You can’t see a lot of (English) information on Noguchi Filing System even using google, probably because the most important link went 404. Wikipedia also doesn’t have much. See explanations on it on 43 folders.

Some help from Boing Boing:

New documents (envelopes) are added at the left end of the “envelope buffer,” and whenever a document is used (i.e., the envelope removed from the shelf), it is returned to the left end of the bookshelf. The result of this system is that the most recent (and frequently) used documents migrate to the left, while documents that are not used often or not used at all migrate to the right. After the system has been in use for a while, the shelf starts to look like the following.

And, one and two posts by Edward Vielmetti. Interesting quote from the second:

Noguchi’s ideas are largely inspired by discoveries related to the use of computers… Nevertheless, when building a database there seems to be no way to avoid using fields, which amounts to classifying. Similarly, the entire process of tagging, be it in SGML or XML formats, involves labeling items of knowledge, often for commercial purposes. The digitization of data in itself does not necessitate classifying, but the use of database applications compels it to a certain extent. Categories, even the most sophisticated ones, once used necessarily reflect the limits of our vocabulary and conceptual horizon.

The strength of it is it is so natural, you can’t digress from it. You can use vertical Noguchi System (meaning you just stack papers on top of other papers) instead of Noguchi’s horizontal one and claim that it’s the most advanced hyperorganization scheme, as Eric Abrahamson says in p. 157 of “A perfect mess.”