Here’s my original proposal:

Been thinking a lot about what it might mean to make Linked Data reliable and resilient. We can do better than just “the LOD cloud” – we can make a web of data that can survive the temporary or permanent loss of a node in the big graph or a set of data sources. Since Linked Data is a natural extension of the web, we have all the knowledge and experience of 20+ years of web and networking developments to apply to building Linked Data systems. We’ve learned a few things about proxying and caching, in particular, and those concepts should apply equally well to linked data. If you’re interested in the “web of data”, whether as a consumer of it in the course of your research or as a producer of digital humanities resources or both, I’d like to highlight some of these issues for you by demoing some work we’re doing in the realm of digital collections in libraries, and to leave you with a few ideas for making your own stuff more resilient.

But then the King of Pop died.

So instead, I would like to demonstrate the shot-for-shot recreation of the famous Thriller video I made last night with an Arduino, Omeka, Processing, crowdsourcing, rectified old maps from NYPL reprinted using e-ink, and a native RDF triple store.