So writing effort posts is hard work.
And then, responses are predominantly from people
who've misunderstood or not read the article, quibbles
about minor and mostly
unimportant points, and so on.
If someone strongly agrees, you may see a few positive
responses with a short quote followed
by "<-this", or
similar, but this doesn't scale well
- to the point where I think most people will refrain from
this most of the time, and certainly where it would be a
repeated action.
Thus, there is an imbalance to the responses in a
negative direction, which may be disheartening to some
authors.
Some commenting systems allow rating of posts in one or
more dimensions. Which shows a more balanced
evaluation of the post's standing, but
only as a whole.
Much less common on the web are annotation systems, which allow
selective markup of a /section/ of a comment with a
further comment, perhaps
recursively. I think I saw this demonstrated on TV, but in
practice I've never seen them live on the web; I think
they're just too hard to manage
in an active commenting system. Possibly hard to
maintain, hard to learn to use, and perhaps in practice
hard to read.
It may work well if a collaboration is curating a larger
article; Microsoft Word has collaboration functions which
does this semi-effectively, but again I
don't think this works at scale.
I propose a hybrid system which combines the location-
specificity of annotations with the ease and scalability of
rating.
It is suitable for one or very few axes; I will discuss the
case of a single bipolar axis, which I will assume is
"agreement".
The gist is that the reader may select a portion of an
article and rate it up or down, depending on whether
they agree or disagree - at which
point the highlighting of that section is varied in an
incremental manner.
This means that sections with which everyone agrees will
be strongly highlighted in the positive manner (typically,
but not necessarily rendered
as a coloured background), sections with which everyone disagrees will
be strongly highlighted in the negative manner.
If, for example, a few users have agreed with a section, it would be weakly highlighted.
My initial thought on the granularity of this highlighting
is that in general it could be the sentence. This would
limit the computational costs of
tracking, while allowing a good measure of expressivity.
I hesitate to define how the highlighting intensity should
progress; some experimentation would be required.
However, it may be that if
someone selects a small part of an article, this should
count for more than where someone has selected the
entire thing. But perhaps not
proportionately. An overall scaling could be applied
dependent on the number of raters, their tendency to
rate and the number of viewers.