re: OHS Comparison Categories

From: Christina Engelbart (
Date: Wed Mar 04 1998 - 08:51:00 PST

[OOPS - I included the wrong message above, duplicate of Differentiators
instead of the Comparison message - here is the CORRECT one]
Here's another contribution from John Rothermel from 1994. John
conducted a 6-month IR&D project at then ESL titled "OHS Technology
Evaluation Project" with the Bootstrap Institute. This summarizes the
categories he delineated for comapring other systems to the OHS model.
Date: Mon, 30 May 94 17:53:58
From: John G. Rothermel <jgr>

In doing comparisons of functionality among different hypermedia
systems, these are the categories I came up with. Other capability
lists (like Jeff Conklin's) are a lot shorter, but seem to be at such a
high level that a lot of important differentiators get left out (e.g.
many of the systems that look basically the same at Jeff's level, don't
satisfy my draft-passing litmus test).

Anyway these are the categories and subcategories:

Information Organization

   Filing - the ability to store and retrieve documents (or parts of
   documents) in named containers (folders / directories / Locator files
   w/ links)

      Items Filed Multiple Places w/out Copying - An item can be filed
      wherever it makes sense. But each filing just points to the
      single system-wide copy. (This is important because if users do
      ubiquitous copying, the linked discussion and other info
      management context of the original is cut off and lost at that
      point as discussion clusters around the copy - e.g. you can get
      two islands of discussion growing around two copies - and the
      twain should meet but won't.)

      Folders Filed Within Folders - the filing capability should
      recursively scale up.

   Linking - the ability to create and follow links between (parts of)
   two documents

      Fine-Grained Location (Src/Dst) - i.e. supports a link from an
      exact spot in a document, to an exact spot in the target document.
      (E.g. v.s. whole file to whole file.)

      Fine-Grained Extent (Src/Dst) - the area of contiguous "text" that
      the link is referring to (from). Should be explicitly delineated
      for the user (reader). Should include sub-object,
      multiple-object, and partial-through-partial object extents (vs.

      Semantic Direction - supports the notion of from-to so that the
      user can sort out semantically-typed links (supports, precedes,
      caused), i.e. which end of the link supported, preceded, or caused
      the other. This is in contrast to...

      Can Follow From Either End - Links should be followable from
      either end (e.g. backlink support). AKA "bifollowable links"

      Storage - e.g. links are embedded in the document itself, in a
      separate associated file, in a DB, or what. [This is in some
      sense a lower-level design (or implementation) question, but has
      repercussions for higher level functionality - like distributed vs
      local / disconnected-portability, access control, whatever.
      [Common academic wisdom is that the link info should be in a
      separate DB - and the documents should remain untainted when not
      in use with a tool that supports linking (Intermedia); but other
      practical experience says this is unnecessary overkill/overhead.]

   Tagging - the ability to associate metainformation (keywords,
   attribute-value pairs, creator/modifier id/timestamp) with
   information items (documents (or parts of documents), filing folders,
   links, or any other information organization construct like "pile" or
   "information set"), in support of (a) retrieval and (b) display
   filtering of the items.




      Multivalued Tags - I.e. supports "property lists" like:

         (attribute, value1, value2, ..., valueN).

   Metainformation - things like indices or catalogs of collections of
   documents (or any other information construct). Stuff that tools can
   hook into in support of finding, displaying, and generally managing
   information. (Tags are an example of metainformation, but this
   category is trying to get at larger collections of metainformation,
   grouped together in order to support finding and organization (so I
   guess a folder would be a primitive example at this level).

Document Management

   Document Architecture - one continuous hunk of text; v.s. allows
   segments of various media (static embedding e.g. via copy/paste);
   v.s. data segmented by the type of data handled by the originating
   tool (where the tool can still work on that segment of the document
   in place (e.g. OLE) (AKA "live embedding")

   Annotations - per-user commentary associated with a sequence of text
   in a document.

   Highlights - per-user background coloring/markup over a sequence of
   text (like link extents without the link). (Along with per-highlight
   metainformation (name, etc.) provides "bookmarks" with extent.)

   Linear Versioning - standard "supercedes" style like DEC file systems
   support (Tops, VAX). Latest is best.

   Parallel Versioning - multiple versions of the same basic information
   co-exist with slightly different packaging. None is any better than
   the others, they're just different takes on the same thing (e.g.
   executive version, implementors' version, instructors' version,
   beginners' version).

   Access Control - control at what levels...

      Group Granularity - owner, group(s), world, system, access

      Information Granularity - whole document, chunk within (object,
      extent, whatever), etc.

Change-Detection / Notification - This gives a *per-user* indication of
what information items have changed out there in the infoverse since
last seen by that user. "Passive notifications" are flags that the user
sees while browsing the info organization (e.g. asterisk by modified
document or folder name, or different color/font) "Active
notifications" are requested by the user for specific documents or
subsets of the filing/linking structure, and user gets pinged by the
system if and when a subsequent change occurs (file updated, folder
contents changed, etc.)

   File Modified (specific chunk?)

   Folder Modified

   Link Web Modified

   DB Updated

Information Visualization

   Filtering - can specify exactly what to present to the screen, or
   exactly what to leave out, based on metainfo attached to the item
   (tag, author, date, etc.)

   Info Organization Structure Display - navigation maps, graphical
   displays of filing or linking structure, etc.

   Graphs / Plots of Tagged Info - e.g. throw all files/chunks in this
   folder onto a timeline plot based on their "Date" tag. Or, ditto
   onto an x/y plot with tag "Cost" on the x axis, and tag "Performance"
   on the y axis (where each item has tags like (Cost, $ 2000) and
   (Performance, 300 MHz)).


   Concurrent Editing

   Bulletin Boards

   Screen Sharing

   Role Support

   Electronic Signatures


   User Written Scripts

   In-house Programming - i.e. source code available, or API available

Data Interoperability

   Within Environment - e.g. copy/paste, v.s. data from various tools
   can be collocated (e.g. in multimedia documents), v.s. data generated
   by one tool can be operated on by another

   With External Environment - import/export, active translation on the
   fly, Augment Reach, API support

Ok, that's all I've got. I sure don't see this as an end-point - just
something on which to start hanging dialog. Next on my agenda would be
a comparison of Augment, VNS, and Lotus Notes using these categories.

Also, despite one box dedicated to "Interoperability Within
Environment", this comparison list doesn't really capture the
interoperability (typically *non*interoperability) of the capabilities
themselves - which is what really makes an environment fly (second-order
effects). It'd be interesting to plot some kind of interoperability
matrix among the various capabilities for each system to be compared; I
bet most would be sparse.

(As a designer, you'd like all capabilities to be orthogonal and play
naturally with each other w/out special cases or dependencies. And in
some sense, you want a minimal set of constructs (nouns as well as
verbs) that are general, and can be applied in a wide variety of
situations. This yields high conceptual integrity.

E.g. Augment's locator files (files of (commented) links) are a great
example of subsuming the concept of filing folders / directories.
Locators are more powerful because (a) they're just another document to
which all the powerful document organization/manipulation capabilities
apply (like editing(!) and structuring), (b) they can point to more
fine-grained pieces of information than whole documents, and (c) can
contain commentary or other metainformation about the item pointed to.
(But it needs to be easier in Augment to do traditional folder
management operations like displaying the list of contents in various
orders, managing file deletions and renames, etc.))

Well, I'm dried up for today. 'Nuff for now.



This archive was generated by hypermail 2b29 : Thu Jan 06 2000 - 11:01:17 PST