes, the anonymity of Bitcoin enables illegal activity and seems to promote built-in tax evasion and asset sheltering. But the permissionless nature of block chain platforms (there will be many) also enables flexible types of peer-to-peer economies that will be necessary to cope with the social disruptions of machine intelligence automation. I concluded my post Monday about deep learning and machine intelligence with a reference to this necessity:
This will not be the year that 80% of the developed world loses their jobs to intelligent machines, but it is not too soon to start figuring out what to do about that eventuality. These same machine technologies can help us redefine and redistribute human value, and we will need to use them for that as well.
If you believe the leaders of Silicon Valley, the falling need for human labor will virtuously coincide with a radical drop in the cost of human necessities. But if and when that future arrives it is bound to not be evenly distributed. How will we, collectively, figure out what to do with human labor and how to distribute essential and non-essential goods and services in a "post-scarcity" world?
The answer, I think, has something to do with distributed applications on networks that have independent trust mechanisms built-in. Centralized networks, like banks and governments and large corporations, change too slowly and are too vulnerable to attack even at their current rate of change. This limitation does not mean that these structures can and should go away. That's the libertarian fantasy. Realistically, centralized structures will coexist with decentralized and distributed systems with each addressing the appropriate scale of transaction and mutability.
To get a clear picture of what these different types of networks are, the diagram by Baran in the image above can be seen as a description of three phases of the web. Web 1.0 was dominated by large walled gardens like AOL and CompuServe. It sucked, but it provided an on-ramp for millions of non-geeks to consume content on the internet through these centralized providers. Web 2.0 is both social and transactional. Although Facebook is an enormous network, each user only partakes of their own decentralized hub of friends and preferred brands. And Amazon is "the everything store," yet it is also a front for millions of vendors that use its network for marketing and distribution. The emerging distributed web 3.0 is computational in nature. Instead of a social graph calculated by a network like Facebook, or a commerce graph negotiated by Amazon, distributed networks allow individual nodes to calculate their relationship independently to other nodes.
If this is sounding a lot like how machine intelligence works, this is not coincidental. What makes learning systems possible is rapid adaptation. To quickly and efficiently allocate resources in a world disrupted by widespread automation (not to mention climate change and ethnic conflict) requires systems that can self-organize and transact. Yes, this is scary, because to most of us these are all black boxes doing mysterious things.
Why should we trust these systems? I have a simple but counterintuitive answer. We should trust automated distributed networks because they obey the same laws of physics as all of nature. According to my friend Adrian Bejan at Duke, as long as flow systems have the freedom to adapt, they will evolve in ways that more dynamically distribute their contents over time.
Bitcoin itself is at an evolutionary crossroads. This crisis explains why investment in the platform is up although the currency itself is down. What makes Bitcoin work is that the owners of individual computers (or collections of computers) voluntarily "mine" the network in return for rewards. The computational work involved in mining is the "proof of work" that validates the transactions and allows them to be added irrevocably to the block chain. The calculus of all of this is mind-numbing, but the rewards decrease over time as computing power is assumed to increase.
Because of its structure, Bitcoin has a built-in scaling problem. Each node on the network has to keep an updated copy of the entire block chain. If Bitcoin remains a niche currency, the size of the block chain might keep pace with the cost of computation and storage. But if it were to expand quickly it would easily require unjustifiably large amounts of energy to maintain. The larger part of the comments to Mougayar's post concern these scaling issues.
Ethereum is confronting scaling directly on their emerging platform. The video below contains a presentation by two of its founders on the different options they are considering. In general, director, co-creator and inventor of Ethereum, Vitalik Buterin, emphasizes that the group is trying to build "a system that works by scaling out not be scaling up." So instead of Bitcoin itself becoming an unstable Jenga tower (each "block" adds to its "height"), Ethereum is looking at different architectures that are less redundant, but still secure.
Finally, just to give you an idea of what this all might be good for in practical terms, there is a project from Ethereum called the Mist browser (see video below). Instead of making a plugin for Chrome to access Ethereum apps, chief designer Alex van de Sande decided to build a dedicated browser. Mist is more of a workbench than a content explorer, but van de Sande's intention is to make it immediately useful to developers and non-developers alike. In the examples in the video he shows how to join a crowd fund campaign and make a marriage contract within the system without the permission of any centralized authority. What is striking to me is how Mist makes the selection of identity and funding options highly transparent and how it gives you clear ways to confirm your intentions at every step without feeling onerous.
These are still early days for distributed applications, but 2015 will mark a turning point for this approach. As with machine intelligence it is still unclear what the surprising "killer app" will be that will make block chain-based apps make sense to civilians. The ability to make smart "things" capable of autonomous transactions could be significant. The built-in financial incentives for mining are a big part of what has driven the Bitcoin phenomenon so far and I think these will drive block chain adoption as well. Most people in the developed world are carrying massive computation power around with them all the time, but few are using this fact to improve their economic circumstance. Simple feature phones in the developing world have had far greater economic impact to date. Look for that to change in 2015.
Bonus Round: See William Mougayar's follow up post, Blockchain Apps: Moving from the Jungle to the Zoo.
And this cool explanatory video about Bitcoin from Vox.
