A week or so ago, Simon Fodden over at Slaw wrote a somewhat exasperated post about the generally poor state of URLs for accessing online statutes. At the end of the post he promised a part 2, in which he’d elaborate on other, more elegant ways to handle the problem.

The issue of usable URLs has been on our mind, as LexPublica must provide sane addresses for a large body of legal information that will span many languages, jurisdictions and types of content.

We think that people need three key things out of the URLs on LexPublica (and, likely, from any site that provides access to a large archive):

1. Convenience: it should be convenient to link to any document on LexPublica (including parts of that document, such as a particular section, clause or phrase).
2. Context: the URL of a LexPublica document should put it in a meaningful context. Take http://lexpubli.ca/english/canada/contract/confidentiality/simple/v1.0 as an example – a potential visitor can guess that they are dealing with a simple Canadian English-language confidentiality agreement.
3. Affordances: each part of a LexPublica URL should be an affordance – that is, a meaningful way to interact with LexPublica. People should be able explore LexPublica simply by making reasonable changes to a URL – as Ramsus^[1] puts it, “Content should be where people look for it.” URLs like http://lexpubli.ca/english/canada/contract/confidentiality/simple, http://lexpubli.ca/english/canada/contract/confidentiality, and http://lexpubli.ca/english/canada/contract should all return meaningful results. So should URLs like http://lexpubli.ca/contract or http://lexpubli.ca/confidentiality.

Convenience

The convenience issue is simple enough to solve – we can auto-generate a short, unique and stable URL for every document on the site. An auto-generated URL for a document would be a random-seeming jumble of 4 to 14 characters, perhaps like so:

http://lexpubli.ca/4pl8-n93f

A scheme like this would allow us to easily provide short URLs for billions^[2] of documents.

Addressing particular parts of a document could be accomplished through anchors and some simple, smart coding that can transform URLs like this:

http://lexpubli.ca/4pl8-n93f#s1.2.3:p1-p4

… into a reference to paragraphs 1 to 4 of section 1.2.3 of the given document. While this isn’t particularly human-readable, it makes for compact and convenient linking.

Context

Context is a tougher problem to solve, as it ventures into the hairy territory of legal information taxonomies. Putting a statute into its proper place in a hierarchy is one thing. Placing a more common piece of legal information, such as an employment agreement, is more difficult due to the sheer volume of employment agreements and the diversity of the information such an agreement might contain.

One solution is to recognize there are multiple relatively-simple taxonomies that a particular document can fit into.

For example (and without a great deal of thought on the exact URLs), the UK Transport Act of 1985 could be accessible via the following URLs:

One URL of the form /{jurisdiction}/{law-making body}/{legislation type}/{legislation name}/{year}. e.g.

http://lexpubli.ca/uk/parliament/ukpga/uk-transport-act/1985

Another URL of the form /{rfc4151 tag}. e.g.

http://lexpubli.ca/legislation.gov.uk,1985:ukpga/1985/67

Affordances

The last issue is making it so that people with different levels of legal and technical skill can easily wade through thousands of documents to find the right needle in the haystack.

While there are many parts to a solution for this issue, in the URL space we’ve adopted the view that documents should be where people look for them. This view implies that our URL system (and other parts of LexPublica that help organize information) must provide multiple ways of finding the same document. At the very least, we aim to cover the most common ways that people might seek to find a particular contract template or piece of legal information.

For example, if someone were to visit the URL http://lexpubli.ca/contract, the site should do its damnedest to provide them with a meaningful result. In this case, it might be an overview of what contracts are and how to find more contract information on LexPublica.

1. Rasmus Lerdorf, best known for inventing the PHP programming language and founding the online community around the language. [↩]
2. Some readers who’ve done the math may be thinking that 14 alphanumeric characters would give us a lot more than of billions of possible IDs. The number of possible IDs would be reduced by filtering dictionary words out of the IDs and by a few features to reduce input errors (such as optionally allowing hyphens every 4 characters) [↩]