The the single best way to drive science and technology: teach everyone to format a json string share it via a URL.
The people on the bleeding edge of science and technology aren't just sitting in futuristic labs, surrounded by holograms and AI tech assistants, spouting one brilliant insight after another.
They are spending most of their days arranging various types of poorly formatted data, be it scientific metadata, purchase orders, journal proofs, grant proposals, interview schedules, their shopping list, or maybe (for a brief moment) actual scientific data. The real productive geniuses are the ones who figure out a way to ignore all this noise for along enough to get something done. It all comes down to logistics in the end.
A huge fraction of this work could conceivably be automated, or just trivial if people migrated to a better data structure. Maybe LLMs will help us bring order here, but that's only half the battle: part of learning to format data is thinking about the formal structure, what consumers will actually need, what parts are duplicated, etc.
20 years ago Tim Burners-Lee was advocating to clean this up via a better structured web [1], but, for various reasons, we haven't made much progress. I suspect a big reason is that the vast majority of people would struggle to understand the purpose of structured data: like basic literacy it's only useful if other people in your community use it.
The people on the bleeding edge of science and technology aren't just sitting in futuristic labs, surrounded by holograms and AI tech assistants, spouting one brilliant insight after another.
They are spending most of their days arranging various types of poorly formatted data, be it scientific metadata, purchase orders, journal proofs, grant proposals, interview schedules, their shopping list, or maybe (for a brief moment) actual scientific data. The real productive geniuses are the ones who figure out a way to ignore all this noise for along enough to get something done. It all comes down to logistics in the end.
A huge fraction of this work could conceivably be automated, or just trivial if people migrated to a better data structure. Maybe LLMs will help us bring order here, but that's only half the battle: part of learning to format data is thinking about the formal structure, what consumers will actually need, what parts are duplicated, etc.
20 years ago Tim Burners-Lee was advocating to clean this up via a better structured web [1], but, for various reasons, we haven't made much progress. I suspect a big reason is that the vast majority of people would struggle to understand the purpose of structured data: like basic literacy it's only useful if other people in your community use it.
[1]: https://en.wikipedia.org/wiki/Semantic_Web