Is ethereum's biggest future upgrade "fundamentally vulnerable?"

That was the claim put forward by distributed systems expert and founding VMware researcher Dahlia Malkhi on Friday, who used those exact words to describe ethereum's Casper protocol at Financial Cryptography 2018 in Curacao.

During a keynote presentation, Malkhi's covered blockchain through the lens of a long-studied area of computer science that explores how nodes come to agreement. Still, it was her critical remarks on Casper, an upgrade that has promised to introduce a better, greener algorithm for the ethereum blockchain, that perhaps drew the most notice.

Unlike bitcoin's pioneering proof-of-work system whereby miners expend real electricity and energy to verify blocks of transactions, proof-of-stake offers a kind of virtual mining, where the coins themselves are locked up (like capital in physical miners) and then perform computations.

As such, proponents say proof-of-stake will have the same theoretical benefit to the system, without real-world power costs and environmental issues.

But while the idea's opposition has been just as fierce, Malkhi's comments mark perhaps the first time a renowned computer science researcher has weighed in on the debate.

She told the audience:

Yet, she softened the blow, adding that she believes Casper has led to "interesting" research that's fueled growing interest and innovation in the field of distributed systems.

As someone who's been studying distributed systems for years, Malkhi's biggest concerns stem from the two properties that should be preserved by the technology – safety and liveness.

Within blockchain systems, these attributes help indicate if transactions will go through properly, and if the system itself will move forward with time.

"I had a conversation with [ethereum's Casper lead author] Vlad Zamfir yesterday," Malkhi said. "He argues, isn't it still useful if it's 'mostly' live?"

In Malkhi's mind, the answer is no, since there are vulnerabilities in how some blockchain projects, such as Casper, are currently laid out.

"We have several decades of experience here," she said.

Her remarks suggest her concern is that not enough traditional academic rigor has been applied to the idea, and that by doing so, key assumptions could be invalidated.

"Seriously, it's very easy to come up with a solution which is not live. It's trivial. The only thing we need to do in this field is generate mechanisms that are both safe and live," she said.

That's not to say Malkhi is not excited by cryptocurrency and the new research and innovation it's spawned.

Elsewhere in her talk, she agreed with the often-cited view that bitcoin helps solve a very important problem in distributed systems – the Byzantine Generals Problem – an insight she attributes to notable investors like Marc Andreessen.

Still, she's quick to say the work isn't over yet, calling Andreessen's comment "insightful" but also perhaps "overly optimistic." As such, her comments suggest she doesn't quite believe any blockchain is architected in way that's free from flaws.

She concluded:

Dahlia Maklhi image via pwlconf website