Decoder
CasperLabs Debuts Its Scalable, Proof-of-Stake Blockchain on Testnet
04.17.2019

CasperLabs, an open source blockchain project focused on research and development, announced the launch of its Node Version 0.2 today along with the blockchain’s native token, called Ectoplasm (inspired by Ghostbusters, aligning nicely with “Casper’s” friendly ghost connotations). The company has hired 21 engineers to work full-time on this project, 10 of whom joined the team in the past three months.

CasperLabs officially debuted on February 22, when the first version of its node software went public on GitHub and the company’s leads demonstrated the project in New York. Developers all over the world gained access to CasperLabs’ code and could start making their own test nodes locally.

That same day, the company announced its collaboration with Ethereum Foundation researcher Vlad Zamfir, now the lead architect of the “Correct-by-Construction” Casper proof-of-stake protocol (more on that later). In a press release at the time, Zamfir stated that he not only hopes CasperLabs’ scalable blockchain protocol is successful, but that it will “put pressure on other projects (like Ethereum) to adopt the technology.”

What is CasperLabs?
CasperLabs describes itself as a “scalable, next generation blockchain” that can be used by businesses, developers, and consumers. Its ultimate goal is to successfully build a secure blockchain that relies on proof-of-stake (PoS) instead of proof-of-work (PoW) using sharding. In other words, the aim is to make a blockchain that scales without sacrificing its decentralization. During the company’s debut event on February 22, CasperLabs cofounder Mrinal Manohar said that bitcoin and Ethereum are the “only two fully decentralized [blockchain] systems out there.” Both are notoriously unable to scale efficiently.

CasperLabs also wants to make its open source project more inclusive. While many blockchain projects use unique languages that developers wouldn’t otherwise know how to use, CasperLabs is using WebAssembly, which supports multiple widely used programming languages. Its core team of 21 engineers hail from all over the world, including China, Germany, Poland, Russia, India, England, and the U.S. “This is the highest caliber pound-for-pound team I’ve ever come across,” CasperLabs cofounder Steven Nerayoff tells us. A cofounder would say that, of course, but he insists the company handpicked engineers “with double PhDs” and careers spent working on Google’s main search, its most valuable product.

Get the BREAKERMAG newsletter, a twice-weekly roundup of blockchain business and culture.

Adaptive Holdings, a privately owned company based in the Cayman Islands, is the backer and parent company of CaperLabs (the research and development shop, not the blockchain itself, which is decentralized but being built out by the engineering team). CasperLabs’ cofounders, Nerayoff and Manohar, are the chairman and CEO of Adaptive Holdings, respectively.

“To date now, we’ve been internally funded 100 percent through the parent company,” says Nerayoff. CasperLabs will seek external funding down the road, says Nerayoff, including through a possible IEO, or initial exchange offering. Through an IEO, people won’t have to contribute that much cash in order to participate, encouraging individuals and smaller organizations to stake the network.

Where does sharding come into play?
“At a higher level, sharding is about taking some composite thing and breaking it down into smaller pieces such that you only have to care about the pieces you’re interested in,” lead Casper developer Michael Birch told us in March. In most blockchains, Birch explained, everyone involved has to be concerned with all of its transactions—even the ones they don’t care about.

With the CBC Casper protocol that Zamfir is developing, blockchain participants only have to concern themselves with relevant transactions. Through the protocol, people don’t have to download full blocks to get information from the blockchain. Instead, they can just “download block headers and can ask full nodes for specific information if they want it,” said Birch. Zamfir is also working on shards’ ability to send messages to other shards. (To be clear, the CasperLabs blockchain is itself an implementation of CBC Casper.)

The tricky part of sharding is choosing which chain to follow. In PoW blockchains, fork choice rule dictates that the longest chain is the chain to build on. “In the sharded protocol, you need to be a little bit careful, because if one shard makes a decision that affects another shard, they have to stay consistent with each other,” said Birch. The way Zamfir deals with this, Birch continued, is by having a “parent/child type structure” between shards, where one shard is dominant over another. If a dominant shard is finalized as a part of the chain, its “child” shard is, too.

“It keeps consistency but feels a little bit unfair because one shard can hijack the fork chain rule of another,” said Birch. To get around this, Zamfir is researching how shards can switch dominance, but this research is “ongoing.”

“I would say that it is unlikely that it will be part of the CasperLabs blockchain platform within the next year,” Birch told us in March.

What are the advantages of PoS over PoW?
You can think of the bitcoin blockchain as proof of concept for PoW. To add a block to the bitcoin blockchain, miners need to solve a complex math problem, which takes a lot of computational power (not to mention the resulting carbon emissions).

“If I’m going to be a miner for bitcoin, I need to go out and buy specialized hardware that’s going to do PoW stuff,” said Birch. “If I just use a run-of-the-mill computer, I’ll always be behind these specialized mining pools.” It’s hard to convince a group of people that’s already invested millions in specialized hardware to “throw it all away and buy tokens instead so they can become PoS validators,” he added.

In a PoS system, those who want to participate, aka “validators,” don’t need pricy mining equipment to take part in the blockchain. Instead, they need to stake tokens against the network. “We think PoS is where blockchain is going,” Birch said.

PoS is “not about the longest chain,” he added. In order to implement the PoS sharding structure that Zamfir and other Casper developers are working on, they’ve had to establish a “different set of rules that tell you how to pick [the blockchain’s] canonical history,” Birch said.

TheChain: Image

So what are CasperLabs’ rules?
“For the CasperLabs blockchain, we’ve chosen what’s called the GHOST fork choice rule,” said Birch. It’s not spooky, though it fits with the blockchain’s spectral theme. Rather, GHOST is an acronym that stands for Greedy Heaviest Observed Subtree.

Each validator is associated with a specific weight determined by how much they stake against the network. When one validator builds a block on top of another validator’s block, the weights of those two validators are combined to determine the heft of the chain. The more validators agree to build on that original block, the “heavier” that chain gets (hence the “H” in GHOST).

In the case of a fork, the heaviest chain becomes the canonical chain. The decision to choose the heaviest chain you see (“O” for “Observed”) takes place again and again, which represents a kind of loose representation of “greed” (the GHOST’s “G”). And instead of a fork, you can envision this like branches on a tree (or “Subtrees”).

What makes one validator weightier than another?
It’s all based on their stake. When a new validator joins the network, they bond a certain number of tokens to the network. That number, essentially, becomes their weight.

Validators in the network can “more or less” choose their weight, said Birch. “Ultimately, the security of the network is tied to how much stake the validators have.” This means choosing reasonable minimum and maximum bond amounts. The minimum creates a barrier to entry, which can’t be too high, or else it excludes people with more modest means. It also can’t be too low, so as to prevent people with bad intentions from taking part in the network without risking their personal finances. (There are other ways built into the network to prevent bad actors. For example, validators who attempt a so-called “bad action” can get their stake taken away from them.)

On the other end, the maximum controls how much sway a potential whale could have in the network. “You could imagine that if minimum and maximum amounts are close to each other, you’ll have lots of validators with roughly equal weights,” said Birch, “but if the maximum and minimum are far apart, [you get a] more diverse range.”

Balancing those two numbers to create maximal decentralization can be tricky, and exact values have yet to be determined. CasperLabs has hired an economist, Onur Solmaz, to eventually help with that. For now, Solmaz is focusing on CasperLabs’ token supply and inflation rates. “We are in the process of expanding our economic research team,” says Nerayoff.

What does the latest Node 0.2 news mean for CasperLabs?
“We have the ability now to actually start spinning up live nodes and stress test the system,” says Nerayoff. No one is contributing real money to the network yet. This will be the first time a Casper protocol has gone to testnet (developers have been working on the protocol since 2017). Meanwhile, Zamfir continues to do research for the Ethereum Foundation.

“While we are launching our own chain, we see our work on CBC Casper as the incubator work that Ethereum could also use in the future if it chooses to do so,” Nerayoff says.