Coding Law

A few weeks ago, I wrote a post reflecting on the intersection of law, code and architecture, prompted by Richard Heaton and John Sheridan speaking about the Goodlaw project.

That was an unapologetically abstract post, so here’s a concrete bit of legislative detail which encapsulates some of the wider issues. And since John is speaking at Teacamp tomorrow, which is a gathering of largely government largely geeks, there may be some interest in applying code optimisation to the law.

For reasons far too tedious to go into, I found myself leafing through the Public Service Pensions Act 2013 the other day, and stumbled across a tiny example of how an absence of architecture and coding can make the law more confusing than it need be (whether it is more confused, I am not qualified to judge).

Tucked away in schedule 7, there is a paragraph defining continuity of employment for the purpose of working out in certain circumstances whether somebody retains rights to be treated under the old (and generally more favourable) rules or falls under the new rules. The basic answer is that people continue to benefit from the old rules if “the old scheme service and the new scheme service are continuous”.

So far, so good. But what does “continuous mean”? The answer is almost certainly not what you think, since it turns out to be “not actually continuous at all”. Or, more precisely:

Continuity of employment

3 (1) For the purposes of paragraphs 1(2)(a) and 2(2)(a), there are to be disregarded—
(a) any gap in service where the person was in pensionable public service;
(b) a single gap of service where the person was not in pensionable public service, if that gap does not exceed five years;
(c) two or more gaps in service where the person was not in pensionable public service, if none of the gaps exceeds five years.
(2) In this paragraph, “pensionable public service” means service which is pensionable service in relation to—
(a) a scheme under section 1, or
(b) a new public body pension scheme.

There are two important issues worth reflecting on there.

The first is how quickly a natural language concept (“continuous”) and its use in a legal context can diverge. If you read – or get a machine to parse – the core provision in isolation, you might think you understand the effect, but you will be wrong.

The second is how the absence – or at least seeming absence – of structure or architecture leads to inefficient and apparently redundant wording. In effect, paras 3(1)(b) and 3(1)(c) are defining “continuous” to mean a period including one or more gaps of no more than five years each. To paraphrase:

• Has there been a single gap in service which is no longer than five years? If so, ignore it.
• Has there been more than one gap in service each of which is no longer than five years? If so, ignore it.

Those are two separate conditions. But they pretty obviously collapse into:

• Ignore gaps in service unless there is one which is longer than five years

I read the original wording several times, because my assumption was that there must be some subtle distinction between the treatment of a single gap and multiple gaps which I was failing to spot. But unless I am still failing to spot it, there isn’t.

It would be silly to draw strong conclusions from a tiny snippet of legislation. But I think it would be worth reflecting on what it would take to be able to express the policy intention with sufficient legal precision, but without redefining words to mean their opposite and with structures which helped understanding rather than hindered it.

I shall be fascinated to see where the Teacamp discussion gets with all of that, and what might happen if we were to recognise not just that code is law, but that law is also code.

Original post