Cosmos
consensus-problem">Jae Kwon and the Consensus Problem
By 2014, the blockchain space had a growing list of competing networks — Bitcoin, Litecoin, early Ethereum testnets, and dozens of smaller projects — all running independently, unable to talk to each other. More fundamentally, most of them were using Nakamoto Consensus (proof of work), which achieved security through massive energy expenditure and introduced probabilistic finality: a transaction might eventually be reversed if enough competing chain work appeared.
Jae Kwon was a software engineer who had worked at Yelp and various startups when he became interested in Byzantine Fault Tolerant (BFT) consensus algorithms — a family of consensus mechanisms from the academic distributed systems literature that predated Bitcoin. BFT systems could achieve deterministic finality, meaning once a block was committed, it was final. No reorganizations. No probabilistic waiting. This property was deeply appealing for financial applications.
In 2014, Kwon published "Tendermint: Consensus without Mining" — the paper that would form the technical foundation of the entire Cosmos ecosystem. The paper proposed a BFT consensus algorithm adapted for public blockchain settings, combining academic rigor with practical engineering considerations.
Tendermint: A New Consensus Engine
How Tendermint Works
Tendermint separates consensus into two distinct layers: the consensus engine (Tendermint Core) and the application logic. The consensus engine handles how nodes agree on the ordering of transactions. The application layer handles what those transactions actually do — updating account balances, executing smart contract logic, or anything else the developer wants.
In Tendermint's consensus protocol, a set of known validators take turns proposing blocks. After a block is proposed, validators go through multiple rounds of voting. A block is committed when more than two-thirds of validators (by voting power, weighted by stake) sign off on it. If a round fails — because validators disagree or the proposer is faulty — the system moves to the next round with a new proposer.
The two-thirds threshold comes directly from the mathematics of Byzantine fault tolerance. As long as fewer than one-third of validators are Byzantine (malicious or faulty), the protocol guarantees safety: all honest nodes agree on the same committed state. This is a stronger guarantee than Bitcoin's probabilistic security model, though it comes with different tradeoffs — Tendermint requires a known validator set, meaning it is not as open as Bitcoin's permissionless mining.
ABCI: The Application Layer
One of Tendermint Core's most elegant engineering decisions was the Application Blockchain Interface (ABCI). Rather than baking application logic into the consensus engine, ABCI defined a clean interface between the consensus layer and whatever application was running on top.
This meant that any application — written in any programming language — could use Tendermint as its consensus engine by implementing the ABCI interface. The consensus engine didn't need to understand what the transactions meant; it just needed to order them and pass them to the application for processing. This separation of concerns made it possible to build entirely custom blockchains (called "sovereign chains" in the Cosmos vocabulary) using Tendermint without being constrained by any single programming model or smart contract language.
Ethan Buchman and the Cosmos Vision
From Tendermint to an Internet of Blockchains
Jae Kwon's collaborator Ethan Buchman brought a broader architectural vision to the project. Buchman, who had been working on decentralized systems and had a deep background in distributed computing theory, helped articulate what would become the Cosmos vision: not just a better blockchain, but an "Internet of Blockchains."
The diagnosis that Buchman and Kwon shared was that the blockchain ecosystem was fragmented by design. Each network was an island. The solution, they argued, was not to pick one winner blockchain and make everything run on it — a path that created dangerous centralization and scalability bottlenecks — but to create a framework for building interconnected, sovereign blockchains that could communicate and transfer value between each other while each maintaining its own governance, validator set, and application logic.
This vision was articulated in the Cosmos whitepaper, published in 2016 by Kwon and Buchman. The paper introduced the term "hub-and-spoke" to describe the architecture: a central hub blockchain (the Cosmos Hub, running a token called ATOM) would connect to many spoke blockchains (called "zones"), with the hub facilitating communication and value transfer between zones.
The Cosmos SDK
The Cosmos SDK is the practical realization of the Internet of Blockchains vision. It is a framework — a set of composable modules — that developers can use to build their own Tendermint-based blockchains without starting from scratch.
The SDK provides standard modules for common blockchain functionality: a bank module for token transfers, a staking module for validator management, a governance module for on-chain proposals and voting, a slashing module for penalizing misbehavior, and more. Developers can use these modules as-is, modify them, or build entirely custom modules that integrate cleanly with the rest.
The design philosophy was similar to web frameworks like Rails or Django: give developers sensible defaults and pre-built components for common needs, while preserving the ability to customize deeply when required. The result was that building a custom blockchain with Cosmos SDK became significantly faster and more accessible than building one from scratch.
IBC: Connecting the Chains
The Inter-Blockchain Communication Protocol
The most technically ambitious piece of the Cosmos ecosystem is the Inter-Blockchain Communication protocol (IBC). If the Cosmos SDK makes it easy to build blockchains, IBC makes it possible for those blockchains to talk to each other in a trustless, permissionless way.
IBC works by having each participating blockchain maintain a "light client" of the other chain — a cryptographic proof that allows one chain to verify the state of another without running a full node. When chain A wants to send tokens to chain B, it commits the transfer to its own state, generates a proof, and passes that proof to a "relayer" — an off-chain process that forwards messages between chains. Chain B verifies the proof using its light client of chain A, and if valid, mints the corresponding tokens on its side.
The elegance of IBC is that it does not require any central trusted party to facilitate cross-chain communication. The security of the transfer is rooted in the consensus security of the participating chains themselves. Any two IBC-compatible chains can communicate without asking anyone's permission.
The Interchain Foundation
The Interchain Foundation (ICF) is a Swiss non-profit that was established to fund the development of the Cosmos ecosystem. Kwon and Buchman ran a token sale in April 2017, raising $17 million in just 29 minutes — one of the fastest fundraises in crypto history at that point. The funds went primarily to the ICF, which disbursed grants to development teams building Cosmos tools and protocols.
The ICF model — a non-profit foundation stewarding protocol development through grants rather than direct employment — became influential in the broader crypto space. It created a degree of separation between the financial interests of token holders and the governance of the protocol's development.
A Thriving Ecosystem
The Cosmos SDK's success is visible in the range of significant projects that adopted it. The Binance Smart Chain (now BNB Chain) was built on a modified version of Tendermint. Terra (before its collapse) used the Cosmos SDK. Crypto.com's Cronos chain, the Osmosis decentralized exchange, Celestia's modular data availability layer, and many others are all Cosmos SDK chains connected through IBC.
By enabling each application to be its own sovereign chain with full control over its validator set, upgrade process, and token economics — while still being interoperable through IBC — Cosmos made a genuine architectural contribution to how multi-chain systems are designed. Jae Kwon's original insight that Byzantine fault tolerant consensus deserved a place in public blockchains turned out to be correct, and the infrastructure built around it became some of the most widely deployed blockchain software in existence.