Where’s Casper? Inside Ethereum’s Race to Reinvent its Blockchain

Screen-Shot-2017-01-18-at-2.13.15-PM
18 January 2017

If you’ve been following ethereum at all, you’re likely aware the blockchain project has a lot riding on something called ‘proof of stake’.

A radical change to how the network forms consensus, the idea is somewhat simple in aim, but potentially history-changing in scope.

The story goes like this: As far back as 2011, some developers began to worry about the energy that bitcoin – which uses a process called ‘proof of work’ (PoW) – would require at scale (some recently estimated that it would consume as much energy as Denmark). Further, bitcoin’s mining firms were already becoming big businesses, and there was fear that this change would negatively impact user growth.

Sure, there is the argument that proof-of-work’s competition created a more secure network (and that pegging mining competition to Moore’s Law led to natural market regulation), but some were eager for a greener alternative.

Enter proof of stake.

To date, the protocol has been embraced by several projects (peercoin, NXT and others) but ethereum’s transition would be unique in that it would essentially switch protocols mid-flight, so to speak. As a result, it’s faced heavy criticism from those who long have alleged that both PoS, and ethereum’s implementation, might not work.

The idea has been met its fair share of skepticism, partly because it’s taking a long time to implement a version that developers agree is secure.

In late 2015, Bitfury Group released a white paper exploring the differences between PoW and PoS, and the attack vectors that could have destroyed early “naive” versions of PoS.

Yet, against this backdrop, ethereum developers have continued working on an alternative, arguing that they’re making headway.

Work has gone through many iterations and materialized in an idea known as ‘Casper’, after the movie ghost, which ethereum creator Vitalik Buterin has described as “consensus by bet”.

Buterin told CoinDesk that the development process is now exploring how to incentivize the functioning of ethereum in a wholly new way, stating:

“The development process involves heavy exploration of both economic and game-theoretic considerations and Byzantine-fault-tolerant protocols, trying to create a protocol that satisfies one of several constraints simultaneously.”

While Buterin’s description may sound complex, what the statement essentially boils down to is that ethereum is still seeking to get its incentive scheme for its proof-of-work alternative right.

As far back as September, both Buterin and fellow developer Vlad Zamfir were working on competing implementations of the idea. Both are essentially betting that the cost of buying physical miners (as well as their competition) can be replicated virtually, given the right conditions.

Now, after years of research, Buterin said, he’s “quite confident on the general principle” though “details need to be worked out.”

Casper, explained

The team often holds public discussions on its research, and has posted three open-source simulations so far (the latest, posted in August, is here), which show how the protocol works and that it can converge on one history.

Recently, Buterin offered a research update covering the protocol, while Zamfir has been running a blog series on the history of his thought process. Developer Rick Dudley, computer science PhD student Loi Luu and others have also contributed to the protocol.

To recap, proof of work, used by both bitcoin and ethereum right now, requires a network of powerful computers to validate transactions, and proof of stake is another means to this end.

The goal of both is the same: for the nodes in the network to come to agreement on a correct transaction history.

PoS hasn’t been easy to implement, seeing several stages of development since 2013, when Buterin stated that ethereum would “likely” switch from PoW.

The first of these stages, which they now call “naive PoS”, suffers from what has been called the ‘nothing-at-stake’ problem. It doesn’t punish actors for validating more than one history — meaning the network could easily disagree on the real history.

“This is clearly a fatal flaw for a technology whose raison d’être is to provide an undisputed canonical order of transactions,” blockchain consultant Ciaran Murray remarked.

What separates Casper (and other more recent versions) from traditional PoS, is that it punishes participants who don’t play by the rules.

Buterin described it with a rough analogy: imagine 100 people sitting around a circular table. One person has a bundle of papers, each with a different transaction history. The first participant picks up a pen and signs one, then passes it onto the next person, who makes a similar choice.

Each participant only gets $1 if they sign the transaction history that most of the participants sign in the end.

“And if you sign one page and later sign a different page, your house burns down,” Buterin added, arguing that this is probably a good incentive to sign the right piece of paper.

Remaining problems

But Buterin and others are working on a few remaining problems.

They come down to “crypto-economics,” a word often heard in the ethereum community to describe what could be an emerging field of study, focusing on how cryptography can be used to help steer participants in a certain direction with the right incentives.

The best-known example might be PoW, where miners have an incentive to play fair and validate correct transactions, or else they’ll lose money. It drives miners to act honestly by paying them bitcoin.

Buterin and others are trying to get players to continue to play by the rules without the miners.

For it to work, the validators need to behave politely, Buterin said. But, there are some situations where it’s more difficult to concoct the right incentives.

One is that, once a block is ‘finalized’ or added to the blockchain, it shouldn’t be possible for a recent block to be replaced unless most of the validators (who are roughly equivalent to the miners) sign a second time to lose their deposits.

Another is that the validators should be barred from cheating competing validators out of their money (a problem called “bounding the griefing factor”).

All in all, the way Vitalik sees it, there are three steps left: finalize the algorithm, review it and build a testnet, and then implement it across all seven of the ethereum clients.

Murray said:

“In attempting to solve the nothing-at-stake problem, the cartel problem and the issue of finality, Casper promises to be, by a distance, the most advanced consensus mechanism to emerge in the public blockchain sphere.”

No deadline

Yet, it remains to be seen when that might be.

“I prefer not to give timelines,” Buterin said, though on social media he recently gave a perhaps too rosy estimate of late 2017.

If that’s the case, even PoS’s skeptics would probably rejoice.

“If there was a proof-of-stake algorithm, I would probably support it,” Bitcoin Core contributor Luke Dashjr said earlier this year on social media. Until then, he (and many others) consider it ‘vaporware’.

For now, the ethereum community appears eager to see a final version.

However, Murray suggested that even then it will probably need further inspection, saying:

”I’m looking forward to seeing a full specification of Casper available for peer review in order to ascertain if this complexity may create new problems of its own.”

Purple light image via Shutterstock