---
title: "Simplicity and extensibility"
date: 2003-07-21
author: "Richard MacManus"
categories:
  - name: "ReadWriteWeb"
    url: "/category/readwriteweb.md"
tags:
  - name: "2003"
    url: "/tag/2003.md"
---

# Simplicity and extensibility

[Tim O’Reilly](https://web.archive.org/web/20040310153405/http://tim.oreilly.com/) writes in [Dan Gillmor’s comments](https://web.archive.org/web/20040310153405/http://weblog.siliconvalley.com/column/dangillmor/archives/001195.shtml#comments): “Simplicity and extensibility should not be orthogonal. And any technology that sets them up as opposed, instead of complements, has clearly done something wrong.”

Note: [orthogonal](https://web.archive.org/web/20040310153405/http://dictionary.reference.com/search?q=orthogonal) means “independent or well separated”.

Tim O’Reilly is talking about RSS2.0 (simple) and RSS1.0 (extensible). Lately I’ve been thinking and reading about weblog topics. There seems to be the same issues of simplicity vs extensibility in this space too, although nowhere near as much mud-flinging.

XTM stands for [XML Topic Maps](https://web.archive.org/web/20040310153405/http://www.topicmaps.org/xtm/1.0/). For a general introduction, check out the [Cover Pages](https://web.archive.org/web/20040310153405/http://xml.coverpages.org/topicMaps.html):

> “A topic map is a kind of index or information overlay which can be constructed separate from a set of resources, identifying instances of subjects and relationships within the set of resources.”

The key things to note are that topic maps are *separate* from the actual content and they are used to organise content into topics or categories. Although XTM was created only in 2001, topic mapping dates back to 1993 and has its roots in SGML. Right there is a giveaway that this spec is a complex beast. SGML is like the queen ant of XML (to borrow [Scoble’s ant metaphor](https://web.archive.org/web/20040310153405/http://radio.weblogs.com/0001011/2003/07/17.html#a3814)) and it has given birth to [many XML ants](https://web.archive.org/web/20040310153405/http://www.wdvl.com/Authoring/Languages/XML/XMLFamily/BigPicture/bigpix20a.html).

The [XTM spec](https://web.archive.org/web/20040310153405/http://www.topicmaps.org/xtm/1.0/) is a bulky insect, weighing in at 100 pages long. But being heavy gives it the advantage of extensibility. Using XTM, you can define not only topics but also associations, occurances, characteristics, hierarchies, mergers – the list goes on.

XTM even has a fancy term for creating a topic: **reification**. The spec defines this as:

> “The act of creating a topic. When anything is reified it becomes the subject of the topic thus created; to reify something is therefore to create a topic of which that thing is the subject.”

Riiiiiight. Now I understand why they used Shakespeare as an example topic in the spec 🙂 But it also illustrates that XTM has a lot of scope and you can define topics for Africa.

Compare this to the [ENT specification](https://web.archive.org/web/20040310153405/http://matt.blogs.it/specs/ENT/1.0/). ENT stands for Easy News Topics and it was built as an add-on to RSS2.0. The authors, [Matt Mower](https://web.archive.org/web/20040310153405/http://matt.blogs.it/) and [Paulo Valdemarin](https://web.archive.org/web/20040310153405/http://paolo.evectors.it/), make a point of emphasizing the *simplicity* of ENT:

> “ENT is intended to be a very simple standard for describing how topic information can be introduced into an RSS2.0 news feed.”

ENT is a lightweight ant, weighing in at only 8 pages. It has only two main concepts: the “topic” and the “cloud”, which is like a map of topics. ENT necessarily doesn’t have the same extensibility, or breadth of functionality, that XTM has. But, here’s the kicker. ENT can **reference** XTM. ENT topics can be linked to an XTM topic map (as well as RDF), via a URI within a cloud. Whoa, lotta acronyms in that last sentence. But the point is, using ENT along with XTM means you get **both** simplicity and extensibility.

And all this [can be done in RSS2.0](https://web.archive.org/web/20040310153405/http://blogs.law.harvard.edu/tech/2003/07/19#a92), and no doubt in RSS1.0 and [Atom](https://web.archive.org/web/20040310153405/http://www.intertwingly.net/wiki/pie/FrontPage) too. Tim O’Reilly is right, simplicity and extensibility don’t have to be orthogonal. You **can** have your cake and eat it too. That is, as long as the ants don’t eat it first 😉

*Originally published on ReadWriteWeb ([archived copy](https://web.archive.org/web/20020204040018/http://www.readwriteweb.com/2003/07/21.html#a83))*