Personal data management, it's been 80 years, and we still don't have a Memex.
Computing for LLMs and other predictable loads - FPGAs are unsuitable because they are optimized for the wrong things.
What's blocking you to start? Do you really need infinite resource to start?
Memex -- Maybe, just barely maybe... First I'd have to develop an actual working hypertext markup language (to allow marking up an existing file, separately, aka annotation) and then a parser to take HTML and translate it to said new language. Then a browser/editor for said new language to import content, save it all locally with timestamps, with tags to mark where it gets used in other documents, so I can purge everything I don't want to keep after 6 months, or whatever works out as reasonable. I'd also have to be able to compile and export a document and the all things it marks up as a package. (Likely a zip file)
Computing/LLMS - I'm going to do a TinyTapeout for a few cells of a BitGrid, but a proper full scale ASIC is outside my budget/expertise. Then there's the matter of compilers to spit out code to run on it.
None of these require actually infinite resources, but they're all outside what I can pull off solo.
jonahbenton•5mo ago
Rehydrate the Sahara
A full recapture/recycle workflow for landfill
Carbon capture (of course), but new ideas needed
Fully recognize eco diversity on the balance sheet