User Tools

Site Tools


Web Montag Silicon Valley 2006-08-14

Silicon Valley
Socialtext Headquarters, Palo Alto

Original notes written in SubEthaEdit by:


  • Tim Bonnemann, Web Montag organizer
  • Harald Rudell
  • Benay Dara-Abrams
  • James Littlejohn,
  • Mario Linge, SAP
  • Sebastian Steinhauer, SAP,
  • Wiebke Poerschke, SAP
  • Jakob Snilsberg, jajah
  • Peter Kaminski, Socialtext
  • Gernot Poetsch
  • Adina Levin, Socialtext
  • Frederik Hermann, jajah
  • Mark Wubben, sIFR, Jot intern
  • Debbie Mac, Liberty Alliance
  • Matthew O'Connor, Socialtext
  • Beth
  • Eugene Eric Kim, Blue Oxen
  • Peter Theony, Twiki
  • Eric
  • Karl
  • Brad Neuberg, Hyperscope
  • Ray
  • Vicky
  • Craig
  • Kristie Wells
  • Alec Dara-Abrams, Sony
  • Dave, Yahoo
  • Ray Seddigh
  • Lynn
  • William Jolitz,
  • Daniel Mettler, IT consultant

Mnemomap Presentation

“realtime search workflow”

Mnemomap does a live, interactive search with lots of integrated data sources – much fancier than the simple list you get back from regular search engines. It provides a “mind map” view of the results and relations to other terms. Out of this map you can create filters to be applied to Yahoo, Flickr and YouTube.

This was a cool international presentation. Simon presenting in English from Germany (at 04:00 in the morning!) with Tim here in Palo Alto operating the slides. It was a double demo; since there were a few folks from here, we used Jajah to do the call setup to Simon's phone and the speakerphone here. :-)

Mnemomap Questions

Q: Are you integrating with any browsers?
A: No plans, but we're interested.

Jajah questions

Q: conference calls?
A: yes, starting tomorrow

Q: do you record the calls?
A: no

Q: pricing?
A: most free, but if you call a non-jajah user, it's at VOIP rates
A: for example, this call with Simon and us, 13.5 minutes, cost 0.28 euros


multimedia content in XML feeds

“Concept and prototypical implementation of an XML-aggregator for multimedia content”: I’ll be giving an overview of my diploma thesis and some of concepts I developed. I’ll probably focus on the aggregation of what I call “universal content” and techniques to automatically handle information overflow.

“Subscribable Content”

Could use additional metadata to determine most important things going on, and present top 20 to users

Introducing “James”

  • “personal information butler”

Q: Is it available for download?
A: No, not yet; will run on Leopard in spring

Q: Only available for Mac users?
A: Only on Mac, depends on a lot of the built-in frameworks, infrastructure (calendar, address book, RSS aggregation), APIs

Q: How will you distribute it?
A: Currently a desktop app instead of a web app. When finished, it should work like Google News, except personalized on your own machine. Needs to be available offline, when I have a little spare time to read the topmost important things. I think the future of web2.0 is that desktop apps and web apps will share data and operation.

Q: Any business aspirations / goals?
A: Cocoa development (done)
A2: diploma (done)
A3: get a job
A4: release this and get really really rich :-)

Hyperscope, Eugene

(Eugene points out that Bill and Lynne Jolitz wrote 386BSD! :-)

Doug Engelbart's original demo at Google Video: <>. (The Mother of All Demo's)

Looking back at the original hypertext systems and capabilities provides valuable lessons for application and information design even today.

Doug's first system: Augment

Eugene's goal: not to make Augment ubiquitous, but to open conversations and spark ideas.

The way to get Doug angry: call him a usability guy. Software for experts needs to be *expert-friendly*, not user-friendly.

Idea: humans are smarter than we give ourselves credit for.


  • dictionaries, how they're organized
  • books: physical structure constrains exploration
  • knowing a little bit about the structure of the document allows me to navigate it quickly
  • another example: bible, john 3:16

Resource: Christopher Alexander's _A Pattern Language_

  • adding italicized summaries, lets readers zoom out and skim high-level

“knowledge structures”

“knowledge containers”

software tools/techniques for information navigation

  • outliners
  • zooming in and out of nodes

There should be a single tool that abstracts the common structure of all documents/knowledge and lets you navigate it.

“Latency matters.” How many people write paper letters vs. emails?

Granular addressability, scrolling to “S” in 2 seconds, vs. ~100 ms to get right there.


  • constrain design to replicate original experience of hypertext
  • immerse ourselves in the original design, then innovate later
  • first two months we just dog-fooded the original tool, we were all converted
  • our hyperscope runs in browsers, and replicates the original experience
  • “viewspec”, an address
  • high-level, granular addressibility
  • transclusion: including the hyperlinked resource inside the hyperlink
    • an <img> which shows the resource behind the src attribute in the HTML page

Hyperscope release party, September 5th – be there or be square!


  • viewspecs
  • purple numbers

Hyperscope team needs:

  • web design
  • logo
  • small image of a dancing hula girl (maybe via Peter Theony???)
  • file transformers
  • see the “ways to contribute” section on the web site

Q: who's “we”
A: Doug, funding from NSF, Eugene, Christina Engelbart, Brad, Craig, Jonathan Cheyer
A: we're the kick-off team, but we want people to join the community and work on the code and the system and the conversations

Q: documents have to be in a special syntax?
A: yes, OPML

Q: the reason the bible chapter/verse thing works is because the document is static. how does this paradigm work as the document gets edited?
A: in the original augment system, there are 3 kinds of addresses, including hierarchical identifiers. hierarchical identifiers address by position, so they refer to different information if the document changes you also need immutable and persistent addresses. in the original augment, those were node identifiers (like purple numbers).

Comment (Brad Neuberg): the keystrokes seems esoteric at first, but they quickly become natural, and now i feel frustrated when working with regular documents

Q: are documents meant to be shared and passed around?
A: yes

Q: granularity: can you address finer than a paragraph
A: in original Augment, yes; in first release of Hyperscope, no

Q: generating identifiers for locations?
A: easy in-document, pretty easy on one web site, unsolved for unique identifiers on the whole web

Q: timeline?
A: 1st release is core capabilities, want to have community discussion on

Comment (Eugene): the original system still runs, on PDP-10 emulator on Linux on Sun; if any of you want access, ask one of us

Plazes, Mark Wubben

  • current web version
  • upcoming: phone version
  • suggestion: rfid: stick a rfid transponder on a wall, and use the rfid reader in the phone to sync to the plaze

Q: range?
A: ~ 4 inches

Q: use phone with GPS?
A: same as reason plazes says, not everyone has a gps

Q: do you have a kit?
A: this model is from

Q: cost?
A: kit, $80

Q: in contact with Felix and Stefan?
A: yes, they're friends, but they haven't seen it because i just put this together yesterday :-)

Q: why not barcodes? e.g.,
A: yes, cool too

Q: embed the rfid reader in the wireless card of a laptop? bluetooth?

Q: the question is, how do you construct fusion mechanisms, e.g., placelabs

Comment: another kind of graffiti!

Other Cool Stuff

Gernot's presentation reminded Pete of another multi-source merged info stream, the Alvin Frame Grabber system from WHOI:

Top Ten German Web Apps:

Angela Merkel Video Blog: [via Google Blogoscoped]

location/silicon_valley/notes_2006-08-14.txt · Last modified: 2022/08/18 15:26 by