Why You Shouldn’t Build a Blockchain - by Daniel Olshansky

Olshansky's Newsletter

SubscribeSign in

Why You Shouldn’t Build a Blockchain<br>What I learned after spending four years building one

Daniel Olshansky<br>Aug 14, 2026

Share

Special thanks to 0xRahul for feedback and review.<br>Preamble : In early 2016, Ethereum was my gateway into the crypto industry. From late 2021 through early 2026, I worked in the industry full-time. I saw the good, the bad, and the ugly. More recently, I’ve had a handful of conversations where people asked me: “So after all these years, what do you think?” This post aims to capture my journey, the opinions I formed along the way, the tradeoffs of having a blockchain, and where I believe value will accrue in the industry moving forward.<br>tl;dr<br>Blockchains are insurance you rarely use. You pay a premium in cost and complexity, hoping you’ll never need it.

Distributed systems are not decentralized networks . These two are often conflated but solve completely different problems.

You don’t need a token . With the rise, adoption, and regulation of stablecoins, there should almost never be a reason for a project to have its own token.

Agentic commerce. Reliable, scalable, and cost-effective digital payments require new infrastructure, and a few large players are leading the charge.

Table of Contents

2014 - 2016: Early Days and Nerd Sniping

2016 - 2018: Losing and Regaining Interest

2021 - 2025: Pocket Network, Grove, APIs and QoS

Distributed Systems Are Not Decentralized Networks

Decentralization Comes at a Premium

Do you need a token though?

2026+: Agentic Payments and the Future of Crypto

Concluding Thoughts and Painful Lessons

Appendix

2014 - 2016: Early Days and Nerd Sniping

In 2014, I did my undergrad thesis at UofT in speeding up fMRI analysis. A graduate student was responsible for migrating PCA analysis over from a CPU to a GPU, and I helped by distributing it over a cluster using Hadoop MapReduce.<br>In hindsight, I should have seen the pattern and capitalized on the fact that Cloud (distributed computing) and GPUs (parallel computing) were the future, but 🤷<br>I was aware of Bitcoin at the time, but didn’t think much of it. When Ethereum came out, my interest spiked because it led with “smart distributed systems” rather than Bitcoin’s “economic anarchism”. Ethereum was the ideal nerd snipe by living at the elegant intersection of distributed systems, cryptography, game theory, decentralized networks and programmable trust, and digital money.<br>2016 - 2018: Losing and Regaining Interest

Before the 2017 ICO craze kicked off, I got interested in prediction markets (e.g. Augur, Gnosis) and Decentralized Autonomous Organizations (e.g. Aragon). DAOs were a cool idea that never materialized - there’s a reason they say that democracy is the worst form of government, except for all the others. Years later, prediction markets caught on through a new generation of companies (i.e. Polymarket and Kalshi), though I’m still hoping those companies shift focus from everyday gambling to leveraging the wisdom of the crowds.<br>In late 2018, I went to a couple of conferences and got disenchanted by suits building out IBM’s Hyperledger. Around the same time, I kept one foot in the industry by becoming an advisor for a project called Pocket Network building a decentralized RPC network. I like to describe it as a marketplace of API providers for canonical data sets, akin to what OpenRouter has done for LLM inference providers.<br>I didn’t pay much attention to the NFT boom, and looked past DeFi summer. I was never a fan of the speculative side of crypto and was more interested in the value that Decentralized Physical Infrastructure Networks (DePINs) could bring through smart request routing and incentive alignment of unused hardware.<br>2021 - 2025: Pocket Network, Grove, APIs and QoS

In late 2021, I got invited as a guest to Pocket Network’s offsite, and ended up joining as a consensus protocol engineer after running into Robert Downey Jr. in Mexico City.<br>Pocket Network’s mission was permissionless infrastructure. The core problem it was trying to solve was access to reliable blockchain data. Why? Because most users and institutions don’t want to run their own infrastructure¹.<br>There were two entities responsible for making this happen:<br>Pocket Network Foundation : A nonprofit entity that stewarded the Pocket Network blockchain, community, and token ecosystem. It acted as a liaison between the mission and the development team.

Grove : A for-profit corporation that led the technical R&D for the Pocket Network blockchain, launched the primary gateway to the network, and focused on customer acquisition and building products on top of the network.

We designed the blockchain to coordinate a permissionless network of infrastructure providers. It included primitives for privacy-preserving requests and a verifiable rate limiter. We also built a multi-provider API gateway.

Along the way, we learned that...