Computers are getting much faster and our mathematics skills are getting better too. This combination is starting to reshape blockchains like Ethereum. This will fuel additional scalability as well as leading to much more genuinely decentralized blockchain ecosystems and more powerful smart contracts.
Blockchains today already consume enormous compute resources, but they are still much more centralized and fragile than many of us would like. Advanced protocols depend on large servers, nearly all of which reside in a few powerful cloud ecosystems, and we’re still in the very early stages of developing truly advanced smart contracts.
Ethereum smart contracts today are typically around 24-25kb and many DeFi ecosystem depend on a web of multiple contracts. There’s no reason to think that we can’t see a future where smart contracts are in the megabyte size, including capabilities like embedded machine-learning models or complex decision trees.
The idea that we should have a 25kb limit on smart contracts will, in time, seem as antiquated as the 640kb main memory limit on early PCs.
To understand how these scientific improvements will change the world of blockchains, it’s worth looking at how we got here in the first place: blockchains use lots of computing power in a way that many would have, once upon a time, considered very wasteful. Again, if you go back to the early days of computing, memory and compute resources were so scarce that people left off the half the year number (The “19” in “1985”) to save space. A proof of work system with thousands of parallel processes would have been considered impossibly wasteful. The problem with blockchains is that they get their security and value from re-doing stuff repeatedly. Everyone is checking balances and calculations and verifying them and trying to reach consensus. If you could just pick one trustworthy party to manage the whole process, we could do this all with 99% less effort. The problem is that we are, currently, rather depressingly short of trustworthy central authorities.
Getting everyone to check each other’s results is something we just couldn’t do in the past because there wasn’t enough compute power to go around. I grew up in a home where computer punch-cards used by my parents in computers were always lying around and my parents had to book computer time the way some of us fight for a table at French Laundry. Fortunately, those days are long gone, and while I cannot assemble a program using punch cards, I do know how to make high performance paper airplanes with them.
Moore’s Law, the observation that computing power seems to double every 18 months or so, rescued us from punch cards. The result is that, as times goes, performance goes up at levels that are hard to comprehend. In 1970, you could get about 1,500 circuits on a chip and, by 2020, it was close to 50 billion.
When it comes to blockchains, this means we can trade something that has become very cheap – computing power – for something very valuable, which is trustworthy data and results. The rise of Ethereum has turned this clever trick into an ecosystem full of practical applications, and that transformation is not yet over, because Moore’s law, though slowing down, refuses to die.
It was long expected that Moore’s Law – which says that the number of transistors on a chip doubles every two years with a little cost increase – would run out of steam sometime this decade. There’s only so small you can make a circuit before the strange effects of quantum mechanics start to make the results unreliable. But that hasn’t happened, yet. The smallest chips use four nanometer-wide circuits today, and the semiconductor industry now has a roadmap to chips with circuits as small as 0.7nm, taking us well into the next decade. (For reference, a silicon atom is 0.2nm wide, which might finally be close to our limit.)
In addition to making faster chips with more logic on them, we’re also getting better at doing mathematics. We’ve become vastly better at a very specific kind of complex mathematical proof that’s critical for blockchains: zero-knowledge proofs (ZKPs). Zero-knowledge proofs are mathematical tools that allow you to prove information is true without revealing the underlying data. This makes it possible to summarize lots of transactions without attaching all the necessary data or to keep the information about those transactions a secret.
ZKPs are essential both to making blockchains handle more transactions as well as enabling privacy for users. The problem with ZKPs is that they are hard to do and need lots of computing power.
In the space of just a few years, ZKPs have gone from proof-of-concept demonstrations to core technologies in the world of blockchain. Part of the credit goes to faster, better, cheaper computers, but it turns out our math skills in this space are improving enormously as well. While nobody has defined a kind of Moore’s Law for ZKPs, our own experience at EY has been very good: the performance of Nightfall, the privacy technology we developed, has improved by a factor of over 10,000 since we unveiled the prototype in 2018.
When we combine improved chip performance with better mathematics, the result should be profound changes to how blockchains operate. The earliest parts of this are already visible: zero-knowledge roll-ups and zero-Knowledge-based virtual machines use advanced math and a lot of computing power to compress and run blockchain transactions on Ethereum. Where we used to need to buy significant server time to run tests of Nightfall, we can now run the latest version on top-of-line laptops.
At the rate things are moving, just about any device, including your phone, should be able to act as a blockchain node and to process transactions on-the-device and not merely sending them into the cloud. You can already do basic ZKP transactions in your browser for chains like Z-Cash. As these capabilities spread, the result could be a much more genuinely decentralized blockchain ecosystem with significantly less centralization of compute-intensive services.
Another significant change may be to increase the allowed size of the largest smart contracts. Today they are limited to 24kb on Ethereum and many of the largest DeFi services need to string together multiple contracts. Allowing bigger smart contracts could simplify services, reduce costs and reduce the opportunities for hackers as well.
For years, we have talked about re-decentralizing the internet. Blockchains have shown us a path forward, but the reality hasn’t always lived up to the hype. Many parts of the Web3 world remain highly centralized. The next iteration of blockchain improvements may give us a new opportunity to achieve extreme levels of genuine decentralization, giving us resilient networks with innovative services. The blockchain evolution is far from over.
Note: These are the personal views of the author and do not represent the views of EY or those of CoinDesk, Inc. or its owners and affiliates.
