#indiewebcamp 2013-01-31

2013-01-31 UTC
zztr joined the channel
#
zztr tantek: when you say 'the web is my reputation service' what do you mean? is there a microformat for endorsement?
#
tantek zztr - yes see Barnabywalters's follow-up comments in IRC re: XFN, hCard etc.
#
tommorris and vote-links. ;)
#
tantek and for explicit endorsement, you can use hReview to post a review of a person on your own site, as you would post a recommendations/testimonials for someone on silos like LinkedIn, or even Flickr
#
tantek tommorris - sadly, vote-links support has died out so it's been moved to an "Archived" section accordingly on the home page: http://microformats.org/wiki/Main_Page#Archived
#
tommorris relying on @rev is a problem
#
tantek yeah, rev is toast
#
tantek plus vote-links were fighting the tendency of people not to link to things they disagreed with
#
tantek the only thing that got (overused/abused) in that realm is no-follow.
#
tantek which either means, don't consider any reputational weighting for this link, or, my CMS/silo/SEO developer was lazy or stupid and put no-follow on every link, or a random sprinkling of links on my site.
#
tommorris well, also, marking those things specifically as things you disagree with, albeit invisibly in a rev attr, isn't something we like doing much
#
zztr are microformats often/ever cryptographically signed?
#
tommorris nope
#
zztr any reason why not?
#
tantek zztr - is HTML ever cryptographically signed?
#
tantek answer that first, and you'll have the answer for microformats
#
zztr when it's in an email sometimes yes
#
tantek then yes
#
tantek since microformats are just HTML
#
tommorris well, there's a few things where crypto-in-HTML does take place
#
tommorris those "stick this meta tag in your HTML to verify" type things often use sha-1
#
tantek so if you want to cryptographically sign your microformats, just cryptographically sign the entire HTML.
#
tommorris yep, and custom headers would be better for that kind of thing
#
tommorris rather than having it in the body of the document. because you kind of then have a regress problem.
#
tantek tommorris - none of those are about the contents of the page itself right? because then there's a catch-22 where adding the meta tag changes the contents of the markup
#
tommorris nope
#
tommorris as I said, regress problem. ;)
#
tommorris calls up Bertrand Russell and asks him to cryptographically sign his set listing all sets that are not members of themselves.
#
zztr which is why it makes more sense to me to sign just a div. the html doc is a container for statements
#
tommorris well, what benefit does it give the user?
#
tommorris *this* particular div is signed. okay, what does that mean?
#
tommorris I can trust the facts I can infer from this div, but not the other divs on the page
#
tommorris also, digitally signing markup is a pain in the ass
#
tommorris because you aren't just signing the text, but the model serialization
#
tommorris it's a pain in the ass with a relatively simple document model like XML
#
zztr in a hypothetical future where people do this, if the browser had a database of known/trusted identities, statements could be published alongside each other. fred has a vcard, george has a calendar item, whatever...
#
tommorris with HTML, if someone changes a <br> to a <br />, are they equivalent?
#
tommorris so you have to run everything through a markup standardiser and get back a DOM object, or a canonicalised bit of markup, and sign that
#
tantek zztr - you may be interested in studying the non-hypothetical uses of the hCard "key" property: http://microformats.org/wiki/key-examples
#
zztr well, the signed pieces would have a life of their own, and the bytes from them would have to remain untouched. that's not such a big deal in my mind
#
tommorris also, let's say I had an hCard for John Smith
#
tantek tommorris - that's easy to work around, you simply include sigs for both the innerHTML and the innerText
#
tantek and if the markup changes but not the text, then you can still do some reasoning about the info
#
tommorris <div id="john" class="vcard" signed="some magical signing string" style="display:none">{some markup}</div>
#
tantek where you really get bitten is whitespace handling ;)
#
tommorris <div id="fakejohn" class="vcard">John Smith with fake details</div>
#
tantek anyway - I'd say start with studying the key-examples before going too far off the theoretical deep end
#
tommorris your browser will now happily tell you that the hCard representing 'John Smith' is crypto signed
#
tantek and then, add your site once you've published your PGP public key - that's imminently more doable/practical than theoretical crypto signing discussions ;)
#
tommorris only it's a different one from the one you are seeing.
#
tommorris where you might need crypto is where you have bits from one person hosted on another person's site
#
tommorris for instance, in some theoretical world, I might want to say to Tantek that I trust him at CSS.
#
tommorris so I could publish that on my own site.
#
tommorris but in theoretical land, he wants to publish that on his own site
#
tommorris so I give him some crypto-signed HTML he can put on his site
#
tommorris such that a future nonexistent user agent could verify it
#
tommorris and check that I actually made the assertion contained therein
#
zztr yep
#
tommorris and he hasn't changed it from being an expert in CSS to being an expert in dental surgery
#
tommorris but, well, okay, great. why bother?
#
tommorris why not just have me publish that on my own site, and retrieve that information from my site when you want to verify it?
#
zztr basically, my stance is that dns sucks as the identity service for the web
#
tommorris you have to make the radical decision to trust that the words printed on tantek.com represent the intent of the person named Tantek Çelik.
#
zztr yep
#
tommorris I'm kind of happy, for practical purposes, to accept that trust for most things
#
tantek tommorris for that kind of sharing various assertions crypto-signing stuff, check out the camlistore project by Brad Fitzpatrick and friends.
#
tantek he actually has something working - at a lower level, but fascinating nonetheless
#
tommorris maybe not if I were selling tantek a million-dollar penthouse in a London skyscraper... but for "I'll be getting a beer at a bar downtown next week" - those kinds of assertions, which are the sort of bulk of things we do on the social web - it's just fine
#
zztr content addressability is where it's at
Loqi, tantek, scor, mxuribe, mxuribe1, catsup, bnvk, andreypopp, Stevef and eschnou joined the channel
#
andreypopp any thoughts on why microblogging via XMPP didn't get traction? I see juick.com over here but it's kinda dead...
tilgovi, Stevef and andreypopp joined the channel
#
tommorris andreypopp: one reason. XMPP requires a always-on running server process.
#
tommorris which is fine if you are a geek like me. I'm sure I could run my own XMPP server.
#
andreypopp tommorris, I don't believe in total indieweb paradise when everybody has their server in the pocket, but decentralized architecture with several servers which federate data seems fine to me...
#
tommorris the problem i have is people need to design for the former
#
tommorris the latter will be a natural compromise
adactio, andreypopp, scor, laurian, mxuribe, friedcell, eschnou and eschenal joined the channel
#
singpolyma andreypopp: juick is a pretty good implementation of the µblog over XMPP idea
#
singpolyma I used it a bit, but network effects were of course an issue there
#
singpolyma For the record, HTTP *also* requires an always-on running server process
laurian_, tantek, laurian, scor, bnvk, barnabywalters, stereoket and eschnou joined the channel
#
tantek tommorris what was that en wp short url again?
#
barnabywalters tantek: would that be http://enwp.org?
zztw joined the channel
#
tantek thanks barnabywalters
andreypopp and stereoket joined the channel
#
aaronparecki.com uploaded /Special:Log/upload "uploaded "[[File:Tantek.com-2010-281-b1-translation.png]]""
wajiii-afk joined the channel
#
aaronparecki.com uploaded /Special:Log/upload "uploaded "[[File:Waterpigs.co.uk-article-chopped-up-translation.png]]""
#
aaronparecki.com uploaded /Special:Log/upload "uploaded "[[File:Aaronparecki.com-2012-10-22-1-translation.png]]""
#
aaronparecki.com edited /translations (+775) "/* Current Examples of IWC Translated Articles */ broke out into headers, added screenshots of links to translations" (view diff)
#
tantek ooh, translations page looking pretty aaronpk :)
#
barnabywalters tantek: why are translations of your content listed at the bottom?
#
tantek.com edited /translations (+140) "yellow backgrounds nothing to do with my translations" (view diff)
#
barnabywalters to me at least they appear more as badges than useful links for non-english speakers
#
tantek badges?
#
tantek because non-primary content goes in the footer ;)
#
barnabywalters but if the reader isn’t an english speaker, that link *is* primary content
#
tantek I don't want to distract readers from the content of my article
#
tantek yeah - I'm trying to figure out a good compromise
#
barnabywalters wonders how reliable the Accept-lang header is
#
tantek see for example how I styled/organized the "Translations:" near the top of http://microformats.org/wiki/hcard
stereoket joined the channel
#
tantek I'm considering something like that for my blog as well.
#
tantek easily skimmable/skippable by English speakers
wajiii joined the channel
#
barnabywalters yeah, I’ve never actually noticed those before :)
#
tantek but easily recognizable by non-English speakers by the presence of links in their own language
#
tantek right - that's a feature :)
#
barnabywalters they certainly haven’t distracted me from reading the main articles
#
tantek YOU as an English speaker are supposed to not notice them.
#
tantek I think they work because they're AFTER the initial paragraph, heading, authors
#
tantek so you start reading, then automatically skip over it
#
tantek whereas non-english speakers will immediately pattern-match on the name of their language in their own language
#
barnabywalters I’m guessing the language name is likely to be more effective than flag glyphs for that, and less visually noisy
#
barnabywalters also flags have the single-country-multiple-languages problem
#
barnabywalters I certainly need to do more work on presenting my translation link
#
tantek yeah
#
barnabywalters but at least french users are likely to see it ;)
#
tantek the flags are distracting and don't scale, but they're mostly ok
#
barnabywalters don’t scale => SVG?
#
barnabywalters or not scaling in the other sense
#
tantek didn't mean that notion of "scale" but yeah
#
tantek that too
#
tantek they're gifs
#
tantek today I'd probably redo them as data: gifs
#
tantek I got the names of the languages in their own language thing from Wikipedia
#
tantek which started this: http://microformats.org/wiki/Main_Page#Microformats_wiki_translations
#
tantek and of course this: http://indiewebcamp.com/Main_Page#Translations :)
Stevef, stereoket, bnvk, morrocco_mole, zztw1 and gardnr joined the channel