Walter Underwood wrote:
> We punted explicit support for ads, so they will continue to show up
> in content and cause more work for Bob.
Uhh... It isn't just me. This work needs to be done by just about
anyone who builds an aggregator if they implement duplicate detection based
on analysis of item contents. Hopefully, that will be much less necessary
with Atom feeds in the future; however, as long as we've got RSS we'll still
need to do text-based comparisons to detect "updates."
The problem with explicit "ad" support was, of course, that people
would be highly likely to use the "ad" tag as a license to delete content
and strip the ads.
I've been thinking that there might be a more general solution in
allowing people to tag certain elements, spans or divs as dynamic content
which could be masked during canonicalization prior to change detection.
Such a generic method of tagging would eliminate the major "damage" done by
ads while providing support for other kinds of dynamic content that doesn't
involve ads. For instance, we've got Slashdot and now Spaces that change a
"comment count" in posts whenever someone creates a new comment. That wrecks
havoc on change detection code... Also, at PubSub, we'd love to insert data
like the current LinkRank or LinkCount, and various other dynamic bits that
should be ignored when comparing two items. If we could create an attribute
like "dynamic" (much like the nofollow attribute) that could be put in
elements, then comparison code would know what bits to ignore -- whether or
not they were ad related.
<span dynamic>
The time of day is: 5:25pm EST.
</span>
bob wyman
