england-national-crime-mapping data?

I am not seriously doubting that it is easy to provide a spreadsheet in machine readable form (or should I say a more easily machine readable form?  Spreadsheets are a hell of an advance on the PDF which was sometimes the only form available).  However, there can be issues about just how it should best be done.

One of those is data revisions.  If all anyone is ever going to want is access to the most up to date estimates, there is not much of an issue.  The old data get overwritten each time.  But some users need access to the old estimates as well.  One of the key questions is often how much and why data have been revised.  If this history is lost, it becomes harder to do this.  National Statistics principles say that past (or even erroneous) statistical releases must not be removed.  So do we maintain an ever increasing number of versions of a time series in directly machine readable form, presumably with some guidance to the user?  Or do we keep the old versions alive in the less friendly form only? 

Even more specific - what about data discontinuities?  It is easy to indicate the existence of a discontinuity in a table.  Less easy, I think, in something which is designed to be picked up automatically by a computer and placed into some other application.  Sometimes, also, the new series is only available back to (say) 2003, with the old series being the only indicator of trends for (say) 1983 to 2003.  Continued access to that old series, flaged up as "old basis", may be essentail for some uses.

It is also quite common for data for a particular year for a couple of categories to be combined, in order to avoid one or both of the individual categories revealing something about an individual or a business.  If you are not careful, the external application is going to end up with totally inconsistent data.

While I am sure there are ways of dealing with these issues, it is likely to involve a bit more than telling someone to press a button which says "turn into CSV/RDF" form or whatever. 

As an aside, I was working as a government statistician when the Open Data initiative was launched (the user name is a bit of a clue).  What it felt like in the immediate aftermath was people in Cabinet Office saying "why have you not already got this stuff in RDF already?".  Most of us had probably never heard of RDF, and it seemed impossible to get any guidance that was understandable to someone who did not already understand it (think of a bad manual for a software package!).  Expectations seem to be more realistic now, with an acceptance that non-randomly named spreadsheets and/or CSV represented a reasonable start, and that the resource available was probably better devoted to other tasks.

So please let's not run away with the idea that everything is easy and it is just lazy or incompetent civil servants who can't be bothered.  The technical transformation may be easy, but some thought about whether the resultant presentation of the data would help or hinder users is not a bad thing to do first.