The idea the authors discuss is of GroupLens; An open architecture for distributed ratings. The architecture specified the format of ratings, the propagations of the ratings and the interfaces for delivering predictions and ratings between news clients and rating servers.
GroupLens is presented as a system for collaborative filtering of news online, to help people find articles they like in a huge stream of available articles. The authors discuss the goals for the design of GroupLens: Openness, Ease of use, Compatibility, Scalability, and Privacy. The authors then go on, discussing the implications of having a GroupLens architecture on the dissemination of news read online.
What I find most interesting in the paper is the foresight shown are some of the social implications discussed. I particularly found the following quote interesting:
Collaborative filtering may introduce many social changes
in the already rapidly evolving Netnews community. For
example, the utility of moderated newsgroups may decline.
New social patterns will have to develop to encourage
socially beneficial behaviors, such as reviewing articles that
have already received a few low ratings.
The authors talk about how certain social patterns emerge if an algorithm does not account for feedback loops to develop. For example, readers with a particular point of view may tend to receive articles of that view alone and fail to see the other side of an issue. So-called “Global villages” will fracture into tribes. The authors suggest moderation can go hand in hand with architectures like GroupLense. This is in fact how many of today’s social media and news aggregators plan to remove bias from the news in this era of fake news. A good example is Flipboard that has curators that work along with an algorithm to maintain content quality and keep bias out.
The idea of killfiles has also been translated to many of today’s applications. You can always filter our topics that you don’t like and select ” see less like these” options on articles on Google News among other places.
The goals of Openness, Ease of Use, Compatibility, Scalability, and Privacy are still valid in today’s scenario, although it can be argued that some of these are more important. Content hosting websites, such as YouTube and Netflix cannot necessarily guarantee Openness and Privacy but do great on Compatibility and Scalability, but they do great on Ease of Use and Scalability.
It can be argued that a rating system is too low a bar to set on who gets to see what information. Automation of the dissemination of news is only as good as the algorithms, or at least the data that machines learn from. Moreover, rating systems tend to favor a surface level review of a service or product. This can lead to some products, services or articles receiving an overabundant or an underwhelming level of attention. Certainly, a cautious and holistic approach must be used when implementing an architecture that enables the spread of information.