Dave, I think I figured something out.

OPML is simple enough to write by hand in a text editor. That’s how I composed this file and this file. But not everybody wants to compose text in a text editor.

It’s much easier to compose OPML files in tools that support OPML, like this one. But not everybody wants to purchase and/or learn how to work in a desktop outliner (their loss).

Then it hit me. We have web-based OPML readers and web-based OPML validators. Why not a web-based OPML composer? The interface won’t be nearly as interactive as a desktop tool, okay maybe we would if someone does a fancy java applet or DHTML thingee, but we’ll achieve goal #1 which is gettting more people to write OPML by lowering the barrier to entry for authors. We could similarly lower the barrier to publishing by providing free hosting for all OPML files authored with the composer.

Sounds like bootstrapping to me. Does this sound crazy enough that it just might work?

I realize now that this post would have been infinitely cooler if I’d simply coded up the app and given everyone a URL. So many ideas, so little time. Fellow hackers, if you want a chance at fame and glory for relatively little effort, here’s your chance 🙂

Update Of course somebody already did it! Check out http://demo.weboutliner.com/weboutliner/. Looks very promising. Thanks to Roland Tanglao for pointing this out.

14 thoughts on “”

  1. >>Does this sound crazy enough that it just might work?

    Yes. I totally agree, and would support it.

    Basically DMOZ has an online editor that sucks.

    Wouldn’t be hard to do one that was better, and spit out OPML.

    Like

  3. What sucks about the Dmoz editor? It’s just a form to enter the data in. It’s not an outline editor, because you don’t need an outline editor: each editor can only edit the categories he is in charge of. Nice and simple.

    Dave sees everything in the form of outlines, even though it tends to be a bad way to model reality. Directories should be in relational database form, with separate tables for categories and for sites at the very minimum.

    The Dmoz data schema could use some work, though. There is no “sort” field for titles or categories, so for example in categories related to personalities you get things like “David Bowie” sorting to D, and “The David Bowie Home Page” sorting to T.

    Like

  4. Dude,
    Stay tuned – any day now – Mr. WebOutliner Marc Barrot is going to reward us with user accounts – so everyone can have their OWN WebOutliner! Then we’re gonna bake it into our social network based upon FOAF – the PeopleAggregator.com.

    🙂

    Like

  6. Why not drop the barrier even more by converting the IE favourates to opml? The converter just converts the data and posts it on a blog, all the input & editing is taken care of by the IE favourates interface.

    Like

  7. We have come up with another solution we will be which will fill a definite void that it out there. What if you built all sides in Flash, as a desktop app, cross platform and portable to any system that has a flash player, yes including PDA’s. Let’s say you built the aggregator the same way and had the abilty to create folders, import V-Cards (not sure about the usefullness of this except that the v-card generation would be in Flash making it much more difficult for web-bots), of course the other end, the editor is also built on the Flash format so it can be a desktop or Web based application. Looking forward to seeing WebOutliner sounds interesting.

    Like

  8. The DMOZ editor might suck, but the modelling’s pretty good. But like Yahoo, it does have a major flaw – hierarchical outlines aren’t a good way to represent reality or the web.

    I very much like the idea of distributed directories, but we need to get away from the tree structure, not carry over the flaw. I think it makes more sense to work *with* the web and use a graph-based model. You can still have hierarchical *views* (e.g. this is Dave’s tree) but you don’t have to put one person above another in the model.

    It also would make sense to use a framework designed specifically for the purpose of describing web resources (the Resource Description Framework) rather than a language designed for simple outlines.

    There’s no harm in creating the World Wide Outline, I just think the time could be better spent working on making the World Wide Web more useful.

    Like

  11. OK. I’m going to drop a bomb here. Most of the work is already done and it just became open source a few weeks ago. I think Viswanath Gondi’s comment is dead on – it should be easy to bootstrap distributed directories in a viral manner using favorites. You need to check out http://www.syncit.com/update.htm . Due to a database server malfunction that wiped out all their accounts, the SyncIT folks have closed their doors and open sourced the software http://sourceforge.net/projects/bookmarksync.

    The SyncIt software consists of a PHP/MySQL or ASP/MSSQL hosting website and C++ synchronizer client (BookmarkSync) that lives in the windows task bar. The bookmark format is XBEL and is similar to OPML. XBEL is a well known format with an existing software base. To link directories, you would just include a link to an XBEL URL, or an XBEL URL with optional XPATH to a sub dir. Clients would be responsible for rendering linked directories. See: http://pyxml.sourceforge.net/topics/xbel/

    I have been using BookmarkSync for about 3 yrs. now and can’t live without it. It synchronizes my bookmarks on my work, home, and laptop computers and allows me to access my bookmarks via the web if I am on another machine (kiosk, friends, etc.). Setting up a new machine and transferring bookmarks is a breeze. I was so hooked that when SyncIT.com went away, I went to the pain to set up my own server just so I could keep using it. Until now, the ability to publish to the web site was not too interesting to me, but I can see that with some slight tweaks to the website we could have exactly what Dave has been evangelizing. Most of it is already there. Check out my test server for an example http://www.pangistics.com/syncit/tree/cview.php?ref=POPULAR&pid=141c. I have made one of my folders public – it’s entitled ‘Search’. Unfortunately, it consists of things I bookmarked about 3 years ago, but it will serve as an example. This raises some questions about link rot, quality of directories, etc. but that should be resolved as Dave suggest, by page rank, or hive-mind linkage. Also, a posting or update time stamp should give a clue as to freshness, and link validators can automate the maintenance process to a large extent. Dead links can be flagged and the # of dead links + update time can give a clue as to the value of the directory. Also, dead links could be redirected to the wayback machine in perpetuity.

    I’m working on business model for a commercial service, but I don’t know yet exactly what that would be. Hosting fees?? Free hosting with revenue from Google AddSense? Write my own enhanced client that is not open source and sell it?

    Like

  12. Why DMOZ sucks. Not just the non-outliner: too many of the editors are petty. I know this from personal experience, and the experience of others.

    It should be completely open.

    Like

  13. I’ve been watching demo.webOutliner.com for a year now and anxiously waiting for something to happen. It is very cool. I’d love to see it as a Radio, Manila, and MT tool.

    Like

Comments are closed.