Once upon a time, in the great dark ages long ago, let's say 1945 - fifty-two
years ago, computers had just been invented well they'd been invented by lots of
people, Babbage, Turing, Zuse, Atanasoff, Eckert and Mauchly, all these people
had come up with essentially the same idea. Zuse is of special interest in that
he tried to sell digital computers in Nazi Germany. Fortunately, the Nazis
didn't know what they could have had.
Anyhow, at the end of the war, there they were with computers and the excitement
of a new burgeoning realm of possibility. And they started accumulating data.
Data for different projects. And this data was kept in files - Now, what is a
file? It's a lump. It's a lump... From the outside, a uniform, indistinguishable
lump. The interior is not addressable.
OK, first they had files, then they had filenames. Gee it makes sense to give
this file a name. Then they said 'Where shall we put all these files? Well, why
don't we make a tape? and put the files one by one on the tape, and then we'll
write down the names on a piece of paper.'
And then somebody had a bright idea and said "Wait, why don't we put the
filenames in a separate file?' Thus was born the hierarchical directory, because
it was recursive and it scaled. And since then, the tradition of lump files with
names and hierarchical directories has been the driving structure of the entire
computer field. So that in documents, one document is one file. Mmm. Metadata -
now what's that? Well, metadata began with filenames. What's metadata? It's what
isn't in the lump. So now there are all kinds of committees trying to decide how
many angels can dance on the head of a file, what is to be metadata of this
kind, that kind, the other kind, all of it because of the lump file structure
and tradition.
All of this was frozen by Unix when, in 1970, the inode table the saying is that
in Unix everything is a file actually in Unix everything is an inode, meaning an
entry in this table, which has an item with a name and then a pathname. And the
entire structure is based on rapid rearrangement of these elements as their
pathnames change. And so preoccupation with file structures where they go in
hierarchies and mapping everything to this model of lumps and names and metadata
and files has been the entire driving structure of the entire computer field.
The industry, the 'computer science' as it's called, whatever. So... finally,
the tradition of one document per file.