#walrus $WAL From Data Islands to Collaborative Networks: Looking at Walrus's Early Experiments in the DeSci Field

I used to think that decentralized science (DeSci) was a cool but somewhat distant concept. Until I recently delved into an early project involving multinational cancer research that utilized Walrus for collaboration. They faced a classic dilemma: the desensitized medical imaging data (such as MRI scans) from multiple hospitals could not be centralized to a single server for joint AI analysis due to privacy and compliance issues, creating 'data islands.'

Their solution is quite interesting, as each participant uploads encrypted raw data to Walrus, and the data itself is not shared. However, utilizing the prototype of the verifiable computing framework provided by Walrus, they can collaboratively train an AI model. In simple terms, the model 'moves' to each data storage node for local training, only aggregating the updates of encrypted model parameters, with the original data never leaving the local environment. Walrus plays two roles here: first, as a trusted and neutral encrypted data repository; second, providing an audit trail for data rights and access.

This case showed me that the value of Walrus goes far beyond 'storing files.' In areas requiring high trust and collaboration, it offers a foundational layer for data collaboration based on cryptography rather than legal texts. Researchers contributing data can receive verifiable contribution certificates, which may allow them to participate in distributions if patents or commercial results arise in the future. This could potentially change the previous situation where data contributors struggled to establish rights and benefits. Although this model is still in its early stages, it has opened a new avenue for fields like biomedicine and climate science that require large-scale data collaboration. The future of science lies in open collaboration, and true collaboration begins with mutual respect for data sovereignty. @Walrus 🦭/acc