Federated Services

Author: Yasmin Curzi

Federation refers to a group of individuals, servers or applications that operate in a decentralized manner, following a common protocol. It typically involves the exchange of data and services between two or more cloud service providers, sometimes for security reasons (logging into a website through another service’s more secure protocols), but also for collaborative purposes – e-mails, a prime example of federated services, can be sent across service providers –, among others. One peculiarity is that the decentralized structure may enable the sharing of knowledge derived from data without actually transferring such data from one server (or from an individual’s device) to another. Specifically, federated learning (also known as collaborative learning) is a machine learning technique that trains an algorithm across multiple decentralized edge devices or servers holding local data samples, without exchanging them.

A more extensive use of federated services could encourage user independence and reduce market concentration since users would tend to be less dependent on market-dominant services (e.g., the case of WhatsApp in Brazil, where using the service is almost necessary for ordinary digital social interactions). In a more extensive implementation, federated services could further improve users’ security and privacy through federated computing, with even further data decentralization. I.e., instead of data being stored in massive central servers, data could remain in the users’ devices and be accessed and processed locally – the result of the computation could then be privately aggregated before being sent to the server, ensuring the proper anonymization is taking place. Overall, it could result in a reduction in the amount of information that is stored by companies, whose often liberal use of user data is hard to assess and (potentially) punish. 

Another off-shoot of federated services are the so-called federated networks, e.g. ‘Mastodon’, which is a decentralized, open-source software for networking services. The infrastructure of these types of networks is composed of several other networks which have their own structure, with their own content moderation features, terms of service, and community policies. They are often part of a larger ‘Fediverse’, which allows interactions between users and members in different communities. Nevertheless, Mastodon has general rules that apply to all communities, not unlike a constitution that applies to all states of a federated country. It is a good example to be followed vis-à-vis interoperability features since it allows users to communicate with other communities without having to integrate them.