Although Bitcoin launched in early 2009, it took until 2017 for blockchains to go mainstream. And only in November 2021 – almost 12 years after Bitcoin’s launch – did crypto’s market cap peak at $2.9T.
Bitcoin’s growth created immeasurable wealth and changed how society perceives money – and who controls its issuance. But along the way, blockchains became victims of their own success. They couldn’t handle all the traffic, leading to long transaction times and high fees.
To understand why that is, we must understand why blockchain networks are also called layer 1 networks, and what makes blockchains different from regular computer networks.
Blockchains vs. Computer Networks
On a basic level, all blockchains are computer networks. Computer networks comprise groups of network participants, known as nodes, that relay data and share computing resources. These nodes can connect to each other in very different ways. There are four main types of computer networks:
Mesh – A node connects to every other node
Ring – A node connects to two other nodes, creating a bi-directional ring
Bus – A node connects only to one other node
Star – A server node connects to client nodes
The star is the most common computer network because it’s fast and cheap. On star networks, the central server node feeds data directly to other nodes, so the data doesn’t have to go through each node on the way to other ones.
This saves network bandwidth and, because the server node provides computing resources directly to client nodes, is highly efficient. However, the price for this performance is high centralization, both in terms of control and single points of failure (SPoF).
In contrast, peer-to-peer (P2P) networks don’t use server nodes to coordinate the network. Instead, each node acts as both a client and a server, sharing computing resources across the network. Such networks solve the problem of centralized control and SPoF, making it ideal for P2P money like Bitcoin.
The cost of decentralization is that peer-to-peer networks tend to be less scalable. This problem applies to blockchain networks because they are secured by P2P network consensus mechanisms. Vitalik Buterin, the co-founder of Ethereum, called this balancing act the Scalability Trilemma (also known as the Blockchain Trilemma).
Early blockchains were limited to offering only two features at once, meaning they’d have to sacrifice on either scalability, security, or decentralization.
What Are Layer-1 Blockchains?
To address the scalability trilemma, blockchain networks began to adopt different approaches. These approaches are called Layer 1s – the base layer of a blockchain network. Bitcoin, Ethereum and Solana are all examples of Layer 1 blockchains.
One obvious way early Layer 1s tried to address the Scalability Trilemma was by increasing the block size. This way, the blockchain could process more transactions within each data block, increasing the number of transactions it could process per second.
Yet increasing the block size would also require node operators to maintain more powerful computers. Fewer people could afford them, leading to greater centralization.
When billionaire Elon Musk proposed increasing Dogecoin’s block size by 900%, Ethereum co-founder Vitalik Buterin pointed out that the blockchain wouldn’t be decentralized if regular users with consumer-grade PCs couldn’t run a node.
Modern layer 1s address the Scalability Trilemma through consensus mechanisms and sharding.
Consensus algorithms underpin blockchain technology. For Bitcoin and other cryptocurrencies to have value, the P2P network has to solve two key problems: double-spending and incentivization.
Double spending is when someone uses the same scarce resource twice (like money). It’s a problem inherent to digital technology because digital files are infinitely reproducible. To solve this, blockchains make each transaction unique through time-stamps and hashes, and by adding them to batches of transactions called blocks. To fake a transaction, a node would have to falsify an entire block.
This is where consensus algorithms come in. They coordinate all nodes of the network in a decentralized manner. For a block to go through, the network has to agree on the validity of the data contained within them. Crucially, if some network nodes submit spurious data, the network can still function as long as the majority of valid nodes control the network’s processing power (hashrate).
“As long as a majority of CPU power is controlled by nodes that are not cooperating to attack the network, they’ll generate the longest chain and outpace attackers.”
– Satoshi Nakamoto, the inventor of Bitcoin
Such network redundancy is called Byzantine Fault Tolerance (BFT). In a decentralized network, it is extremely important for the network to be operational even if some of its nodes do not work properly. Otherwise, it would ground to a halt.
In addition to addressing the double-spending problem, consensus protocols provide incentives for nodes to keep processing transactions. This is equally important: why would anyone sacrifice their computing power and pay an outsized electricity bill for free?
In the case of Bitcoin, node operators called miners expend computational resources. For their trouble, they receive block rewards as BTC. This is known as proof-of-work (PoW).
Other blockchains, such as proof-of-stake (PoS), use validators as node operators. Instead of expending energy-hungry computing power, validators rely stake (lock up) resources – coins – to achieve the same consensus coordination goal. For instance, Ethereum requires a 32 ETH stake to become a validator. After validators have staked funds, they receive a cut from each transaction fee.
Malicious actors have different obstacles to overcome. With Bitcoin, they would have to have greater CPU power than 51% of the network, which is virtually impossible to achieve given its size.
Bitcoin network’s annualized energy consumption equals Thailand’s, at 204.5 TWh. Attackers would have to accrue over half of that power to commit to a coordinated hack. Image credit: digiconomist.net
With Ethereum, they’d need to have the largest ETH stake – immensely wealthy, in other words. However, the attacker would have to be prepared to lose that wealth; the entire network would lose its value as soon as they processed a fraudulent transaction.
While most new L1s use PoS, they aren’t always better at scalability. Solana, a PoS blockchain, suffered multiple outages after its traffic load increased during the last 12 months. Its staking protocol was of little use when nearly half of its nodes were hosted on just five data centers.
Solana offers theoretical network throughput of 50,000 transactions per second (TPS). That’s a lot more than Bitcoin’s ~5 TPS – but what good is it if it isn’t decentralized?
Another layer 1 scalability solution is sharding, which partitions a network into small databases called shards. Each shard runs its own transactions and adds blocks with its own nodes. By distributing processing across lots of little shards, the load is taken off of the main consensus mechanism, resulting in a higher TPS.
However, since each shard is smaller, it’s easier for an attacker to amass the funds or computational power necessary to overwhelm it. For this reason, sharding has yet to be proven on a large blockchain.
Ethereum is leading the way and plans to implement sharding after it transitions from PoW to PoS consensus in 2022. Ethereum sharding, scheduled for 2023, will compartmentalize Ethereum into 64 shards.
The network will attempt to address sharding’s security concerns by randomly assigning nodes to shards, including randomly reassigning nodes to other shards.
Other sharding experiments aim to resolve the Scalability Trilemma completely. Swiss-based Distributed Technology Research Foundation (DTR), consisting of seven universities, launched the Unit-e project in 2019 as a scalable global payment network. Another project, Radix, partially orders shards rather than framing them on a single timeline, like Ethereum does.
Get Smarter on DeFi and Web3
Get the 5-minute newsletter keeping 80K+ crypto innovators in the loop.
Are Layer-1 Scalability Solutions On The Way?
Tampering with a blockchain network is a delicate matter. Most people are already suspect of crypto. Bitcoin managed to assuage those concerns over a decade, so its layer-1 upgrades are more conservative.
The latest Bitcoin upgrade, Taproot, introduced Schnorr digital signatures. They let the network batch multiple transactions together to cut fees and increase scalability. However, Bitcoin still prioritizes layer 2 solutions for true scalability through the Lightning Network.
The same is true of Ethereum, with dozens of layer 2 networks built on top of layer 1.
Top 10 L2 solutions for Ethereum. Image credit: L2beat.com
In both cases, L2 protocols take the workload off of the main L1 chain, process it elsewhere, and feed data back to the L1 in a far more efficient manner. L2s employ a variety of scalability technologies to accomplish this, as noted in the table above.
However, an ecosystem of L1 and L2 networks is complicated. Tokens have to be transported across blockchain bridges, and each dApp has to be integrated into each L2. By contrast, engaging solely with L1 networks would make life easier for developers and users.
L1s such as Cardano, Algorand, Elrond, Fantom, Avalanche and Harmony have all tried to fix the Scalability Trilemma, but none have the footprint that approaches either Bitcoin or Ethereum. Still in their infancy, it’s too early to conclude whether even blockchains with operational mainnets have vastly improved upon BTC or ETH.