I recently completed two weeks of fast design for The National Archives (TNA) to sketch out some visual patterns and other views that could be used at the aggregate level across millions of archival records held at Kew and the other 2,500 archives across the UK whose catalogues are part of the TNA Discovery service. The deliverable was a bucket of files and drawings so there’s nothing to click around really, but I can write about what we did and share some pics…
“I asked George to look at how we might help people understand what collections are held nationally around certain themes. The National Archives’ holdings are large and difficult to get an overall sense of but this is compounded by adding in data from the hundreds of other archives recorded in our systems. What would it be like to try and get a sense of key national collections around topics like migration, LGBT history and the performing arts? How could we convey the spread and coverage of different collections?”Jo Pugh’s brief
Such a short piece of work doesn’t especially have a process. We thought it would be easier to just hit GO, and see what happens. (Thanks to Jo for being OK with not being certain what the outcomes would be. This is so valuable (and essential?) to exploratory work like this.) But, I interviewed Jo to make sure I understood what he needed from the work, who his audience was, and what he wanted to do with whatever we made. I explained that I would need to situate myself in the giant TNA dataset and website, absorb it so I understood the skeletal data structure and current exploration structure. I need to be able to confidently do those two things before I can suggest ways to re-present that same information in a new way, which was ultimately what we were trying to do, with the new concept of Topics.
I always prefer to work with real data instead of fluffy made-up placeholder data too, so Jo provided me with three topics we could explore using gigantic search queries: migration, LGBT histories, and performing arts. These were great because I could be exploring the actual result set, albeit trapped in search facets and the maze-like archival record UI.
The first step I usually take to situate myself in a new dataset is to look at the superset of fields available in a record. In this case, I gathered field labels as I went along navigating dark corners of the Discovery service, building a list by exploring. Gaps were supplemented by Jo, and we found another list in an old email that was pretty similar, but interestingly also a bit different. I enjoyed reconciling it all, and thinking about grouping the fields. The Discovery UI just draws all fields present in any records out in a table format dumped in a page from the record, and that isn’t great for scanning. You can do much more for your information legibility by positioning related things proximally.
Usability of vast, thin catalogues
“Maze-like” is justified by the three key questions Jo reported visitors to the Discovery service having:
“Where do I look?“
This might just be a reaction to a very text-dense layout without much helpful hierarchy. Everything looks the same, and there’s no real sense of location or depth. It’s easy to be looking at an amazing set of archival materials but for them to appear just like every other one. There is a structure to any archive, and it’s really heterogeneous. Archivists describe different structures narratively, and work to ensure each archive’s arrangement is reflected in its catalogue record(s). A fonds is the name for the very top level of an archive:
“What do I type?“
I think this question might be a results of the expectations searching Google and the web have created in us all. This amazing technology has made us all feel like anything we want will just come up, and this just does not work in individual cultural collections (yet), no matter how big and important they are. Searching individual thin records is often very unsatisfying. Even when you know exactly what you’re looking for it can be hard to retrieve in an online search that isn’t Google.
“What am I looking at?“
Archival UIs can be quite disorienting. Archives are organised structurally, in fonds, series, sub-series etc, but it’s probably not crazy to say that no two archives are the same. So, even as you’re moving up and down one structure, any expectations you have from the previous one you were just looking in will be challenged by the next one. But, it’s also not easy to see the overall shape of any one fonds and its contents, and that was an idea I began tickling in the sketches.
Busting out the pencils!
It was a real pleasure to make some hand-drawn layouts using our three topics and exploring the Discovery service to find real results. It was great to take a bit of time here to think through ideas about how to show these topics’ constituent aggregate data points. How can we show how the migration archives have changed over time? Who are the prominent characters in LGBT histories? Which archives hold the most performing arts materials?
Showing shape to help with orientation
As well as doing what’s now becoming baseline spelunking design – showing search facets visually, and more textually, interlinking more views, showing total counts/ordered lists/taxonomies where available – I wondered if there is a way to show the top level size and organisation for each archive fonds, so they could be distinguished visually (while realising this is not accessible to visually impaired visitors, although could be audio-described interestingly, I reckon). I wondered if dendrograms could help show these shapes, so set about doing some experimenting with Observable and real data from the Discovery platform, hand-cranked into these lovely shapes:
Perhaps some kind of visual like this could help people navigate around and differentiate fonds from each other… we’ll see.
In hindsight I’m mildly surprised how much stuff I generated for the TNA team – I suspect this was a direct result of being locked down for 12 months? I made a basic HTML site too, to get a feel for how it might be to navigate all those records from these higher level constructs like topic, subject, creator and so forth. I was definitely encouraging TNA to expose much more of the hidden taxonomic structures they use to organise things – those sorts of lists are really accessible, I reckon.
Here’s what Jo wrote for me when I asked him for a quote for this blog post:
It’s very valuable working with someone who has such an outstanding grasp of usability and interaction design principles and is familiar with the cultural heritage domain but at the same time doesn’t have the baggage that comes from working with our specific systems and data every day. Someone who can ask ‘why aren’t you making more use of that?’ – the answer is often to ask ourselves the same question.
I’m well aware of the value of overviews – macroscopic, zoomed out summaries of big messy collections offer a way in to the complexity and help sensemaking by researchers. But that’s a rather abstract conviction. George was able to systematically demonstrate how to represent aspects of our data in visual form, encourage us to make more extensive use of underexposed categorisation and rapidly sketch expected information journeys. The role of an aggregator like Discovery in such journeys is by no means completely obvious and it certainly got me thinking some more about how we best serve users who are often looking for a digitised object which we not only cannot readily provide but likely doesn’t exist. There’s no shame in being a treasure map (if you’re a good one) but there’s no point in pretending that a map and a treasure chest are functionally identical.
We will definitely be taking these ideas further in the future and in the meantime we have a cornucopia of wireframes, sketches and visualisations to help us. Not to mention dendrograms. Turns out I love a good dendrogram.Jo Pugh
The deliverable was basically “what ever you end up with” as a bucket of files, and a presentation of the work to TNA staff. I’m secretly hoping the parcel of work I made gets popped in The National Archives somewhere, so I took a moment to write a note to the future in the form of a README for the .ZIP bucket:
README 19 March 2021 This folder of files represents the work undertaken by George Oates at Good, Form & Spectacle for Jo Pugh at The National Archives (TNA). It was sent as a .ZIP to Jo via WeTransfer on 19 March 2021. The brief from Jo to George was to help sketch out options for new Topic pages, and we picked the example topics of migration, LGBT, and performing arts. George set to work for about 10 days in February 2021, and this folder contains all of what she produced. George presented this work to a group of about 30 TNA staff on this day, and apparently Jo made a recording of that. She and Jo worked closely together, talking every day or two throughout, and Jo provided several data samples and example search queries so George could approximate real result sets using the TNA website that was live in February 2021. George reported on her progress using the tna-progress-log.key file which she has also saved as a PDF. The Google Sheets files may be of interest, as they are George’s attempts at assembling and massaging TNA data structures to “situate” herself in the information system. The html folder contains all the hand-cranked HTML pages George built to investigate how Topics, Creators and Archives might appear in a more browsable interface. It’s rough, but most links should work (to other pages within the folder). This site is also online at time of writing at http://goodformandspectacle.com/tna, but that probably won’t last forever. The project folder is labelled 020 The National Archives UK because that is the 20th client of Good, Form & Spectacle. Enjoy!