Why Blockchain TPS numbers fail so often in the real world
Blockchain transactions per second (TPS) numbers are often taken as a performance metric, but they don't tell the whole story about whether a network can scale in practice.
Psy Protocol founder and former hacker Carter Feldman told Cointelegraph that TPS figures are often misleading because they ignore how transactions are actually verified and transferred in decentralized systems.
“A lot of pre-mainnet, testnet, or independent benchmarking tests only measure TPS on a single node. At that point, you can call Instagram a blockchain that can hit 1 billion TPS because it has one central authority that validates every API call,” Feldman said.
Part of the problem is how most blockchains are designed. As you try to go faster, the load on each node becomes heavier and decentralization becomes stronger. That burden can be reduced by separating transaction execution from verification.
TPS numbers ignore the cost of decentralization
TPS is an accurate measure of blockchain performance. If a network has a higher TPS, it can handle more actual usage.
But Feldman argues that most headline TPS figures represent idealized settings that don't translate to real-world input. The impressive numbers do not show how the system works in decentralized conditions.
“The TPS of a virtual machine or a single node is not a true measure of the blockchain's mainnet performance,” Feldman said.
“However, the number of transactions per second that blockchain can handle in a production environment is still an accurate way to measure its use, which should mean distortion.”
Any full node in the blockchain must ensure that transactions follow the rules of the protocol. If one node receives an invalid transaction, the others must reject it. That's what makes a decentralized ledger work.
Related: Firedancer accelerates Solana, but cannot reach full potential.
Blockchain performance takes into account how fast a virtual machine can process transactions. But bandwidth, latency and network topology matter in the real world. Therefore, performance also depends on how transactions are received and verified by other nodes on the network.
As a result, TPS figures published in white papers often differ from actual network performance. The benchmarks that separate execution from relay and verification costs measure something closer to virtual machine speed than blockchain metrics.
EOS, the network where Feldman is a former block producer, broke initial coin offering records in 2018. The white paper suggested a theoretical level of around 1 million TPS. That remains an eye-popping figure even by 2026 standards.
EOS never reached its theoretical TPS target. Earlier reports said it could hit 4,000 transactions with favorable settings. However, according to research conducted by blockchain testers at Whiteblock, under real network conditions, the result drops to 50 TPS.
In the year In the 2023 Leap Crypto, Solana's validator client FireDancer tested 1 million TPS, demonstrating that EOS had achieved what it could not. The client has since been rolled out, with many authenticators running a hybrid version known as Frankendasser. The Solana typically operates around 3,000-4,000 TPS today under live conditions. 40% of those transactions are voiceless transactions, which better reflect user activity.
Solving the linear scaling problem
Blockchain throughput is usually linearly related to workload. More transactions reflect more activity, but also means nodes receiving and validating more data.
Each additional transaction increases the computational load. At one point, bandwidth limitations, hardware limitations, and synchronization delays make incremental increases sustainable without sacrificing decentralization.
Feldman says that overcoming this limitation requires rethinking how accuracy is verified, which can be done with zero knowledge (ZK) technology. ZK is a way to ensure that a set of transactions are executed correctly without requiring each node to re-run those transactions. ZK is often pushed as a solution to privacy concerns, as it allows authentication without revealing all underlying information.
RELATED: Privacy tools are rising behind institutional adoption, says ZKsync dev
Feldman argues that through repeated ZK-proofs he can also ease the burden of stretching. Simply put, it refers to evidence that corroborates other assertions.
“You can take two ZK-proofs and generate a ZK-proof that proves both of these proofs are correct,” Feldman said. “So you can take two pieces of evidence and make one piece of evidence.”
“Let's say we start a transaction with 16 users. We can take the 16 and make them into eight proofs, then we can take the eight proofs and make them into four proofs,” Feldman said, illustrating a proof tree where multiple proofs eventually merge into one.
In traditional blockchain designs, adding TPS increases the authentication and bandwidth requirements for each node. Feldman argues that through a proof-of-concept design, utility can be increased without disproportionately increasing the verification costs of each node.
This does not mean that ZK will completely eliminate sales results. Generating evidence is computationally intensive and may require specialized infrastructure. While authentication is cheap for ordinary nodes, the burden shifts to authenticators doing the heavy cryptographic work. Adapting evidence-based authentication to existing blockchain architectures is also complex, which helps explain why most major networks still rely on traditional execution models.
Performance over raw emissions
TPS is not useful, but it is conditional. According to Feldman, cash figures are less meaningful than economic indicators like transaction fees, which are more indicative of a network's health and demand.
“I would argue that TPS is the number two measure of blockchain performance, but not just in the production environment or where transactions are processed, but only if they are transmitted and verified by other nodes,” he said.
Blockchain's dominant and existing design has also influenced investment. You can't simply tie in evidence-based validation without redesigning how transactions are designed around sequence execution.
“Initially, it was almost impossible to collect money for anything except ZK EVM [Ethereum Virtual Machine]” said Feldman, explaining the Psy Protocol's previous funding issues.
“The reason people initially didn't want to fund it was because it took a while,” he added.
In most blockchains, higher TPS means more work for each node. A news report alone does not indicate whether that workload is sustainable.
Magazine: Ethereum's Roadmap to 10,000 TPS Using ZK Technology: A Dummies Guide
Cointelegraph Features and Cointelegraph Magazine publish long-form journalism, analysis, and narrative prepared by Cointelegraph's in-house editorial team and external contributors handpicked with subject matter expertise. All articles are edited and reviewed by Cointelegraph editors in accordance with our editorial standards. Contributions of external writers are commissioned for their experience, research or opinion and do not reflect the views of Cointelegraph as a company unless expressly stated. Content published in features and magazines does not constitute financial, legal or investment advice. Readers should conduct their own research and consult qualified professionals when appropriate. Cointelegraph maintains full editorial independence. The selection, mission and publication of features and magazine content are not influenced by advertisers, partners or commercial relationships.
