Here we outline what we mean by the mirror layer, and its significance.
This layer embraces two of the hottest topics in the digital domain: data and AI.
# Data Data,, data ownership and data sovereignty are hotly contested when huge amounts of data - for example on traffic, but also on identity - are routinely scraped across the internet (across national borders) and analysed by powerful computing setups under corporate control (for example, by the search-engine corporations).
Data is controversial because you can never easily be sure who is able to access data that you think is 'yours', and private, and because you absolutely can't be sure what data about you - your identity, and the patterns of your web-mediated social interactions - is held by whom, and exploited for what purposes.
Large parts of the 'free' social media universe, for example, are expliclty designed as honey-traps where corporations can harvest data about you and your social relationships. It's a secret rent (if we think of land) or a toll (if we think of highways). Eye watering amounts of money are made in the data scraping industry, and the profiles that the sector produces are very significant in enabling (mostly corporate - either private or State) entities to taregt . . whatever: false news, prosecutions, delivery of benefits or rights, political messages; or good ol' adverising, but now narrow-cast (rather like 'smart bombs') into finely defined percentiles and demographics.
# So-called articial intelligence So-called artificial intelligence mobilises enormous amounts of scraped data - laregly from websites, where presentations of arguments and accounts of phenomena are placed - and powerful, complex families of pattern-recognition algorithms are trained to be able to respond to user queries with a simulated natural-language response which can look very much like a human-drafted account from a document on the web. These families of algorithms combined with eye-watering amounts of scraped data are called 'large language models' (LLMs).
# Mirrors Large-scale operations like these enable some actors - who are powerful and can pay - to mobilise profiles of social, economic, cultural, political, sexual, etc life. These actors 'know' things about communities and collectives that the communities themselves don't know. They know what is trending. They know where the most activity is happening, and the least. They know who - by location, by category - traffics with whom, and how this shifts.
What the machinery almost never does, is to provide members of a community with that same kind of information about the profiles of **their own activity** in the collective. What we don't have, by and large, is **mirrors** of our digital lives in the collective.
Wow! There's a thought!
This is a hugely underdeveloped field. Rather than 'small scale AIs', of which there are some - specialised by fields of professional practice for example, or sometimes by geography, trained on specialised collections of data - what if we have a discipline of 'making mirrors for communities'?
Now, there would be something for the Federation to facilitate!
> The mirror layer is **a commons of data, and of pattern recognition machinery** and of trained responses to 'conversational' queries, which mirrors the activity of a community to the community.