So here we are. Week 51. Haçienda that then. As we say up north.
The usual pattern of weeknotes is to sign off on a week of work. This week, as our crack team of librarians and computational experts head off on a much needed and deserved vacation at home, we’re signing off on the year. And what a year. Innately possessed of sunny dispositions, we remain convinced that things can only get better. But we’ve been mistaken before. For now, we put our desire to give 2020 the kick up the bum it so richly deserves firmly on the record.
If you read no further, we wish all the very best to you and yours with whatever festival you choose to celebrate. And a much happier new year for all of us.
So then. About the work…
Much of our Jianhan’s time has been spent tidying up machine generated visualisations, to be compliant with the new logic gate based approach to procedural pedantry. Ably assisted by librarian Jayne and computational expert Michael. As we mentioned last week, we think the new pictures are really rather snazzy - if a little more verbose than those that preceded them. But with verbosity comes precision. At least in this case.
Jayne and Michael spent an hour on Thursday talking through the changes with librarians Claire and Emma. We suspect both were a little horrified to see the spectre of GCSE maths resurrected in their working lives. Claire speculated she might still have a logic gate stencil set somewhere. Probably nestled alongside some strawberry scented erasers, a propelling pencil and a picture of Bros. We had, overly ambitiously as it turned out, hoped to sign off on Jianhan’s work or at least provide feedback for a few more tweaks. Instead, we’ll sit down again next week and chat through what we’ve done, tease out whether our naming and typing of steps is somewhat redundant, and ponder what - if anything - we plan to do about that. There’s also a suspicion the machine maps might need a key. But as yet no agreement on what should be in it.
Until we’re happy with the procedure visualisations we’re blocked on the rewriting of work package visualisations, as we intend to base the latter largely on the former. But this did not stop young Robert and Michael applying their by now somewhat blunt, tiny chisels to writing up design notes on the parsing of work packages with logic gates. And rather pleased with their work they are too. You should take a read. If one of you is Santa, you might also consider coming to the aid of our itinerant chisellers.
Unable to make progress on the work package visualisation work, Jianhan cast around for another quick task. And Michael was more than happy to lumber him with writing reports to check that the number of inputs and outputs to steps matched expectations. This was occasioned by Jayne and Michael accidentally reversing an arrow on their hand-drawn proposed negative statutory instrument map and adding it to the data without noticing. Which lead to a NOT gate having two inputs and no outputs. Not a thing a machine expects to encounter. Jianhan’s code gathers every step connected to a route in a procedure, loops through those steps, counts the number of inbound and outbound routes in that procedure, checks whether the numbers meet the expected counts, and raises an error if they don’t. It’s already proved its value, catching an OR gate that Jayne and Michael had accidentally created as a NOT. A glitch that is now fixed. By next week we hope to have a daily email detailing everything we’ve got wrong. Just imagine. And Jayne and Michael will check the checking without having to mither Jianhan any more than is strictly necessary. Another example of machines being helpful to the cause of good data governance.
The business of the OR being added as a NOT led to Michael adding a new card to check that all of our PNSI steps are of the appropriate type. Sorry Jayne. But now he’s wondering if this could also be automated - our step names follow a predictable naming strategy, so it certainly should be possible. But that does, once again, raise the question of whether our naming strategy is correct. Which puts us back at the start of the circle. More thought needed next week. If we’re capable of thought next week.
Jayne and Michael convened in pixels on Tuesday to make a tweak or two to our treaty procedure maps and data. Correspondence with Alex had revealed that both the International Trade Committee and the Joint Committee on Human Rights were more than likely to open calls for evidence as part of their scrutiny of treaties. So two new steps and accompanying routes have been added. Alex also mentioned we should probably add an inquiry opened step for the JCHR. So that has also been added and routes adjusted accordingly.
Jayne has been filling gaps in her all too busy schedule with the addition of legislation citation blobs to our assorted procedure maps. So far she’s managed to polish off five of the eight, and this week she added another. But not by doing any work. In a conversation with Michael, they came to the conclusion that adding legislation citations to the made affirmative procedure would not be possible because the legislation informing procedure for such instruments is scattered far and wide. So this note is merely for completeness.
An email that had bounced its way through Martyn, Mr Evans and assorted other clerks finally found its way into Peter’s inbox and an answer was forthcoming. Our treaty map allows for debate on a motion that the treaty should not be ratified to take place in a Delegated Legislation Committee. Arabella had questioned whether that was ever likely to happen or whether the Government might find time on the floor of the House. Or, failing that, whether it might be debated on the floor during an opposition day. Peter got back to say that - whilst this is not a thing that’s happened - it’s theoretically possible that the Government might throw those opposing the treaty a bone, and allocate to a DLC. This under public Standing Order 118, he believed. Given the treaty procedure is chock full of things that have never happened and given we already have these steps and routes, this was another happy occasion on which no work was needed. Thanks Peter.
On the subject of standing orders, having converted the ParlRules data to a schema that’s more amenable to our needs, Robert and Michael set about converting the relational model to something more ontological and OWL-like. Those with an interest in standing orders are encouraged to read their comments and get in touch if anything is unclear or inaccurate. A conversation with our friends in Oxford is pencilled in. But like much else, that will have to wait until the newer, better year.
A couple of things have cropped up recently that made us wonder if it might be reasonable to extend SKOS to make broader a class rather than a predicate. The first was occasioned by our exploration of the House of Lords hanging files - first encountered back in week 40 - where we’d like to maintain some record of taxonomic structures over time. A quick Google revealed other people who had grappled with this problem had settled on named graphs to bind properties into time spans. But we are suspicious about named graphs. As we are suspicious of all means by which computers hide their brains. The second was occasioned by our thought experiments on how we might maintain a single set of Concepts and combine them into different structures for different consumers. Both led Anya and Michael to an audio-visual telephone call with Silver and Julian. Who know about such things. The idea of class-ifying broader was not laughed out of school but we need to give more thought to the tradeoffs here.
Back in week 46, Anya and Michael spent an hour or two cutting and shutting some pictures they’d scribbled with assorted Library types. They’d already cobbled together a FRBR based record model and an agent based publisher model to capture the outputs of assorted parliamentary research services. And they were keen to add more details around reviews of such content and any resulting revisions, withdrawals and archivings. Or what they used to call the R words before the words changed. This week they adjusted their diagram, removing redundancy with an additional layer of subclassing, and began turning it into turtle. Which they then turned into HTML. You might notice that the diagram is a little scrappy and that some of the classes and properties are still lacking comments. That also is a job for next week.
An edition of weeknotes would not be complete without one of our regular egg timer updates, and this week is no exception. On Wednesday, JO Jane broke the news that the House of Lords would rise on Thursday for Christmas recess. The same announcement followed from the House of Commons in short order. Librarian Jayshree updated our assorted Google calendars in double quick time, our sync script did not let us down, and the new data was sucked successfully into the egg timer. All dates returned matched those expected by JO Jane and she declared herself happy. Which obviously makes us happy. Testing with instruments laid during recess is the next hurdle we must cross. More news on the other side.
Anya had wondered if we should perhaps add a legal disclaimer. It’s not been a great year and we don’t want to start the next one with being sued by the gubmint. So, with help and advice from Alex and Edward, our disclaimer is now live. Thanks both.
For the sake of completeness, Robert has also tweaked the fonts a little. So everything is a little less bold. Which proves to Michael that whining pays dividends.
Anya, Ian, Robert and Michael got together on Tuesday to expand their list of source systems, adding details of which systems feed which platforms. Boxes were ticked for both the data platform and the search and indexing service. The data.parliament column still needs attention. But there is time. This is not our last year. Even though it has at times felt like it.
Meanwhile, Librarian Jayne has been busy helping our colleagues in Software Engineering test changes to the Statutory Instruments website. Made affirmative instruments are now showing the correct date for making, links between PNSIs and subsequent SIs and vice versa are now in place and the filter for instruments with motions tabled now returns the correct number of results. Top work all.
And with that, we bid you so long. And indeed farewell. And hope to see you on the other side.