Contents 1 Itches & Scratches 1.1 successful projects 1.2 consolidating profiles 1.3 easy way to publish personal profile 1.4 sharecropping and site death 1.5 user interface frustrations 1.6 URL shortener service problems 1.7 URL shortener fragility 1.8 non-printsafe short URLs 1.9 writing code over and over 1.10 CASSIS proof of concept 1.11 OpenID too hard to consume 1.12 Twitter bottleneck 1.13 diso 1.14 summary 1.15 thanks

Itches & Scratches

2010-199 Federated Social Web Summit talk by Tantek

• many building blocks - towards a Federated Social Web
• hopefully you might be able to (re)use some of them

successful projects

• most successful projects
• are people scratching their own itches

consolidating profiles

itch: no easy way of consolidating profiles

scratch: XFN rel-me (since 2003)

• http://gmpg.org/xfn/and/
• http://code.google.com/apis/socialgraph/

easy way to publish personal profile

itch: no easy way to publish my preferred profile

scratch: hCard (2004)

• http://microformats.org/wiki/hcard-authoring
• broke thru 2 billion recently (blog post)

sharecropping and site death

fast-forward to 2008-2010

itch: tired of sharecropping and site death, untrustworthy content hosting

• 2008-03 CNET killed Consumating.com
• 2009-02 SixApart killed Pownce.com
• 2009-12 Yahoo killed GeoCities.com
  • 2009-10-26 Tears in the rain blog post by Jeremy Keith on Geocities's death.
• 2009 Google shutdown Dodgeball.com
• 2010-05-14 Google killed Etherpad.com content and URLs
• 2010-08-04 Google announced end of life for wave.google.com
  • less than 3 months before, Etherpad users were encouraged to transition to Wave.
• 2010-09-30 Six Apart shutting down Vox on September 30, 2010 (going read-only on 2010-09-15)
  • irony: Six Apart encouraged Pownce users to switch to Vox
    • 2010-09-03: pointed out by Heidi Cool:

• person-specific sharecrop vulnerabilities:
  • 2008 - BoingBoing "unpublished" all articles by Violet Blue (>100 posts) (see: William Gibson Completely Deleted from BoingBoing Archives)

and

• Twitter fail frustrations

scratch: DIY indie web, start with my own site

• use my site as my identity - XFN+hCard+OpenID pretty much solved this
• post tweets/notes on my own site
  • syndicate out to silos - that's where my friends listen/post

http://tantek.com/

• server-based Twitter client
• called it Falcon http://tantek.pbworks.com/Falcon
• the fastest bird, when trained, can drop things off, pick things up for you

user interface frustrations

itch: crappy Twitter permalink UI, etc.

http://twitter.com/t, http://twitter.com/t/status/18803073634

• inconsistent typography (sans vs. serif ?!?)
• navigation deadend
• no URL embeds
• poor auto-linking (e.g. "federatedsocialweb.net" )
• auto-shortening when not needed!

scratch: improved UI on my site http://ttk.me/t46n3

• Helvetica - at least it's consistent
• next/prev buttons/shortcuts (arrow keys, j/k)
• easy URLs for copy/paste linking
• better auto-linking (handle URLs without http)
• better ellipsing

URL shortener service problems

itches:

• tr.im being down, shutting down, going away
• bit.ly being blocked

scratch: my own shortener domain

http://ttk.me/

everyone should have one (in addition to their own domain of course).

go get one.

URL shortener fragility

itch: url shorteners are fragile (Joshua post), databases are fragile (database dies, IDs die)

scratch: algorithmic URL shortening

• Whistle
  • what you use to call a trained Falcon
  • introducing "algorithmic permashortlinks"

non-printsafe short URLs

itch: non-printsafe short URL encodings

• http://search.twitter.com/search?q=bit.ly - note use of "O", "l" etc.
• print evidence: http://tantek.pbworks.com/ShortURLPrintExample

scratch: print-safe NewBase60

• http://tantek.pbworks.com/NewBase60

writing code over and over

itch: tired of writing same logic front/back-end

• http://bulletproofajax.com/

scratch: CASSIS language subset/framework

• runs native in Javascript and PHP
• http://tantek.pbworks.com/CassisProject
• http://tantek.com/cassis.js
• open source

CASSIS proof of concept

itch: amazon shortlinks too long

scratch: ASIN.cc

• proof of concept of CASSIS, NewBase60, algorithmic shortening

OpenID too hard to consume

itch: OpenID PHP libraries too big/complex

• http://openidenabled.com/php-openid/ or http://www.janrain.com/openid-enabled

scratch: RelMeAuth - generalize form of TwitterAuth model

• http://microformats.org/wiki/RelMeAuth
• Python open source proof of concept
  • http://github.com/ptarjan/relmeauth
• PHP proof of concept
  • http://github.com/themattharris/relmeauth
  • live prototype: http://tantek.com/relmeauth/ (was live demo: http://apps.themattharris.com/relmeauth/ )

Twitter bottleneck

itch: twitter being down - bottleneck to update other sites

scratch: implement PubSubHubbub support (easy) + Atom (bulky)

• goal: unbottleneck myself from Twitter
• e.g. updates my Buzz directly

http://www.google.com/profiles/tantek

discovered:

• Buzz not really support Atom with XHTML (avoiding double-escaping)
• Buzz special treats Twitter - can't emulate yourself (yet)

Maybe Activity Streams can solve this.

diso

itch: DiSo lack of apparent progress, don't want to install/maintain/update WordPress, database backup tax

complexity favors oligarchies, rise of the oligarchies

scratch: reframe DiSo2

• indie web first, oligarchies second

conceptualizing DiSo 2.0:

1. personal site+shortlink domains.
2. algorithmic URL shortener.
3. hAtom store (no DBA tax - easier to maintain/backup than MySQL).
4. hAtom activity stream (DRY).
5. update notifications to PubSubHubbub (PuSH).
6. your own PuSH hub.
7. content-type-specific syndication to specific sites (e.g. text notes to Twitter, geo checkins to Foursquare, photos to Flickr, events to Upcoming, etc.).
8. reverse syndication of comments+tags+notes from said specific sites.
9. edit/comment authorization via RelMeAuth (OAuth + rel-me).
10. personal OAuth endpoint.

summary

steps you can take today:

1. use your own site as your identity
2. publish on your own site: HTML first (XFN, hCard, hAtom), XML next (Atom, ActivityStreams), API last
3. syndicate with PubSubHubbub

Most importantly, build things that scratch your own itches first.

thanks