So, I guess there was a bit of an extended kerfluffle over the weekend. And as these things usually do, it started out from a very cool idea: You can build a web directory as an outline of links, with some nodes "farmed out" to outlines hosted elsewhere—all through the magic of transclusion.
For example, say you want to host a directory of resources on web development. But, you might like to let me maintain the section on web syndication feeds. Well, I can toss you a URL to my outline of RSS and Atom links, and you can just pull it into a branch of your "superset" outline—kind of like an RSS subscription, really. Whenever I change my outline, yours will automatically get updated with my work.
Now, imagine this going off into infinity in both directions: Outlines including outlines including outlines. Outlines included by outlines included by outlines. It's a world-wide outline.
So, anyway, TechCrunch's brainstorm lead to Dave Winer's kudos and Scoble's OPML evangelism and an implementation challenge. But, to this, James Robertson responded by calling OPML a "really, really crappy format".
From this ensued a splattering of posts in various places chiming in on both sides:
- OPML is a sucky and under-specified format, with implementations subject to approval by one guy.
- OPML is a working format already in use by lots of code, so offer something better or shut up.
Now, Shelley Powers of Burningbird says that the "put up or shut up" attitude is wrong, that it's "bad technology". And, though I do agree with her, the problem is that the usual suspects involved in these sorts of kerfluffles fall on two sides:
- We want to see good, well-specified agreements before we code something useful.
- We want working, useful code that we'll agree is good when we see it.
While group #1 is willing to talk/shout things out and reach consensus ahead of time, group #2 wants to forge ahead with machines in motion and reach consensus through popular implementation. So, members of group #2 will never take group #1 seriously until they've "put up", because that's their process.
Personally, I sympathize with the dirty ways of group #2. But, I've become convinced that what group #1 does is best over the long term, as some of the early successes of group #2 may become tottering unbalanced stacks of plates later on.
So... What to do? Bah, I don't know. But, against my better judgement, I feel like I have an idea or two to "put up".
In the meantime: Members of group #1, stop arguing with members of group #2—you're not speaking in quite the same languages, and you're not going to convert anyone. Just nod & smile, walk away and come up with a better idea, come back and show why it's better.
(Oh yeah, and whatever happened to OML replacing OPML?)