Archive of UserLand's first discussion group, started October 5, 1998.

Re: XML needs a killer app

Author:Paul Snively
Posted:6/29/1999; 12:18:10 PM
Topic:XML needs a killer app
Msg #:7933 (In response to 7911)
Prev/Next:7932 / 7934

XML needs a killer app. You won't find it on the W3C website. It won't be an editor, or a spreadsheet, or even a search engine. I don't expect competitors to really work with each other.

I think this last point is a given--it's exactly the problem that true software integration has whenever there's a system-level "lingua franca," whether it's AppleEvents or COM/OLE or CORBA: to the extent that the functions exposed to interprocess communication are some developer's family jewels, there's a disincentive to exposing them to an RPC mechanism. Exposing your data/object model through XML has the same (perceived) risks. The folks who end up winning here are the big players anyway: the Words, the Excels, the Visios... being scriptable or XML-based, for them, is just another way of ensuring tie-in/buy-in. Meanwhile, the rest of us are left to scramble for what's left of the mindshare/marketshare pie.

The killer XML app will be so compelling, such an immediate market force, that it will make competitors forget about killing each other.

I frankly doubt this; precisely because XML is a metalanguage, I expect its "killer app" to be a lot more subtle. But you allude to this yourself in the very next few sentences.

HTML was like this, but you can't just wish for it to make it happen. It has to happen all on its own.

Once again, I'm inclined to doubt this: XML's killer app will likely not arise by accident, but will be a novel implementation of XML's being poised to successfully straddle the world of data and the world of code. I actually suspect that XML's "killer app" will resemble Java in one crucial aspect: people with relevant technical background will look at it and say "that's nothing new; we've had tools with properties X, Y, and Z for thirty years!" It will be a function of bringing those properties together in the right one place at the right one time, and then--like it or not--mercilessly hyping the result in order to establish first mindshare, then marketshare. The big lesson of Java isn't technical; it's that in order to succeed you need perceived ubiquity, and in order to achieve perceived ubiquity you need marketing. If your marketing is successful, ubiquity becomes a self-fulfilling prophecy. So it will be with XML's "killer app."

Web content syndication may be the killer app.

You must be getting tired of reading me saying "I doubt it," but... :-)

"Content syndication" is a legacy media concept that doesn't translate well to the Web. It's based on the idea that there are informed, educated, high-value content developers who in one fashion or another "broadcast," whether literally through technology or indirectly through syndication, their uniquely valuable information, insights, and/or opinions to the consumer masses.

Rightly or wrongly, that's not the model upon which the Web works, and not for lack of legacy media's attempts to make it fit that mold, either (read Michael Wolff's "Burn Rate" for an at turns amusing and depressing look at a legacy media guy's efforts in this regard).

The Web is inherently more communal, democratizing, leveling than this. Freedom of the press belongs to those who own one (this needs attribution, but regretably, I lack the original reference). With the Web, anyone with a computer, a text editor, and a Tripod account ;-) is "the media," with everything that implies. Caveat Emptor.

It could be distributed apps that run equally well on all popular operating systems.

XML + Java stands to offer truly enormous gains here, but I still don't see the "killer app."

The role XML will play is an enabling role. You may find out that the killer app for XML is built in XML on a "How this is Done" page, but it won't be in your face.

Exactly, so what will it be?

I believe we have hints based on the reflective nature of XML. Specifically I believe the big hit will stem from the metadata efforts, the fact that RDF represents the evolution and distillation of decades of experience in knowledge representation, and the growing recognition that far-flung, disparate databases of information aren't enough. Thanks to efforts such as <http://www.cyc.com> and <http://www.framerd.org>, or perhaps something Bayesian Belief Net-based, databases will give way to knowledgebases, something like RDF will become the "lingua franca" of the knowledge thus represented, and Web-based software as a collective will become appreciably "smarter." Think of it as taking advantage of the existence of WordNet, but with common sense reasoning included as well.

Thoughts?

There are responses to this message:

This page was archived on 6/13/2001; 4:51:07 PM.

© Copyright 1998-2001 UserLand Software, Inc.