Skip to content

Associative trail

Definition

An associative trail is a named, persistent path through a collection built by joining items in pairs: the user brings two items into view together, names the trail, and ties them with a mutual code written into both, so that whenever either is in view the other can be summoned instantly; repeated joins accumulate into a sequence that can be run through like a book, branched with side trails, and reproduced whole for insertion into someone else's store.

Explanation

The primitive is a symmetric join, and where it is stored is the design decision that matters: the code goes into blank code spaces on each of the two items, each recording the other's index number, so the link belongs to the items rather than to a central catalog and is traversable from either end. Trails are built while working rather than planned in advance — someone comparing the short Turkish bow to the English longbow ties an encyclopedia article to a history, inserts a comment of their own into the chain, and branches onto a side trail through textbooks on elasticity when materials turn out to matter — so the path is a by-product of the inquiry, not a separate cataloguing task. Three properties make a trail more than a linked list. An item can belong to many trails at once, so paths compose rather than partition the collection the way a filing scheme does. Trails persist and do not fade, so a path built years earlier is re-enterable from its head and traversable at speed, stopping at interesting items. And a trail can be photographed out and passed to another person to link into their own store, which turns a private convenience into a shareable artifact and yields the consequences Bush drew from it: a profession of trail blazers who establish useful paths through the common record, and an inheritance from a teacher that includes the whole scaffolding of reasoning behind their conclusions rather than the conclusions alone. This is a 1945 proposal whose mechanism was microfilm and photocells; the design is the direct ancestor of hypertext and of every linked note system since.

Key Properties

  • The primitive is a symmetric join — a code written into both items, each pointing at the other, owned by the items and not by a catalog
  • Trails are named, entered from the head, and traversed forwards or backwards at variable speed
  • One item can sit on many trails, so paths compose instead of partitioning the collection
  • Side trails branch off a main line of inquiry and carry annotations of the reader's own
  • A trail can be copied whole into another person's store, making a path of reasoning a transferable artifact

Relationships

  • Memex — is the store this mechanism runs on, and supplies the two preconditions it needs — items held side by side, and a permanent place on each item to write the join into
  • Selection as the bottleneck of the record — is the constraint this mechanism attacks, substituting a link the reader laid down for a descent through someone else's index
  • Cross-run observational memory — carries the same insight into agent systems — recording the path taken, not only the result reached, is what lets later work resume from prior work instead of re-deriving it

Applications

Designing knowledge bases where the link rather than the folder is the unit of organisation, and where backlinks are reciprocal because the join lives on both items; publishing the path of an investigation alongside its conclusion so a reader can re-walk it; reading bidirectional-link note systems, hypertext and shared reading lists as implementations of an eighty-year-old specification.

Sources

  • https://www.theatlantic.com/magazine/archive/1945/07/as-we-may-think/303881/

See Also