xFolk thoughts

I agree with Tantek's sentiment to keep xFolk simple unless
there is  
justification to make it more complex.  When I wrote xFolk,
my  
intention was to have a format that would allow tagged links
to be  
harvested, not so much to display lists of the links that
were thus  
harvested with proper citation back to the linker and
reference to a  
harvest date.  That's great because people get what it's
for without  
too much trouble, and as a result they don't have too much
trouble  
implementing.

At the time I wrote xFolk (last iteration was last summer) I
thought  
that by convention, you could assume that such a tagged link
came  
from the page on which you found it and was authored by
whoever  
authored the page.

Now, I can see how that might not be satisfying to many, but
I also  
think the issues of author and authored-on-date is outside
the scope  
of xFolk.  These issues may be in the scope of something
like hAtom  
which seems to make designations about author and author
date for  
whole blocks of content on a page.  You could also quite
possibly use  
a citation format of some sort if you wanted to say that a
particular  
tagged link came from a particular source.

The problem with creating specific solutions for things like
author  
and authored-date at the xFolk level is that you can come up
with a  
billion implementations for each specific case.  Why would
you do that?

Bud
On Aug 27, 2006, at 14:59, Tantek Çelik wrote:

> On 8/27/06 11:11 AM, "Brian Suda"
<brian.sudagmail.com> wrote:
>
>> I have been playing around with xFolk and now that
more sites are
>> beginning to implement this along with Pingerati
indexing them. I had
>> a few thoughts that could be added to the spec.
>>
>> 1) There is no date for when the link was tagged or
added. This would
>> be an optional property. It could also be extracted
from  
>> "surrounding"
>> microformats. It makes sense that if you are using
hAtom (which as a
>> datetime published and updated) that you could use
that for the xFolk
>> if the xFolk is contained within the hAtom entry.
>
> Brian, based on what real world link tagging examples
do you think  
> there
> should be a date in xFolk?
>
>> Maybe there was a reason to leave-off any sort of
date, can  
>> someone clarify?
>
> Other way around.
>
> We don't need any reason to leave off anything. 
Things are left  
> out by
> default.
>
> There needs to be a good justification to put things
in, starting  
> with being
> part of the 80/20 implied schema of existing use in
real world  
> examples
> published on the *Web*.
>
>
>> 2) Who added this link? Another optional property
to nest withing an
>> xfolkentry could be an hCard and/or if the page has
an <address
>> class="vcard"> then that can be used
as well. I know it is pretty
>> obvious that the owner of the site posted the
xfolkentry, but once  
>> you
>> begin to aggregate xFolks links, we should give
some sort of credit
>> from whom the xFolk was gleaned.
>
> Again, ditto.  See above methodology.
>
>
>> I know xFolk is purposely pretty lightweight and
these two items
>> should be optional. What do people think?
>
> Just use hReview to "review" the URL. 
hReview has dtreviewed and  
> reviewer
> to handle both of these.
>
> Keep xFolk simple as intended.  Note that you can
overlap and  
> include both
> xFolk and hReview semantics if you wish on tagged URLs.
>
> Tantek
>
> _______________________________________________
> microformats-discuss mailing list
> microformats-discussmicroformats.org
> http://microformats.org/mailman/listinfo/microforma
ts-discuss
>

_______________________________________________
microformats-discuss mailing list
microformats-discussmicroformats.org
http://microformats.org/mailman/listinfo/microforma
ts-discuss

Thanks Bud for your thoughts. I know xFolk was developed
pre-microformats.org, so a lot of those ideas and
discussions where
never documented into the archives.

On 8/27/06, Bud Gibson <budthecommunityengine.com>
wrote:
> I agree with Tantek's sentiment to keep xFolk simple
unless there is
> justification to make it more complex.  When I wrote
xFolk, my
> intention was to have a format that would allow tagged
links to be
> harvested, not so much to display lists of the links
that were thus
> harvested with proper citation back to the linker and
reference to a
> harvest date.

--- if that was the original intention, they i can't argue
too much with that.

> That's great because people get what it's for without
> too much trouble, and as a result they don't have too
much trouble
> implementing.

--- I agree, low barriers to entry are key. The two main
points i
brought-up could be made optional - which wouldn't make it
any more
difficult to implement.

> Now, I can see how that might not be satisfying to
many, but I also
> think the issues of author and authored-on-date is
outside the scope
> of xFolk.

--- i would argue that it is only outside of the scope
because you
wanted it to be. The implied schemas from all your examples
have a
Date and Time which you choose to omit in favour of
simplicity. You
are modelling what people are ready publish, but you are
only modeling
a portion of it.

> The problem with creating specific solutions for things
like author
> and authored-date at the xFolk level is that you can
come up with a
> billion implementations for each specific case.  Why
would you do that?

--- i don't think i follow you here? how would xFolk
specifying an
author and time stamp create a billion implementations? i
would think
that the ability to nest an arbitrary microformat inside
creates the
billions of implementations... now i have to look for hCard,
but not
when it is inside an ITEM property, but if it is inside a
location in
a vevent it is something different etc....

If your intention was never to add these things into xFolk,
then that
is END OF STORY - no sense arguing. I was just confused at
why things
that are present in all your examples were not encoded into
xFolk -
and i think you have answered my question and now it is
documented and
we have something to reference in the future.

Thanks,
-brian

-- 
brian suda
http://suda.co.uk
_______________________________________________
microformats-discuss mailing list
microformats-discussmicroformats.org
http://microformats.org/mailman/listinfo/microforma
ts-discuss

On Aug 27, 2006, at 5:04 PM, Bud Gibson wrote:

> I agree with Tantek's sentiment to keep xFolk simple
unless there  
> is justification to make it more complex.  When I wrote
xFolk, my  
> intention was to have a format that would allow tagged
links to be  
> harvested, not so much to display lists of the links
that were thus  
> harvested with proper citation back to the linker and
reference to  
> a harvest date.  That's great because people get what
it's for  
> without too much trouble, and as a result they don't
have too much  
> trouble implementing.

One thing I find useful in folksonomies is that they can
represent  
popular opinion on a particular topic at a particular time.
One  
benefit of adding dates to individual items in folksonomies
is that  
it allows an aggregator to more aptly weigh tags over time.
This may  
not seem all that valuable at this point, but as the web
ages, I can  
see it becoming a more important issue.

Here's an example: Links to Mel Gibson's bio page (not
implying any  
relation to Bud). In aggregate, over the years, Mel will
have been  
tagged as handsome, professional, talented, etc...he was
long held in  
high esteem by lots of folks. Recently, though, his actions
have made  
him sort of a parody of himself.

So, an aggregator that's displaying info about Mel
Gibson's bio might  
still show that he's considered more handsome,
professional, and  
talented than irresponsible, because the aggregator will
have  
collected more of those positive links in volume, even
though that  
doesn't show an accurate picture of the public perception
of him  
right now. Conversely, in a year, the tags may have caught
up with  
him, reflecting a more accurate current popular view,
completely  
eliminating the fact that he *was* at one point
well-respected.

One could argue, then, that the history of the folksonomy of
a  
particular bookmark is irrelevant, and we don't want to
facilitate  
people looking  back, but if that's the case, one must also
consider  
that a week after Mel's arrest, when you look at the
folksonomy of  
his bio page, what it is displaying is, in fact, nothing
more than  
history -- and not the current popular opinion -- if the
tags aren't  
weighed by date, but only by volume.

It could be that we expect aggregators to simply datestamp
entries  
when they're scraped, but that necessarily puts any new
aggregator at  
a disadvantage to its earlier competition -- scraping old
items  
without dates is the same as new.

I have no answers, but thought this might be an interesting
enough  
issue to explore.

Best,
Kerri


_______________________________________________
microformats-discuss mailing list
microformats-discussmicroformats.org
http://microformats.org/mailman/listinfo/microforma
ts-discuss