UBC's Zite (after sufficient training) was consistently excellent at surfacing high-signal content from a variety of sources. It declined after acquisition by CNN and has become nearly invisible after incorporation into Flipboard's "Cover Stories". Now that Flipboard has turned off the backend which powered the Zite app, the next best options are:

  - manually curate Twitter lists for reading in Flipboard
  - manually curate RSS feeds 

In the age of burgeoning AI and powerful clients, it is tragic that we have such little investment in automated filtering where all data is private to the endpoint and portable between devices. Or public sharing of training data for news algorithms. Where is the GPL-equivalent for data at risk of being orphaned?

We need standardized legal clauses for startup founding engineers to deter "Orphan Software Works". E.g. source code is escrowed at the time of each acquisition, and if the acquired technology disappears from the market because all acquirers have closed or abandoned the product, it must be released to the public as open-source, with or without patent grants or training data.

Code and learning from dead startups can fertilize research fields for future crops, instead of disappearing into legal black holes.