By the way, how does the machine implement Huffman encoding

The machine obviously would do the best job of any database form in existence, well, aw shucks, I do have a microcode engine though.

Huffman encoded attribute patterns. Quick, before Apple patents it. Why do I imagine this stuff?

Actually it is the more cotinuous version of 80/20. The 80 most comman patterns are identified by a what? a byte of attribute code!. These are right there, in the machine, the basic Dublin core. Where is the 80% of the remaining 20%? One hop away, a table look up. The next? two hop look up, and so on. The machine would define an attribute called Huffamn encoding.

MyGraph:Huff$12

The number, or constant string yield something lika a : a,b.d.e,e,f

One of the 128 most common patterns of nests found. Maximum entropy indexing is sort of a phantom when the smallest usable unit is the byte and table hop, you get most of it with common sense. But just two bytes in the link field and I bet 95% of attribute/entity matches are made in a hop, using Huffman encoded patterns.

But, any entity, pick a container vessel, a sub graph scheme. Use that mostly, just make sure you have defaults for unused attribute spots. Then when the your entity gets too important, pick a different, more complex container. Like a growing business, pick a container a little larger than you need, have spare capacity. Using this methods, easily, we could cover 16,000 combinations of container schema, plenty byte space. Put them in tables, an idea I have seen for storing attribute data in short small tables. But simply use TE script to describe the schema:

TE_Schema:(int$_.date$_,_,_._), TE_More_of_the_same:(_),,

That script, to remind us all, is exactly storable in nested form, minus the parentheses.
Anyway, define it to look really weird, impress your friends even. But if you put it in a table then a syntax even can go grab the pattern, fill it in with discovered data.

What is happening? Networks do this right now, with bunches of pages, optimal caching. Humans do it, its called thinking and acting. We already speak the same language.