Tackling climate change: electrify everything, decarbonize electricity.
Locational marginal pricing w/in 45s …
(possibility for distributed updating in ~real time)
Joel on Underlay:
The current web / world conditions us to have a belief that a URL or chart is enough background. That’s ridiculous. Working on the Underlay Project —> incubated at the Knowledge Futures Group at MIT. Building a global decentralized graph db, w/ automated provenance. Another way of pitching it: this is for knowledge what IPFS is for bytes: A universal resource where you can describe the data you want and have it resolved.
We can’t verify data by its hash, so data has to carry provenance with it. We’re just building a system where dependencies are explicit and we can compose sources losslessly. If that sounds fun, or you care about the inevitable future of knowledge, come talk to us. We’re looking for collaborators.
Comparisons (suggested by others):
Kipster, Andre Cruz: can use lazy loading, storing CRDTs on IPFS, and later on there will be pinning servers to guarantee persistence of certain ones, for persistence.
Notion, as data metastore. put WP on it; distribute. (Kip)
Library authority-alignment network: Started work w/Freebase before 2012.
New effort w/ Stanford, DPLA, others in 2017, in spare time; stopped March ‘18.
A new alternative for publishing science: Raw data + analysis, all things whose provenance traces back to you and yours. Recognition based on a combination of volume, clarity, usefulness of grain-size, interest of result, and quality of reuse. (cf Drew Wright, StanfordLibrary)
Future of IPLD
Computation in a Merkle forest: want pure/reproducible semantics; a “query planner”; caching of subcomputations
Building a programming environment: data and type definition, schemas; a lang of value literals, w/ human syntax; serialization lenses, selectors, queries w/ binding, primitive recursion, equijoins; object model for APIs, db-style indices, CRDTs.
DTube | media over IPLD