[OLPC library] Annotations, revisited

[Samuel Klein]

On Thu, Jan 14, 2010 at 11:28 AM, Walter Bender <walter.bender at gmail.com> wrote:

> This already works as part of the Read activity, AFAIK.

I wouldn't say that annotations aggregate; they are simply a bit of
metadata that can be associated with an instance of a document.  At
best, I could get an instance of your doc with annotations, and later
add my own to that instance.  But unless we carefully exchange the
latest version of the doc between ourselves, and delete previous
versions of that doc on our systems, things get confusing.  And
there's no way for two separately marked-up documents to merge their
annotations.

>>  - to have a global namespace for comments/annotations/reviews
>> allowing searching for such annotations made by other people
>
> That is not something we do yet. Is there a good exemplar to look at?

This is an unsolved part of the puzzle that Andreas was particularly
interested in.  It's generally unsolved for comments/reviews/notes on
works or products of all sorts, though many siloed solutions have been
tried over the years.


>>  - for the above to be integrated simply into the normal process of
>> discovering, opening and reading a book
>>
>
> I am not sure how to integrate the global namespace into the normal
> process if connectivity is not universal. We really need to be careful
> re how global network access is tied to Sugar activities as it breaks
> things for children in many places: Peru, Nepal, etc. The "cloud" is
> not a practical reality yet in many (most) of our deployments.

It is if you consider a group of students in one classroom to be a
local cloud.  The cloud model maps pretty well onto that scenario.
Right now the extent to which we support sharing annotations (from the
first point above) doesn't integrate simply into discovery, opening,
and reading.

* local-network discovery doesn't work.  It's a push sharing model
only; I have to be online and actively sharing a doc for you to find
it.
* tracking of variations on a file over time doesn't work as expected.
 That makes it hard to aggregate comments even within a single machine
and user over time on similar docs.
* clustering of documents, activities, or other entities by name or ID
across the local network doesn't work.  If I share a doc with you and
you reshare it with others and it comes back to me, there's no way to
associate the new version I encounter with the one I already have.
* 'find a book' search (when searching an online or schoolserver
library) doesn't support searching annotations, or even finding
whether annotations exist -- annotations are treated as something
transient associated with a document found on the network or in a
journal, not something lasting.


SJ