The following are notes from a session by Douglas Bowman, the first speaker at Webstock 2006. My note-taking rambles somewhat, but as Doug finished speaking I was left wondering "when is a standard not a standard?" Doug presented an interesting idea that raises more questions than it answers (e.g. Should these common practice class names/ids be translated between languages? Should sites offer a patterns discovery service?):

Doug and others (e.g. John Allsopp) have analysed the markup of a number of web sites. In some cases they observed a tendency for interface developers to use consistent identifiers and class names in similar page regions (“header”, “footer” etc.). In other cases there are marked disparities.

Doug suggests there may be value in a distillation of best practices; the derivation of a common approach to markup; a semantic web with a lower case “s”. As demonstrated by CSS Zen Garden, a regularised approach to markup need not constrain the designer. The approach could lead to more efficient deployments and the enabling of user created stylesheets. He sees benefits for:

1. Developers: consistency, interoperability, time-saving
2. Vendors: reliable targets, possibilities for customisation and autogeneration
3. Users: skinnable applications, user stylesheets

Borrowing from Christopher Alexander's work in A Pattern Language, and from Java patterns, Doug proposes the identification of various commonly used patterns:

1. Site patterns: blog, wiki, academic site, store, application, etc.
2. Page patterns: contact, news, map, results, etc.
3. Module patterns: profile, entries, archive list, etc.
4. Interaction patterns: drag + drop, expand/collapse, inline edit, etc.

Doug identifies a gap between page and module patterns. Doug recommended the O'Reilly book Designing Interfaces by Jenifer Tidwell, and noted the resources available on webpatterns.org, lukew.com, SixApart, and the Yahoo! Developer Network (“action patterns”).

Doug stressed that this common approach must be neither prescriptive, nor limiting. He has no intention of producing a standard for use in validating documents. He visualised the distillation process as a de-centralised collaboration: asking questions, seeking answers, documenting, sharing and learning.To avoid starting too early down the wrong track, Doug defined four principles:

1. Start simple
2. Identify common patterns
3. Re-use popular taxonomy
4. Enable global participation

[Doug went on to hint at the possibilities of microformats such as hcard, hcalendar and XOXO, noting the value of these coupled with a Firefox plugin such as Tails]