New Account Types and data equivalency (was post TB 3.1 mailnews backend plans)
kent at caspia.com
Fri Jul 16 18:40:35 UTC 2010
I guess I'm going to have to quit saying the MoMo isn't really
interested in new account types! Glad to see that you are thinking about
To really get the value of an aggregated solution, we need to think
about the issue of data equivalency when you aggregate disparate
sources. If we don't we will end up with a mishmash of stuff that
doesn't really make sense as an aggregate.
The OWL spec from the RDF world has thought about this some
(http://www.w3.org/TR/owl-guide/) with concepts of "Data Equivalency".
To use a simple example, when is a Twitter feed coming from the same
person as an email, or a blog posting, or someone's Exchange email
account instead of the same person's home account? I'm not proposing
that we re-RDFify mailnews (still hate that name 'mailnews' by the way),
but we should not let our reaction to the RDF debacle prevent us from
learning something from them.
Looking even further ahead, one of the Big Ideas for Gigabird should be
a robust way to combine together disparate data sources in useful ways.
"Show me everything that idiot rkent said today so that I can rebut his
ramblings everywhere" might be a sample use case. How will we accomplish
this? We do it badly now, my address book usually has many entries for
the same person.
It applies much further than simply contacts. Items have many different
times, many different categories. Is a twitter # search field like
#moz10 more like a tag, or more like a folder? For that matter, is a tag
really any different than a folder?
It would help to start if we would define the basic model of what we
are. For example: "Mailnews provides an aggregate view of data that can
be represented as individual items in a tree of categories (aka
folders), with metadata for each item that includes at least a text
summary (aka subject) and body (which could be the expansion of the link
in a twitter message)." If we could define the basic underlying model of
out items, and how they are extended and aggregated, we could go a long
way toward defining the data model that Gigabird needs to succeed.
More information about the tb-planning