2020 - Week 19

The week kicked off with the very best of intentions, as weeks often do. Robert and Michael planned to take the latest data from their friends in Oxford, reshape it slightly and squirt into the Standing Orders application they’ve cobbled together. And Librarian Jayne and Michael were fully committed to drawing up a straw man procedure map for census related statutory instruments, in a bid to have a bit of paper to check off when the SI finally lands into Parliament. This would have been by necessity a very crude straw man, because the procedure is some weird hybridisation of negative and affirmative. It happens so infrequently no one they’ve yet spoken to is prepared to make any promises as to how it’s likely to work or even how it worked last time. As it turned out, both of these plans came to naught. If there’s one thing they need to learn, it’s when you’re in the midst of a complex evolving situation, the first thing you stop doing is planning. They’re sure Dave would agree here. And Mike.

That said, some progress was made this week. On Monday, Librarian Jayne, Anya, Robert and Michael donned their headsets and worked through Mr Evans’ email on how to count sitting days when a House ends up sitting for two days without interruption. It turned out their understanding of sitting days as described in the Statutory Instruments Act 1946 had been somewhat naive and their edge case worries were misplaced. Although the same cannot be said for sitting days described by CRAG. Typical. And, thanks Paul.

The same conversation saw our correspondents wrap their heads around Journal Office Jane’s Word document describing how she calculates scrutiny periods for assorted types of instrument. Via the medium of shared screens they turned what Jane had written into some pseudo-code, which Robert and Michael expanded on as they continue to build out their procedural egg-timer application. As repeated ad nauseam and then some, five of the six procedures they currently support feature a clock of some kind. Often several. The rules around the clocks are set out in statute. They are somewhat deterministic and although definitely not complex they can certainly be described as complicated. End dates for procedures are calculated by JO Jane and distributed to assorted offices around Parliament. Librarian Jayne takes these dates and enters them via the procedure editor application into the data platform. From there they are grabbed by a SPARQL query, and presented on both the SI and treaty tracker websites. As the end date calculations take account of sitting days in some fashion and since sitting days are liable to change - especially in the midst of a plague - JO Jane must frequently recalculate and Brarian Jayne must frequently re-enter the dates. This is a laborious and tedious task, so they’ve set about trying to decant the rules from Jane’s head and turn them into prototype code. The data is sourced from a pair of spreadsheets of Parliament periods and sessions and from a set of Google calendars storing sitting and non-sitting days for both Houses. Michael got the Google calendar sync mostly working last week and fully intended to move on to prototyping the actual calculations, but the mostly working bit proved to be an itch that had to be scratched. So scratch he did, becoming completely distracted by fixing all known niggles with the syncing script. Which now works for every use case he has so far managed to think of. Itch scratched, Robert and Michael wrote up some notes for how the calendars should be populated, then ploughed on with coding up the calculations. Jayne added more rows to the Parliament period spreadsheet and filled in session dates back to 1979. So far, Robert and Michael have managed to wrangle together one calculation - the Proposed Negative Statutory Instrument scrutiny period. This clock kicks off on the first joint sitting day following laying and continues until both Houses have each ticked off ten sitting days. Or whichever House gets to ten last. Their egg timer application now has a working form and everything seems to be coming together rather nicely. The only caveat being, the date the calculation returns is quite wrong. Because it is based on sitting dates that Michael made up. Next week they hope to get some accurate sitting day data and code up more calculations. Although they have learnt not to plan.

Tuesday opened with an email from Rad, letting us know that our bid for funding for further collaboration had been received warmly. But given current circumstances, some tweaks were required to ensure plague compliance. A new version has been agreed and we keep fingers crossed that i’s will be dotted and t’s crossed. And that one day, we’ll get to go to Oxford again. Please.

Comment editing continued throughout the week, with Robert and Michael sharpening their chisels on the election model. This is still a work in progress but a work in progress that progresses every day. Slow and steady but always stepping forward, the sounds of their tiny chisels can be heard echoing from Battersea to semi-detached suburban Surrey.

Thursday night saw the now habitual red wine and Zoom session between Anya, Silver and Michael. Anya has had a long term goal to make the UK Parliament controlled vocabulary useful as a tool for web browsing. But the taxonomy is itself quite granular and the way it’s used to index content is also quite granular. So Anya’s keen to see the taxonomy rolled-up into broader concepts in a fashion that’s more nuanced than simple transitivity. Over wine and pixels last week, Silver suggested we might use the way the BBC indexes news articles as a route to identifying wider themes and events. So this week he’s taken the taxonomy and a sample set of indexings of Hansard and trained a machine to do entity extraction against BBC news articles. By seeing how the subject terms used by the House of Commons Library cluster around articles and how the BBC chooses to index the same articles with their own taxonomy, he hopes to generate a fingerprint of Library terms that map across to a wider BBC concept. The set of articles he used for testing date from around the time the job of managing the Wales football team was up for grabs, and the machines are showing a worrying obsession with John Hartson. Still, other parts are promising and we await next week’s wine and pixels with hope not yet abandoned.