The Technical Evolution of Vannevar Bush’s Memex (2008)

coincidentally i am watching these treasure trove of videos:

https://archive.org/search.php?query=subject%3A%22Vannevar+B...

after reading the above, highly recommended to watch those!

After reading “As We May Think”, the thing that struck me wasn’t so much the predictions that Bush got right (which were many), but one particular technology that he missed: digital image processing and printing.

Everything he describes is based on photographic reproduction of handwritten or traditionally printed material, with an extremely small number of annotation bits encoded by exposing film to individual lamps. He more or less describes barcodes, but misses that they can be printed on the product packaging without special equipment. The idea that a computer could draw things was completely unimaginable.

I actually think that this oversight helps sell how visionary his ideas were. This really isn’t a paper that could be written today, because there’s this really fundamental thing that everyone takes for granted that’s conspicuously absent— It helps anchor the text in the time it was written.

I just posted this in another thread, but there are some very similar ideas I am implementing here: http://onemodel.org

(Its strength is not in document storage, though it can. Also not really about version control but that could come later: it is about brainstorming fast, rearranging things fast, exporting as outline documents, and not being as limited as if the info were all textual.)

A hierarchical organizer: AGPL, desktop- and keyboard-oriented, highly efficient, I use every day because org-mode was more awkward to me: http://onemodel.org . (My progress has been slow lately but I have plans: feedback welcome, especially if on the mailing list.)

Anything can be nested anywhere, so I put effectively all my notes in it on everything. In college it would have been a huge help. It exports to text outlines (w/ or w/o legal numbering) which I've found useful when sharing info before the real sharing feature is implemented.

(The idea behind it is to reduce knowledge to an atomic level, as an object model: things we know have relationships to other concepts, can be expressed as measurements, etc. But it is cumbersome to create an object model anew, for every app. Right now, it is really good at supporting a big, efficient list of lists (based in postgresql), with attributes, and very minimal support for defining classes on the fly as a side-effect of use. But I really hope to be able to add anki-like (spaced repetition / flashcard-like) features, internal scripting, hosting, and secured sharing of info between instances, so it becomes like a wiki in convenience, but more efficient and computable rather than piles of words for everything).