They were kind enough to leave a comment explaining more at the bottom of that post.
Here is our reply to them:
Strictly speaking new boundaries don’t come into effect until elections are held (usually in May). That’s because the existing representatives continue to represent the same constituents until then.
This is not a reason not to publish the data – publishing it in Boundary-Line could be done early and often, just with the addition of an ‘enforcement date’ or ‘start date’ field in the data. There are many data sets that have a concept of time in them. This could also extend to an ‘end date’ field too, when we know particular boundaries are going to stop existing.
OS creates the data for new boundaries according to the decisions of the relevant authority, but they are not final until each has cleared Parliament as a Statutory Instrument (SI).
As above. However, OS is missing out an important part of the data you publish: as the owner of the ID system used to refer to these shapes there is a responsibility to ensure that these IDs can be defined and used as early as possible, even if the new shapes can’t be published for other reasons. Issuing IDs is the thing that allows people and computers to know they are talking about the same exact thing, and this is not possible in this case until the publishing authority issues a new ID.
As a side point, this is why we decided to make election IDs decentralized – we don’t want to be the authority on issuing new IDs.
As a result it’s difficult to publish a definitive set of boundaries very much in advance of an election.
Our Boundary-Line opendata product (which is created from the SIs) is updated in October and May, but we aim to have the full set of new boundaries in Election Maps by the end of March.
It would be a lot easier to publish the data if it wasn’t “published” every 6 months. Everyone who publishes data should be moving towards maintaining registers and away from “binding data” in the way we might have bound a book in the past.
We understand that the complexity of procedures can lead to confusion and frustration, and we are constantly looking for better ways of working.
Actually, the worry and confusion about licensing that OS imposes on, for example, The Boundary Commission, is the worst part. This could be fixed very quickly with the publication of a public blog post, explaining that authorities and The Boundary Commission are able to publish shapefiles at any stage of their creation under ‘presumption to’ publish’, as Jeni points out above.
To summarise, here are the improvements OS can make:
- Publish IDs of the new areas in advance. This needs to be as early as possible. Maybe draft IDs could be published alongside the draft areas the The Boundary Commission create. Without these, no one has a shared (digital) identifier for the new data that can be relied on.
- Publish the shapes with a ‘start date’ and ‘end date’ fields, as laid out above.
- Make it clear to those creating the data that they are allowed to publish it directly, without fear of legal recourse from you (by writing an easy to understand blog post about it).
- Stop publishing 6 monthly “data books” and start publishing APIs. For example;
- Offer a product like mySociety’s MapIt API. This is the best resource for geographical information in the UK.