There was a lot of talk about corporate wikis, measuring corporate wiki participation, and some wikis “competing” for use in the corporate area.
Oddmuse is definitely not a “corporate wiki” in this respect. I think the problem is the process of decision making: Someone starts with a list of features and goes to collect more items for that list by talking to other “stakeholders” in the company. This is what you end up with:
Yeah, I’m being mean.
Damn, I forgot WYSIWYG, blog integration, CMS, mailing list integration, bug tracking system integration, calendar integration, … there’s so much out there!
I arrived half an hour late because I was looking for my wallet. The first keynote of the day is sort-of interesting. He says many obviously true things, so the interesting nuggets per minute count isn’t too great, which is why I pulled out my laptop to write this.
The speaker is Stewart Nickolas. He’s talking about web widgets. He wants to enable widespread mashups on the widget level – secure communication between widgets on a single page, coexistence of unrelated widgets on a page. But he has a demo! That’s pretty cool. Widgets “broadcast” events such as their selection.
One interesting site he shows is Programmable Web: “keep-up with what’s new and interesting with mashups, Web 2.0 APIs, and the new Web as Platform.”
An interesting API is OpenSearch which allows you to publish how your site is to be searched, apparently. Other people can read that specification and build a search form for your site, for example.
Random website mentioned at the end: Newstin. I wonder what this was about.
Later, not much happened. I started playing Go on TedErnst’s board: He had a full 19 × 19 board, he played with some people on a 9 × 9 section and I played with some people on a different 9 × 9 section. The funny thing is that Ted himself learnt how to play Go at WikiSym 2006! Hopefully we’ll have spread the meme and even more people will be playing Go at the next meeting.
After lunch it was time to go. I faintly remembered the last day not being too great at other conferences, so I was chagrined to learn that there was an interesting program in the afternoon and an actual walk (!) scheduled for the morning after the last day. Next time I’ll contact other people earlier in order to sort out those issues.
I took the metro back to the airport together with Andreas Gohr. I was very tired having averaged 6h of sleep every night.
My plane is 40 min late, but the airport offers “free access” via wireless. Great, I think to myself. It turns out that you get free access to local content only and need to pick an ISP to reach the outside world. Sucks. But then I discover that one of these providers is the same provider that they used in my hotel. I stayed at the Pestana Porto Hotel which I liked – excellent breakfast, excellent location. Internet connectivity was bad, probably because my room was on the top floor. But being able to keep using my account is great, of course.
Looking back I think the coolest discussion I had was talking with Andreas Gohr and Christoph Sauer on the various namespace implementations we have, including an excursion into access control lists. The coolest moment I had was other people being interested in the running code I had to show for multilingual wikis.
George P. Landau
The morning started with a speech by George P. Landau on the development of the Victorian Web – a hypertext that started way before the Web. His basic premise: “the World Wide Web is essentially Hypertext 0.5.” I like it.
Stuff he mentioned about older hypertext systems:
- One to many links – he really likes it
- Previews of the target
- Marking (highlighting) the relevant passage of a link target
- Incoming and outgoing links
- Concept map integration (eg. a graphic site-map)
- Overlapping link anchors
- Multi-window environments that permit both spatial hypertext and collage writing
- Read-write medium (add links & texts) – apparently this is the only thing that wikis reintroduced!
- Multiple versions of hypertext – personal and shareable; essentially links are stored separately from the text and can therefore differ between versions.
- Backlinks – he really likes that, too
- Backlinks also apply to named anchors, so it would show you a list of documents linking to your current document, and it would highlight the documents that had links to the particular region you’re currently at.
- Eastgate's Storyspace – also makers of TinderBox (the conference chair, Mark Bernstein is “Eastgate’s chief scientist”
- Southampton's Microcosm
He also went into the question of why the web turned into a killer application. He started with benefits compared to Gopher, continued on to embedding images, and mentioned some wiki benefits. These include wikis being a read-write form hypertext, wikis enabling intended and uninteded collaboration, enabling multiple points of view, offer expandable or extendable documents, allows for different hypertext genres, simultaneous emphasis on multiple approaches and unresolveable problems. Apparently a point of view not shared by Wikipedia.
The Victorian Web is actually not a wiki. It’s a website where the author encourages contributions via email. He doesn’t believe that the delay between submission and publication is a problem. Collaboration and discussion turn out to be submissions of separate but linked documents. This gives his site a static and stable appearance. Documents just… stay. A totally different contribution and discussion culture compared to what has developed on wikis.
- Architectural Component of Hypertext Systems – This is chapter 5 from Jakob Nielsen’s book Multimedia and Hypertext: The Internet and Beyond, Morgan Kaufmann Publishers, 1995.
- When a Wiki is not a Wiki – introduction to the keynote
- The Lively Kernel: A Wiki of Active Objects – introduction to the keynote.
- SVG editing was already on my mind in 2006 (2006-08-31 Oddmuse and SVG Files) and the problem is still not solved. I’ll have to look into this Lively Kernel thing!
Too bad I didn’t find any really interesting sessions today. There was a session on the state of WikiCreole, companies hiring wiki gardeners and reportings on their excellent results. There was a session on distributed wikis collecting pitfalls but not finding any solutions.
I also spent some time sitting in the sun, doing nothing. I didn’t feel like listening to paper presentations because I usually feel that the authors aren’t good presentors.
There was an excellent bus trip through town, a boat trip up and down the river, and dinner at a restaurant. That really gave all of us another opportunity to talk to new faces, and so I ended up talking a lot with Christoph Sauer & Andreas Gohr (JPS Wiki, Doku Wiki, Wiki Matrix), Alain Désilet (multilingual wikis, WikiSym organisation in general), and others.
Too bad I apparently missed some excellent paper presentations.
The first part of the day I spent in the Babel Wiki 08 workshop. We listened to lightning talks, wrote ideas and question on pieces of colorful paper and organized them. And then we started our first breakout session and I joined the “free dictionaries” group.
We looked at the following sites:
- Le grand dictionnaire terminologique – multiple translations of terms per domain (two levels deep)
- Omega Wiki – words and phrases are grouped around concepts
- LEO – Alain thought that translators would love it; he also liked the integration of a forum for new terms and phrases
- BEOLINGUS – looked a bit visually confusing
- Wortschatz Portal – based on web crawling
- IATE – multiple target languages including less popular European languages
- United Nations Multilingual Terminology Database
- Joint Inter-Agency Meeting on Computer-Assisted Translation and Terminology – similar to the Oddmuse:Translation Extension it allows you to specify a source page and revision (automatically instead of manually) and will notify readers if the translation is out of sync.
We talked about various issues. One of the participants worked for Omega Wiki (which is based on MediaWiki) and pointed out the issues he thought were important:
- user interfaces (free software in order to improve on them)
- free software
- data structures (eg. SKOS RDF)
- mining, ie. bootstrapping the dictionary based on other sources (eg. Wikipedia, Wiktionary)
Stuff related to my own code:
- Oddmuse language guessing and filtering of search & RecentChanges by language preferences
- Translation support using the Oddmuse:Translation Extension. I should improve that and improve the user interface. Here’s an idea: For any page A, show a button “Translate” which asks you name a page B, if it exists and is already tagged as a translation of A, update the revision number. If it exists is not tagged as a translation of A, warn the user and ask whether to insert the translation tag anyway. If the page does not exist, create it with the translation tag.
- I should make it easier to create and tag translations of pages in general. Assuming I still want to encourage Emacs Wiki to turn into a multilingual site.
Most of the day I spent at the Babel Wiki workshop.
I also spent dinner with eleven other people in the old town of Porto. (Porto Flickr Group) Porto’s old town is very old and signs of decay are everywhere. Having just finished the book about Inés de Suárez by Isabel Allende I see the signs of an old sea-faring colonial power everywhere.
I also had a chat with Mark Bernstein and others over a wide range of topics. Is Wikipedia governable? What papers to read in a graduate course on wikis? Andrea Forte asks back: “Just wikis?” Is there a point in developing alternatives to MediaWiki? At work everybody seems to know and want it to the exclusion of everything else. Mark was clearly opposed to that idea. The importance of non-web solutions for some hypertext applications. For example a spatial wiki where documents are represented by little colored notes that are dragged around the screen; in a collaborative environment you’d like that movement to be visible to other users in real-time. That’s hard to do using HTTP. Using Eclipse (or an application running inside a browser?) as a front-end to a wiki infrastructure to edit code. You don’t need to download the entire code base, you don’t need the development tools on our local machine, all the building happens on the remote machine.
I was also in an OpenSpace session on Wiki Security, but I didn’t learn anything new. Users still like audit trails and permissions tied to identity. I’m still not sure I need it. I pushed my own agenda arguing against Big Brother solutions and explained my spam fighting strategy: The idea is to always spend less time on it than the spammer. Always. If spammers drag you into an arms race, you just invest the minimal amount of energy to solve the problem. Avoid burnout at all cost.
I’ll be in Porto for the next few days. Say hi if you see me!