Blogs
Humanbased Blog
Writing on verifiable human expertise — data provenance and attribution, model training infrastructure, and contributor ownership.
Latest
From one piece of data to an asset Hand data over and you lose control of it — you cannot even tell a breach from a leak. XnY Protocol keeps the books at the level of a single contribution: fingerprints, shares and grants recorded on a ledger that belongs to no party. Why the URI is already final before the upload finishes A contributor hits submit and gets a permanent address within seconds — while the file has not reached Arweave yet. The URI can be fixed that early because identity and delivery are separated. Pricing the Human Layer: Instrument Fit, Allocation Rules, and Settlement Design for AI Training Data A full research report: what forty real-world data deals, nine pre-registered experiments, and one adversarial red-team say about when royalties are the right way to pay for data — and how to split them when they are. Published as a formal report, not a summary. What 295 Agentic PRs Taught Us About Code Review A retrospective of two months of Humanbased monorepo PRs and retained Crosscheck logs, with practical findings on PR shape, review cost, model routing, and product opportunities for AI-native engineering teams. Gasless Adapter: Enabling x402 Protocol Support for Non-Upgradable ERC20 Tokens Gives immutable ERC-20 tokens like $XNY ERC2612/3009-style gasless capabilities through a capability layer that enables legacy tokens to support new standards without requiring upgrades.
Browse by topic
AI & Data
Evolution of Large Model Data Engineering: A Paradigm Shift in Knowledge Extraction Efficiency The core driving force has shifted from computational expansion to knowledge extraction efficiency through four paradigms: large-scale crawling, fine-grained annotation, AI-synthetic data, and human-machine collaboration. The Minds Behind the Data: What We Learned from Our Top Contributors Qualitative insights into contributor motivations, behaviors, and expectations from interviews with eight high-performing airdrop contributors with diverse expertise. Open Robotic Data at Scale: Ecosystem Formation and Implications Future value lies in high-fidelity, physically complex, and training-ready datasets as basic manipulation tasks become commoditized and major ecosystems consolidate around standardized formats.
Protocol
From one piece of data to an asset Hand data over and you lose control of it — you cannot even tell a breach from a leak. XnY Protocol keeps the books at the level of a single contribution: fingerprints, shares and grants recorded on a ledger that belongs to no party. Why the URI is already final before the upload finishes A contributor hits submit and gets a permanent address within seconds — while the file has not reached Arweave yet. The URI can be fixed that early because identity and delivery are separated. Gasless Adapter: Enabling x402 Protocol Support for Non-Upgradable ERC20 Tokens Gives immutable ERC-20 tokens like $XNY ERC2612/3009-style gasless capabilities through a capability layer that enables legacy tokens to support new standards without requiring upgrades. Building Sovereign Data Lineage: A Decentralized Storage Architecture with DID and Proxy Re-Encryption Enable data circulation without relying on platform trust through a decentralized storage architecture combining DIDs, proxy re-encryption, and verifiable credentials. ERC-8004: MCP on Ethereum A resume template for AI agents that states who I am, what I can do: a detailed explanation of the ERC-8004 protocol for agent identity, reputation, and validation on Ethereum.