Timestamps are in UTC.
tantek: ping
we need irc ping-o-matic
yakk, what do you mean by that?
if you ping tantek, then you also ping everyone who's interested in what tantek thinks
it's a pub-sub model
B-)
or it could ping across all the networks
or it could proxy pings to other protocols
we don't actually need this - it was just a silly idea :)
<yakk> or it could proxy pings to other protocols
Like poke-in-the-arm protocol
Are ya busy? Hmm? How about now? Can you talk? Hey?
poke poke poke poke poke poke poke
yakk: talk to termie
he has already built that
synced w/ his cell phone
bergie is lives in Finland and blogs at http://bergie.iki.fi/blog/ and Midgard CMS developer
[[citation-brainstorming]] M http://microformats.org/wiki?title=citation-brainstorming&diff=0&oldid=6522 * JoeAndrieu * (+0) Schema -
Atamido is Paul Bryson, http://orangeman.commo.de/
[[picoformats]] N http://microformats.org/wiki/picoformats * Chris Messina * (+575)
[[picoformats]] M http://microformats.org/wiki?title=picoformats&diff=0&oldid=6523 * Chris Messina * (+48)
[[Talk:Main Page]] http://microformats.org/wiki?title=Talk:Main_Page&diff=0&oldid=6524 * Rrrrrrrrr * (+16135)
[[Talk:Main Page]] M http://microformats.org/wiki?title=Talk:Main_Page&diff=0&oldid=6525 * Tantek * (-16135) Reverted edit of Rrrrrrrrr, changed back to last version by RobertBachmann
[[Special:Log/block]] http://microformats.org/wiki?title=Special:Log/block&diff=0&oldid=0 * Tantek * (+0) blocked "User:Rrrrrrrrr" with an expiry time of infinite: spam
bergie is lives in Finland and blogs at http://bergie.iki.fi/blog/ and Midgard CMS developer
http://www.25hoursaday.com/weblog/CommentView.aspx?guid=96747D35-9190-483B-95E2-5B1CCAB48DF2#ab1f016e-400c-409c-9ea2-71a7bf9b2f60
trovster is a web developer from the UK who writes on http://www.trovster.com and runs www.csslounge.co.uk
Jonnay is a programmer, graphic designer and musician. He blogs at http://blog.jonnay.net and his music is at http://www.jonnay.net
gsnedders is a 14 year old idiot from Scotland and pretends to have a website at http://geoffers.uni.cc/
bergie is lives in Finland and blogs at http://bergie.iki.fi/blog/ and Midgard CMS developer
[[screencasts]] http://microformats.org/wiki?title=screencasts&diff=0&oldid=6526 * DrewMcLellan * (+120) this year -
pnhChris is Chris Casciano, blogs at http://placenamehere.com/ , and a member of the Web Standards Project.
dglazkov is Dimitri Glazkov (http://glazkov.com) and lives in Birmingham, AL, USA (-6:00 GMT)
oh look.. cork'd has hcards in their profiles
briansuda is brian suda of X2V fame
KevinMarks: dupes in the flickr listings? http://kitchen.technorati.com/contact/search/casciano
but not for this one: http://kitchen.technorati.com/contact/search/quincejam
oh you know what
hmmm
is it picking up on changes to friends lists
or other content
and relisting
i wonder
that would explain that difference briansuda ... recently added contact
yeah, the search is conducted over the WHOLE document, but ONLY the Microformat content is presented in the results
not exactly how i'd do it, but it's not my company
yeah.. it really could use some more advanced search options
but some of that gets real messy
"search for chris in fields 1,2,3,4,5,5,6,7"
but i'm not sure how to resolve the update to a card and re-ping issue
i guess delete all old ones at that url... can't really rely on fragment identifiers
but its not my problem either
directly at least
the more interesting question is how can we publish data to make results like this -- http://kitchen.technorati.com/contact/search/chris -- be smarter
and more useful
the other issue which will start to pop-up is the flooding of invalid data. I can add an hCard about you and put loads of false information.
which is a nearly impossible task... at least that's the answer i get after thinking about it for 30 seconds
pnhChris, you're right that it's picking up any change to your hCard contents
tantek: no, i think it is.. with multiple pings
but storing / displaying them as multiple records
we're being a bit liberal with showing changed content
yes
we'll likely start collapsing some of that relatively soon, but for now found have left it in to help with debugging etc.
it is definitely a bit more of an "open" technology preview in that regard
.. but espeecially with sites like flickr that hold some info private.. there really isn't anyhting stopping me from registering as bria suda with public info X,Y + Z that correlates to the same attributes as your "real" data
we really wanted to give the community something to play with soon and iterate based on feedback from the community
.. flickr being benign... unless you're posting goatse
the problem of spoofing content on the web is certainly not a new one
but s/flickr/a white power friendster/ or something else you wouldn't want to be assoicated with "you"
nope
but making assumptions during aggrigation in a bit newer
or at least
requires understanding that
hence why there is very little aggregation in the results right now
one of the big reasons I developed XFN with Matt and Eric was to provide at least the beginnings of a building block for various forms of distributed implied trust and recognition
even then, you'd have to control all the points... or a list of all the valid points
no, you don't have to control "all the points"
it's more about raising some results above others
due to greater implied trust/recognition
rel="me" let's you explicitly indicate that various different profiles are the same person for example
those can obviously be strongly preferred in any aggregation
so the white power friendseter adds XFN
and you're back to an identical record as the one on friendster
See http://gmpg.org/xfn/and/#idconsolidation for more
no, you're back to an island
the point is that by other folks indicating that you are a colleague etc., that information is implied to be more trusted
the first occurrences of such bad data will likely be spam rather than any kind of overtly negative stuff as you suggest
at least if present behavior on blogs is any indication
sure
but the implications on aggrigating for less spam-worthy results/names .. and common names in general (chris) are there as well
but pingerati doesn't do just blogs
right
but the fact remains the same
economically spam provides more incentive than hate
before blogs people spammed meta keywords etc.
its definitely a wider data problem
i dunno
but from a publisher standpoint i'm not looking to solve /that/ problem as much
as i am providing enough info to correlate records
need some sort of voting/karma system for microformats search :)
and that's where spoofing on an individual leve becomes a problem
.. or linking valid recods
the beauty of decentralized database is that the aggregator has control of what to include or not
instead of needing to control what to store or not
... and doing so for people / contacts that don't neccesarily have a "home page" where they can house a list of all their mentions with rel=me
cgriego is Chris Griego (-06:00)
just rambling at this point, i know.. but its what hits me when i see the search results starting to build records wise
dglazkov - XFN provides sufficient information to imply some sort of karma
though I'd have to think hcard would imply some of that karma
vcard class="url" says "them" just as directly
(as in my chunkysoup.net vcard.. with url of placenamehere.com but no xfn)
no, hcard does not imply any karma - it only self-describes
hyperlinks to *other* folks imply some karma, as modern search engines have already taken into account
hyperlinks+XFN take it one more step and imply even more
so is address class="vcard" class="url" not descriptive enough to make assumptions about connectedness.. .to say its soemthing that the thing at the url that wrote?
seems silly to me
does your current search do anything w/XFN?
i'm not sure what it "says" sematically if that's not it
be aware that the current proposal to include IM accounts is class="url" href="aim:foobar"
it says that is one of the urls associated with that hCard
dglazkov, we're indexing some XFN but we're not surfacing it yet
but what do you do with that on aggrigation
cool
certainly more then any old link to that URL
no?
if not, why?
btw, tantek, are you fully parsing HTML (tidy, etc.) or pattern-matching until you see what you're interested in?
Chris, perhaps. It will take some more analysis.
dglazkov, we're using a combination of checking the fields and doing some relevance as well
just because XFN is an established way to make that connection, i don't think that leaves hcard implying nothing
the first cut was to make it as easy as possible for people to search
watch this space for project announcement
i mean any aggrigator can certainly decide on their own how they want to use that information
dglakov, great!
I'm not saying it implies nothing Chris, I'm just saying it doesn't imply karma
it is not a measure *between* individuals
whereas XFN is
the 'url' property in hCard/hCalendar etc. *does* imply a greater association for aggregation, but does not imply any karma, or any kind of "recommendation" of one individual by another
then we're talking past each other.. becuase i'm talking about linking records from a publishers standpoint so that aggrigators can work with the records better
and making that connection between, say, placenamehere and chunkysoup
you're not talking about any form of "karma" or reputation as classically discussed then
or joe smith at lfickr and jo smith at corkd
that's identity consolidation
and accomplished with bidirectional rel="me" links
for one, we've been using rel="me" with Technorati profiles and embeds for quite some time
so when you claim a blog, it links your profile to your blog with rel="me"
and if you include the embed code in your blog and have it link to your profile, then it does so with rel="me" as well
but as i said.. if one doesn't have a "home page" where they can provide the corresponding rel me
or hasn't
then you have no place to put the hCard right?
the "if one doesn't have a home page" argument is basically akin to the "if one doesn't have an email address" argument of about 20 years ago
it's effectively a non-problem, or will soon be so
the reciprical linking requires a place where i can keep links to 30 sites I have profiles with and keep up with whichs ones hav epublic or private profiles as i subscribe to them on a whim
with 70m+ people with their own myspace profiles etc., this is not a problem
though most of those proviles, in the context of your new search.. if they get listed will have the one way hcard class=url
and what do you do with that.. or how does that factor in
was all i was going for
you were working on a different problem
mlinksva is Mike Linksvayer and from Creative Commons
anyway
as i stated.. its mostly just rambling from me... and .. well.. don't have much more time for that
:/
not today anywa
way
i guess to me the question is what is the goal of searching for hcards.. its to get a specific peice of information on someone, to build a big profile/consolidated vcard, or to see where what they touch (profiles, comments, blog posts, mentions)... xfn only helps with some of that... and relying on xfn hurts some other (e.g. hcard'd blog comments that will never have xfn, and thus may drop in certain search output that relies o
and with that
xfn is for indicating the social ties from which you can infer other facets of trust
publishing and searching hCards essentially enables a world wide distributed addressbook
RobertBachmann is Robert Bachmann <http://rbach.priv.at/> and lives in Austria (Timezone: UTC 02)
but i'm having trouble seeing that jump from publishing individual snippits of contact information makes individual consumption on an individual basis easier as well as offering some context clues to something that is the basis for a world wide address book
especially with short hcards in blog posts, comments, mentions, etc. that have little value to add to that global address book, but might have more value in identifying connections
if netflix goes behind my back and makes a public profile page that gets indexed.. its certainly me
that doesn't change that its a prfile that can be connected (via email or url or whatever else) back to me, or not
so how do you aggrigate anything with that level of detail
or do you not
and if you dont', what benefit is searching the web
for contact info
or you get a subset of data for one or two promeinet people with lots of friends and the rest gets lots
lost
.. becuase you just can't aggrigate the rest
stuff that works on the individual level... i can go to my friend jo smith's web page and import their data to my address book and know i have that information.. but i don't know how searching the web for hcard'd info would be a useful alternative
particularly if they weren't my friend where i'd konw their home page already but someone that i needed to chase down
searching the web for hcard'd info may be how you would find your friend's contact info web page in the first place
that's the point
people use traditional search engines right now to search for other people
and you get tons more false positives
hCard improves upon that experience significantly
and not only that, but eliminates the "copy paste each piece of info about the person one property at a time" tediousness once you *do* find that contact info
it certainly won't be perfect or solve all possible problems
but then again, that's not the point, nor is it a goal of microformats
we leave perfection and solving all possible problems to smarter people with more time on their hands ;)
tantek, if i search for you i get my flickr page because you are a contact of mine, and my hCard appears. That's pretty close to a false positive
not really brian
it does indicate some relevance which is acurate
accurate even ;)
of course we'll continue to tweak the algorithms and improve them as the dataset grows and we gain more experience
you can imagine how when those contact lists/links are marked up with XFN, then it becomes even easier to do so
and think about it this way brian, if someone were to search for me, and all they found was your flickr page, at least now they may have potentially found a way to contact me
so I don't think it is a false positive at all, it's just of less relevance than my flickr page
true, but as the index grows it might become overwheling to get 12,000 results for 'john'
how is that any different than looking up "Smith" in the white pages phone directory?
and yes, the world is a big place with lots of people
although you might be able to do some really interesting 6-degrees of seperation with the results
right - that's where the XFN will come in
The difference is that you KNOW 'smith' is the last name, where as with your search 'smith' is on the page
the difference is semantics versus occurance
right, so you'll want to list results with the first or last name of "smith" before those with additional-name, before those with mentions or nearby text etc. -- that's the point of relevance ranking
agreed.
and when we do end up with hundreds of millions (or even billions) of hCards on the web, it will present a very interesting challenge to user interface designers
how to build a UI for an addressbook of that many people?
even today's social networking sites have problems with that
even *personal* address books are growing - it is not unheard of for people to have thousands of contacts in their address book, and current address book UIs just don't cut it
(these numbers are true despite many so-called experts clinging to the overly hyped "150" number etc.)
but these are good challenges to have
good opportunities for people to figure out new applications, new services, new interfaces
the problem isn't just that there are billions of hcards out there.. its that 500 of them are "me" because of the way hcard is used outside of profile pages that I directly have a hand in creating
and all have various levels of detail and quality
well, both are challenges
think about it this way, computers and PDAs are just barely getting to the point where they can sync the contacts of *one person* somewhat reasonably - even now, all cases that I know have some data loss due to inconsistent feature support
syncing (or merging/aggregating, same difference) the contacts from *multiple sources* is even more challenging
yet not impossible
it's just going to take more clever code etc.
which is why i can see the use on an individual import basis of hcard
but i can't make that leap in context
and to some extent, again your problem of "hcard used outside of your own profile pages" is the same as web pages today
you're not stating a new problem at all
or that leap in application from extracting info on something you have, to getting info from an object you don't konw about
you're stating an old problem, which hCard does a better job at than preexisting solutions
no, i'm just thinking very literally about the problem
with the things and uses i see infront of me and that i know are being published about one data item, me
and thinking of how to better manage that
they were being published about you before hCard even existed
and the mess was even worse
just not abstracting that into a broader content
that's easy to forget
sure
that's my larger point
we're not solving all the problems with just this one step
but we are making things better
of course there is still much to do
but i worry that there isn't enough definition to do much more then we did before
there's plenty
you know which hCards have more details
and which are merely tokens of your presence, like comment authorship
on an individual occuracne and with me looking it over, yes, i can see that
and similarly, search engines can tell that too, now that those details are marked up with hCard
whereas before, that was *much* more difficult (if possible at all), and of very poor accuracy
[[hcard]] M http://microformats.org/wiki?title=hcard&diff=0&oldid=6527 * PatRamsey * (+202) Examples in the wild -
yay... looks like I convinced a developer to index microformats from our crawls
if they don't I"
I'll work on it on my own time
it'll be nifty since no ping is needed
hober is Edward O'Connor and works for EVDB on http://eventful.com/ and lives in San Diego, CA (-08:00)
what are you crawling?
cgriego is Chris Griego (-06:00)
ah, Alexa. we could feed you page pings if you want them
hmmm
that might come later
dunno if you know about our new product
called AWSP
we are developing some samle apps
and I think I've convinced them to do a microformats search
dunno if it'll be all uf though or just hcard
I'm excited about it
I was going to start on it myself last night but I forgot my password
speaking of which
kingryan is ryan king
cgriego is Chris Griego (-06:00)
Excuse me, I have a question. Someone told me to go to freenode because I want to make IRC for my website but I cannot understand how to.
Could anyone possibly help me with this?
tantek is Tantek <http://tantek.com> and works on Technorati and develops microformats <http://microformats.org>
Excuse me could you possibly help me with something? I was told to go to freenode because I want to make IRC for my website but I cannot understand how.
question: Is there a place that plainly defines what a microformat is supposed to do without all the buzzwords? :-)
microformats allow you to encode useful information in a zero-sum kind of way
the technology that does it is already available, it takes advantage of pre-existing standards
right, I grok that. I guess what I'm trying to do is explain to someone how they differ from pre-existing Technorati tags.
but allows you to manipulate complex data-types in an way that is atomic from a user inteface perspective
standards driven, like XML?
actually microformats encodes all data in xhtml
so it's really a formalized convention for authoring xhtml
although you can use html
the reason behind that is because html/xhtml is already consumed with decent support
ok. So, for common things there will be common microformats - like the hcard example. But for particular things, there will be other tags that can be derived.
yeah, makes sense.
knowprose, note that Technorati tags use a microformat themselves called rel-tag
aha! that's what I was wondering!
technorati employs people that work on microformats
so there is a tight integration thre :-)
there
ok :-) Now I'm learning something.
the interesting thing is, I think Yahoo now employs more people that work on/with microformats than Technorati
knowprose: yeah, tantek is the connection there... tantek is a major force behind microformats, and also the cto at technorat
Well, I'm just someone trying to negotiate meaning so I can explain it to NGOs. Gotta know it to write it, that sort of thing.
NGO-folk have a tendency to ask for things that they don't need, and have a tendency to do things that they do need.
think of microformats as the way to publish and share information on the web with higher fidelity
yeah, you see - that's the kind of phrase that gets NGOs drooling and causes me headaches. :-)
for example, if an NGO wanted their contact information to be easily found and shared, they would publish it with hCard
Right.
or if an NGO is planning a series of events and wants more people to know about them and add them to their calendars, then they would publish their events listing with hCard
hCalendar that is
or use eventDB to take care of it for them
for "issue" NGOs, whenever they take a position on some political leader, some piece of legislation etc., if they wanted their evaluation/review/rating of those people/laws to be more easily found and passed around, they would publish such reviews with hReview
bewest, right
Makes sense. So now I understand the ground level. My disconnect was understanding the difference between technorati tags (which I already use) and microformats. That's solved.
Now I'm grokking this.
for all of these, to make it easier, the NGO should just use a tool or service that supports microformats
right. That's where I come in. :-)
:-)
but they key here is that microformats are simple/easy enough that the NGO's own web authors/designers can easily add them in themselves
adding microformats is easier than publishing an RSS feed for example
you don't have to be a programmer
anyone with decent (X)HTML+CSS authoring/writing skills can use microformats
That rules out most people within international NGOs. :-D
but it's closer, and they're learning.
considering disaster standards now. I'm on the WorldWideHelp team as well.
knowprose, the point is, pretty much anyone who is literate can be taught how to author HTML+CSS
Angelo and I worked out a consistent set of tags to use...
and thus microformats makes use of very widely available skill sets
tantek, I understand that, but anyone who is literate can pump their own gas. Some people choose not to. :-)
i like that. using microformats is as easy as pumping gas (or plugging in your electric car ;) )
but I do see the potential. :-)
don't say that! I had a Chief in the Navy who didn't know what the black round thing under his hood was! :o
in NJ you can't pump you own gs :-)
pnhChris is Chris Casciano, blogs at http://placenamehere.com/ , and a member of the Web Standards Project.
thanks. I'll stew on it and write about it later... I think now I can browse the Wiki and not get confuffled. :-)
bergie is lives in Finland and blogs at http://bergie.iki.fi/blog/ and Midgard CMS developer
[[what-are-microformats]] M http://microformats.org/wiki?title=what-are-microformats&diff=0&oldid=6528 * Tantek * (+1693) Add Yours Here -
briansuda is brian suda of X2V fame
hey briansuda
hello
so, I was working with the hcard tests earlier, and x2v doesn't seem to be working on test #35
not sure why
cool
OK, what version of X2V are you using?
latest from hg
what is the error? because it seems to work for me?
I have:
changeset: 42:12f6926962c2
tag: tip
parent: 41:2f8763a2bbca
parent: 38:77cef3da3046
user: unknown@t-lva-suda-co-uk.local
date: Mon Apr 24 10:20:09 2006 -0500
summary: merged with Ryan
you have the lastest version of the tests?
probably not... let me check
here's the error I get:
~/microformats/tests ryan$ bin/test-xsltproc.pl hcard/35-include-pattern
FAIL hcard/35-include-pattern
10,11d9
< FN;CHARSET=UTF-8:James Levine
< N;CHARSET=UTF-8:Levine;James;;;
oh, i was looking at test #5 not #35
reset, start again
and the web version doesn't work either: http://suda.co.uk/projects/X2V/get-vcard.php?uri=http://microformats.org/test/hcard/35-include-pattern.html
you know you can run just one test, right?
bin/test-xstlproc.pl hcard/35-include-pattern
yes
k
ok, i know what this is.
yeah?
this is an unresolved issue with the include pattern
what's the issue?
the IDREF, i was advocating that the CHILDREN of that be included, but in the test it is one the same element as class="fn" and NOT a child and therefore ignored
ah, gotcha
The easy fix in XSLT would then allow class="vcard fn ..." which we are avoiding.
i never got around to seeing there was another way, all these confernces got in the way
DanC, tantek and I all emailed about it, i'm not sure if it was on the list or not.
well, let's air the issue out and resolve it first
I don't think so
we should email the list and see if we can get consensus
i think it is on the mf-dev list
http://microformats.org/discuss/mail/microformats-dev/2006-April/000091.html
you're right, I found it
DanC's suggestion of changing .//* to descendant-or-self::* opens abit of a pandora's box
well, let's ignore x2v for the time being
'cause its not always going to be the only parser
(and its already not)
[[include-pattern-issues]] N http://microformats.org/wiki/include-pattern-issues * RyanKing * (+122) added link to issues email
These logs were automatically created by mflogbot on chat.freenode.net using a modified version of the Java IRC LogBot.
See http://microformats.org/wiki/mflogbot for more information.