Well then. That’s the first week of Lockdown 2 out of the way. It’s been dark, it’s been cold and Michael’s hot water has packed up. Leaving him even more miserable than usual. In that sardonic, northern fashion we’ve all been forced to grow used to - if not love. Nevertheless, progress has been made. Serious progress.
Silver and his colleagues at Data Language continue to investigate what went wrong with our orchestration of Member data from the Members’ Names Information System (MNIS) to the data platform. If you tuned in last week, you’ll know we suspect that some degree of manual intervention was involved and that stopped working when some of our (wom/m)anly intervenors left the building. There is also a suspicion that the intervention involved massaging the source data for the House of Lords into a shape that matched our House membership model. And that the massaging was intended to place noble bottoms in the correct seats over time.
Because MNIS assigns both House incumbencies and constituency incumbencies to Commons Members, it’s possible to inflate the model to have different people sit in the same ‘seat’ over time. The seats only change as and when constituency boundaries change. But the data for the Lords only has House incumbencies and the rules around membership are somewhat more complicated. If we’re to cope with historical bottoms, we need to deal with life peers, hereditary peers, excepted hereditary peers, Law Lords, named bishops and hierarchical bishops. Life peers are easy to deal with. The seat is created alongside the peerage and ceases to exist upon either death or retirement. Whichever comes first. But we have a real problem with named bishops and hereditary peers, where we’d like to say that different people sitting by the same ‘virtue’ occupy the same seat.
Luckily we have something of a start in the database Robert and Michael purloined from David. To make that work they need to pick apart person details from peerage holding details from peerage details. And get a better understanding of which properties live where, and how the chain of inheritance through remainders and special remainders works. So on Thursday, Anya, Robert, Ben and Michael fired up Zoom for a chat with David. Silver was supposed to join but events intervened when one of his children, or possibly his dog, left a tap on and flooded his house. Each of our many peerage related questions were answered and a new candidate schema emerged. Robert and Michael have work to do to migrate David’s database to match the candidate model and more work to do to get this model shipshape for the data platform. Thanks as ever for your time David.
Early on Monday morning, as dawn broke, our Jianhan made a start on converting our procedure model - currently based on route types - to something more logical. By the end of the week he’d done most of the preliminary work necessary to equip the procedure editor to create, update and delete procedure step types - business, logic gate and decision - and associate them with steps. Feedback from Anya, Jayne and Michael led to a couple of interface tweaks which Jianhan is digging into as we type. His super rapid progress can be seen on yet another Trello board. Top work Jianhan.
It all goes to show that, sometimes, there’s no substitute for a good plan. But only sometimes. There are other occasions when your journey sees you arrive at a river, there is no bridge in sight and the only way to cross is to feel the stones. Such an occasion occurred this week, when Michael awoke in a cold sweat at the realisation that the logical interpretation of a NOT is negation and not preclusion. And that a step having not been actualised would emit a FALSE, which the NOT gate would turn into a TRUE, which would bring the target step into play. Not something anyone would desire, as we’re sure our dear reader will agree. He was at least able to wash off the cold sweat with a cold shower before gathering the gang, unshodding them and forcing them to paddle through pixels.
Which is how he came to spend Friday morning with Anya, Jayne, Jianhan and Robert, talking and drawing slowly through the problem. He concluded with a nonchalant, ‘well, how do we solve that then?’. At which point Jianhan solved it. Eight minutes into an hour long meeting. Now we assume an unactualised step does not emit a FALSE, but a NULL. If indeed, a NULL can be emitted. Robert explained that we’d moved from binary to ternary logic, if only to impress upon the assembled the advantages of his formal - and paid for - education. To emerge from a cold shower with only coffee to heat the bones, and find yourself faced with a shift from binary to ternary logic, is no way to start the day. Robert also made some noises around whether the maps were intended to be based on language models with heuristic interpretation via grammar existing externally. Or whether they were pure ones and zeros. And nulls. With no externalities. We think we’ve settled on the latter, although we need to experiment. Thought experiments also need lab time. The conversation continued and both librarians and computational experts seemed to agree that - should pure logic be the goal - there should only ever be one route into each step. All combinations being combined by logic. We suspect Samu would agree. Or we like to hope so.
A wee while back we enquired of Commons clerkly colleagues as to whether a decision on a non-fatal motion in the made negative procedure could ever be deferred. An eventuality we’d already catered for in our draft negative procedure. The answer came back that, “Divisions on non-fatal motions triggered after the Moment of Interruption are deferred unless a previous motion agreed to by the House has been agreed to.” Your correspondents admit they have no idea why there’s two agreeds in that sentence. But they’ve learnt not to question. So Jayne and Michael once more applied procedure map scalpels and now deferred divisons on non-fatals are also in the made negative map and data.
On the subject of negative procedures, Librarian Jayne has annotated both of our maps with citations of the legislation that informs them. We would encourage any nerd at the intersection of legislation and parliamentary procedure - and in possession of reasonable eyesight - to take a look and tell us what we’ve got wrong.
Last week the question arose of whether treaties were indeed ‘allocated’ to the EU International Agreements Committee, as both our maps and data suggest. So Jayne checked in with Dominique and we learnt a treaty might be allocated to either the IAC or its parent committee. Proving ourselves quite correct, we backed away and did nothing.
Just as we aim to one day link from our procedure maps to informing legislation, we’d also like to link steps and routes to informing Standing Orders of both Houses where applicable. To this end, we’ve been collaborating with a research group in Oxford who’ve been working on turning House of Commons public Standing Orders into data. This week, they poked us over email to say the latest release of their data is now available. Michael has downloaded the data, attempted to wrap his head around what’s changed and started to process it into our database. So far there is only a list of dates to show for his efforts. But everything has to start somewhere. Thanks Rad, Tom, Niels.
On Wednesday, Librarian Jayne and computational expert Michael made another assault on the summit of motion mountain, only to eventally be driven back by the cold, the snow, and a degree of confusion on the matter of questions not being decided. An email to Mr Evans has been dispatched. Or will be. Questions not decided aside, progress was made. The journey through non-forthwithery is now much simpler, and redundant routes around the withdrawal or lapsing of amendments have been removed. That said, much looks likely to change now we know our logic gates have acquired a third dimension.
Following feedback from Tony, Librarian Jayne has been making steady progress on adding comments to our library of SPARQL queries. This week she polished off the comments for queries involving motions around statutory instruments. As we always say, if you have an interest in learning SPARQL and a passing interest in parliamentary procedure, they’re really not a bad place to start. And we assume, if you’ve read this far, both could be true. Click a link, tinker with the query, see what happens. And let us know what else you’d like to see.
Although the egg timer is nominally complete, we all know that nothing is ever complete. And without maintenance all things eventually submit to entropy. So this week, young Robert has been busying himself with fixes and tidies. These were, in short order:
Switching the application to https. After all, we don’t want random strangers interfering with the boiling of our procedural eggs.
Removing the dependency on external CSS. Which should make things less brittle. It definitely makes things faster.
Making all links more visible in dark mode, to avoid Michael’s eyes breaking any further than absolutely necessary.
Rewriting his comment parser to include links from code line numbers to their GitHub equivalents.
Robert and Michael also spent a good hour turning what had been some ‘unless elses’ into ‘if elses’. They had been using ‘unlesses’ to ask if the laying date of the instrument corresponded to the type of day that should be counted, and if not, heading off into the future to find the first date that did. Because there is no work to do if the laying day is a countable day, they had no ‘else’. Which was fine. But last week, following a chat with Arabella, they added ‘elses’ in order to set the first day of scrutiny to the laying day. Nobody has ever said ‘unless x else y’, and now neither does their code. So that’s good.
If you joined our adventures last week, you’ll know that Jianhan spent a wee while turning the legislation.gov.uk links stored in the data platform from http into https. Because we prefer to store URLs as they are and not rely on the fickleness of redirects. This week he picked up a similar job; links to early day motions having changed in the run up to this summer’s website redesign, some of the URLs we store were of the old pattern. Again redirects were in place but we thought better to update at source. So Jianhan wrote a query and found… ten. An amount which proved easier to fix by hand. Librarian Jayne stepped in and all EDM links are now fixed. It has caused us to question what other URLs we might be storing that have since changed, but that’s a matter for next week.
On Tuesday, Anya and Michael had the second - in a series of two - meetings with House of Commons Library research and editorial types to chat through editorial policy and what to date they’ve been calling the ‘R words’: review, revise, retain, retract and ‘rchive. This time with Grace, Lina, Stefano and Bryn. A new sketch was sketched which turned out to be not entirely dissimilar to the sketch that was sketched in the first meeting.
Anya and Michael spent part of Friday, bodging the two sketches together. Or ‘synthesising design concepts’ as they like to say. It feels like there are further conversations to be had before the sketch settles down and they’re ready to turn it into Turtle. But not a bad start they thought. Not a bad start at all.
Unfortunately, yes. Anya and Robert have been gamified. Because there’s no longer anything to do of an evening - besides watching ITV3 mystery crime dramas and drink - they’ve been battling each other and a host of other folk to match the highest number of Parliament taxonomy terms to Wikidata concepts. Anya has been heard to complain that the interface has a tendency toward the unresponsive on her browser. Whilst also suggesting that Robert might be waking in the night with the sole aim of beating her. Your correspondent will not comment on these allegations. Whilst he is entirely committed to the idea of linking data, in practice he finds it hard to imagine the appeal of staying awake to click buttons with not so much as a biscuit by way of reward. Instead he tends to take to his bed and soak up as much warmth as possible before hitting his morning shower. Which, as he may have mentioned, is currently cold.
Nothing was learnt this week.
