CasperLabs’ Mrinal Manohar: “Proof of Work is Just Not an Efficient System”

Avi Rosten
  • 19 May 2019

Casper Labs

CasperLabs is a relative newcomer to the crypto space, but with an impressive technical pedigree.

With the aim of building a “fully decentralized, sharded and scalable next-generation blockchain,” CasperLabs is building a new proof-of-stake blockchain, the design of which is being led by Vlad Zamfir, formerly a lead researcher for the Ethereum Foundation.

While the PoS vs. PoW argument is far from settled in the crypto community, CasperLabs believes that PoS is the most viable means of achieving scalability.

I caught up with Mrinal Manohar, CEO of Adaptive Labs, CasperLabs’ parent company, to discuss the project. We talked about his history in the space, why he believes blockchains must move away from Proof-of-Work, and the company’s relationship with the Ethereum Foundation.

When and how did you get into the crypto space?

So actually it's a two part answer - I got into decentralization in 2002. I was an open source contributor to the BitTorrent project, not on the core BitTorrent stuff, but stuff like magnet linking on some of the applications.

And then I got into blockchain specifically in 2010. My roommate was one of the largest miners in San Francisco, but I started investing in the industry in 2012. So I started pretty early.

And when did you start CasperLabs?

So CasperLabs was founded on October 31st 2018. So it's fairly recent. Research into CBC Casper, which is led by Vlad [Zamfir] has been ongoing since 2014. So it's it's fairly developed. But CasperLabs, the R & D shop, that's releasing a blockchain that primarily works on the Casper protocol, is newer.

How would you describe in a nutshell what CasperLabs does?

“The ability to scale a blockchain without any compromise.”

What's happened is, in my opinion is that only Bitcoin and Ethereum are 100% decentralized because anyone can access the network at anytime - there's no permissioning, they are leaderless systems. What's happened since their creation is that in order to scale the systems, people have done it in one of two ways.

One way is some form of mild centralization, either through delegation, voting processes, randomization, etc.

The other way is through layer two solutions, and layer two solutions are really necessary - Visa has Square PayPal and Stripe on top of it for examples.

But we need layer one to be great.

Really where we're different is we're saying “no, you can do this with total decentralization.” We don't want to trade decentralization for scalability. Now, you might not get a million transactions per second, but you can get a few thousand, fully decentralized.

What's the motivation behind Casper's embracing of Proof of Stake, is it technical, environmental, both?

I'd say technical more than environmental, although there are environmental consequences. It is mostly technical, because if you think about it, the fundamental issue with proof of work is that it limits scalability, because 90 to 95% of your processing power is generating hashes, or random numbers. So it's just generating a bunch of random numbers. And once they're generated, they have no utility. So you spend 90 to 95% of your processing power, and it’s just wasted.

On the other hand, with a proof of stake system, you're reorienting the 90 to 95% that's wasted and actually doing useful work, such as securing the network, making sure consensus has happened, and processing transactions. Just fundamentally, because of that fact, there's a huge amount of throughput that's being wasted on a proof of work system. So proof of stake just by design is going to be more efficient, faster and more scalable.

Now in terms of the environmental factors. Yeah, it's not a good thing to waste 90 to 95% of energy, irrespective of where you are in the climate change debate, wasting energy is probably not a good idea regardless.

It's an added bonus.

The motivation really is we should have efficient systems and proof of work is just not an efficient system. Proof of work has gone away in a lot of cases before: most email spam filters were proof of work a decade or two ago, and now none of them are. Distributed denial of service (DDOS) attack prevention systems were all proof of work. They're not anymore. The reason why they exist early on in any industry is because they're easier to implement than proof of stake or other proof-based systems.

What about Lightning?

Lightning is a layer two solution. I think layer two solutions are great and they're going to really help blockchains in a similar way to how Square, Stripe and PayPal helped Visa.

But you need the underlying layer one to be efficient, because the amount of security you're giving up is inversely proportional to the speed of the network.

If the network's really fast you only have the batch a small number of transactions, if the network really slow, you have to batch a lot of transactions, but layer one is way more secure than layer two because it has way more validators. So you're giving up too much security to layer two if the underlying layer one isn’t fast enough.

So what you’re saying is it's better to get layer one right than trying to build a great layer two on top of shaky layer one?


I view layer one as sort of the foundation of a house. And I view layer two as the house. I actually anticipate that the end users are likely to interact with layer two more than layer one. And we're happy to just be a great layer one. I think layer two is very necessary. Industries grow because of that. Payments grew, because of all the stuff built on Visa. Uber etc. grew from Google Maps and Google Maps is used way more now because of Uber and Waze. Layer two helps layer one.

I view them as a necessary part of the ecosystem, but bad foundations are not sustainable.

What's CasperLabs' relationship with the Etheruem Foundation? Will there be room for collaboration in the future?

100%. We are open source purists and we are decentralization purists.

A lot of us on the team were early investors in Ethereum. So you know, we love it. We think it's a great platform. There's certain reasons why the transition to proof of stake might take a little longer than people would like. And we're launching a chain that's pure proof of stake. But if Ethereum wants to use that code, we'd be more than helpful in helping them use it. In fact, if anyone at Ethereum - and this is an open invitation - wants to speak to our developers and exchange ideas on how best to get the CBC Casper-like implementation, we're completely open to that. We view them with nothing but respect, we're standing on the shoulders of giants, trying to push the industry forward. It's not an antagonistic relationship.

I always find it ironic when people criticize earlier Blockchains like Bitcoin’s and Ethereum’s for apparent technical problems that their newer blockchains try and solve...

Yeah it's like saying, Charles Babbage and Alan Turing sucked, because their computers couldn't play games.

Which is how we view all the initial protocols. We're nothing but respectful and we want to collaborate as much as we can.

What do you have in store for 2019 and beyond?

 We should be feature complete by the end of the year, meaning we should be code complete and have the system up and running by the end of the year, at which point we'll go into a testing phase. The testing phase could be anywhere between four to 10 weeks, hopefully on the lower end. So we'll launch by the end of the year, or very early next year. We also have a whitepaper that people can read if they want to look at the tech specs. We're mostly heads down in development and partnerships.