My buddy works as an editor at a publishing house, polishing manuscripts for dozens of authors every year. Once the book is published, authors earn royalties while he gets a fixed salary, so if the book sells like hotcakes, it doesn’t really affect him. He says he’s come to terms with it, but sometimes seeing a book sell hundreds of thousands of copies still stings a bit. In the AI economy, there’s a layer of people in a similar boat, namely those who are actually contributing training data. They are the bottom-tier suppliers of the entire AI industry but hardly see any long-term gains in the current setup. @OpenLedger has coined this mechanism as the contributor layer, implementing a Proof of Attribution mechanism that establishes an exclusive distribution system for this layer.
The contributor layer encompasses more people than most might think. Professional data labelers are one type, domain experts and researchers fall into another, and content creators are included too—anyone who organizes valuable expertise into datasets and uploads them to Datanet counts. This group has been invisible in the AI value chain; their contributions vanish once packed into model weights. #OpenLedger 's Proof of Attribution makes these invisible contributions visible. At its core, the Infini gram algorithm traces back the output of each inference to specific training data tokens, automatically distributing $OPEN tokens to the respective contributors based on the actual impact weight of each dataset on that output.
What I find most interesting about this design is not the tech itself but the fact that it gives the contributor layer an identity for the first time. In the past, when you contributed data, you had no clue how much it was worth, nor did you have any proof that it was used. Now there’s a record on-chain, attribution is calculated, and contributors can finally say, 'This is my stuff.'
Currently, the accuracy limits of attribution weight calculation pose a challenge. When multiple data sources jointly influence an output, achieving a distribution ratio that everyone agrees on is tough. Another issue is that most people in the contributor layer aren't Web3-savvy; the barriers of on-chain operations, wallet management, and token swaps can keep a lot of genuinely valuable domain experts out of the game. For the contributor layer to truly be activated, the usability of the product is a bigger hurdle than the tech itself.
The contributor layer encompasses more people than most might think. Professional data labelers are one type, domain experts and researchers fall into another, and content creators are included too—anyone who organizes valuable expertise into datasets and uploads them to Datanet counts. This group has been invisible in the AI value chain; their contributions vanish once packed into model weights. #OpenLedger 's Proof of Attribution makes these invisible contributions visible. At its core, the Infini gram algorithm traces back the output of each inference to specific training data tokens, automatically distributing $OPEN tokens to the respective contributors based on the actual impact weight of each dataset on that output.
What I find most interesting about this design is not the tech itself but the fact that it gives the contributor layer an identity for the first time. In the past, when you contributed data, you had no clue how much it was worth, nor did you have any proof that it was used. Now there’s a record on-chain, attribution is calculated, and contributors can finally say, 'This is my stuff.'
Currently, the accuracy limits of attribution weight calculation pose a challenge. When multiple data sources jointly influence an output, achieving a distribution ratio that everyone agrees on is tough. Another issue is that most people in the contributor layer aren't Web3-savvy; the barriers of on-chain operations, wallet management, and token swaps can keep a lot of genuinely valuable domain experts out of the game. For the contributor layer to truly be activated, the usability of the product is a bigger hurdle than the tech itself.