I’m Jenn Phillips-Bacher, and I’m here to tell you about how the Wellcome Library took a different design approach to an increasingly common set of library problems.
We all have lots of digitised stuff and physical stuff lots of metadata evidence that search isn’t always the best way of finding stuff, especially for non-experts wondering how to increase serendipity in a digital space
Over the next 30 minutes I’ll share with you our R&D project to find out What’s In The Library.
To give you a bit of background, let me tell you about us.
Wellcome Library, a research library in London that is open to the public and free to join.
Founded by Henry Wellcome - a Midwestern-Anglo-American like me, whose pharmaceutical company patented the TABLOID, a compressed tablet = pills as we know them.
Henry was an obsessive collector whose associates would trawl the world’s auction houses and sales, searching for artifacts and objects from many of the world’s cultures. He was also an avid book collector and bought lots of materials relating to the history of medicine, including related subjects such as alchemy, witchcraft, anthropology and ethnography.
On his death in the 1936, his vast collection was bestowed upon his new Foundation, and the Wellcome Historical Medical Library was formed. Covering all kinds of topics, from anatomy and surgery, to botany, quackery and the medical professions, nutrition
...and this guy (from an alchemical compendium)
We still actively acquire archives, books, manuscripts, artworks, film and video, all relating in some way to cultural contexts of health.
In short: we are a vibrant, special collections library with AMAZING STUFF
AND HERE’S WHERE I START MY STORY
The year was 2012 and I was working as web editor
We were in the beginning stages of our mass digitization programme, digitizing 22 archive collections relating to history of genetics, over 3000 genetics books, 150 years of London public health reports, artworks etc
To make it easier to access this new digital content, we took an approach to our content that was built on catalogue discovery and curated editorial content
As part of our redesign we built a deep-zoom viewer that would be used to display all of our amazingly high quality digitized content
which we then open sourced
Created what was effectively an online exhibition, or microsite, for our History of Genetics digitised materials - comprised of short essays describing the collections, an interactive timeline, and subject-based lists of digitised books
Made it possible to link through to the free digitised materials through our discovery layer (we use Encore, FWIW)
….and then moved on to something else. And we waited to see what the usage of the digital content would amount to. And we waited. And not a lot happened. We never saw huge gains from our digitisation efforts.
Through interviews, focus groups, testing and analytics, we found 2 Big Problems
The first problem is that click-throughs from the catalogue to the digitised materials weren’t happening at a rate we’d hoped
The second big problem is people were bypassing the links we carefully embedded in the essay texts
No effective way of linking a user from an item in the catalogue, to the contextual, curated information on the microsite
Our digitised content tells many stories, and connects to other parts of the collection in sometimes surprising and uncurated ways
But what we’d effectively built was a serendipity-free cul-de-sac!
From a content strategy point of view, the model we’d built would be completely unsustainable in the long term.
With an ambition to digitise as much of our unique collections as legally possible, we are looking at a massive increase in scale.
In fact, to date, we are ingesting nearly 6000 new items a month, through our in-house digitization and work with the UK Medical Heritage Library via Internet Archive.
Catalogue discovery and old models of online exhibitions are not going to hold up to this level of digitization activity.
We needed to rethink how content discovery could be redesigned for the future – and how we can make a truly digital library collection website that takes advantage of the inherent beauty of the web: linking a world of people, concepts and ideas.
In 2015, Mitchell Whitelaw published a paper in Digital Humanities Quarterly in which
he reviews some of the literature around the activity of Browsing, and provides several case studies of what have come to be known as ‘generous interfaces.’
Generous interfaces are becoming the norm for museum collections, but not so much for library collections.
He suggests that…. “Keyword search is ungenerous: it demands a query, discourages exploration, and withholds more than it provides.”
With this in mind, I wanted to reimagine the library website WITHOUT SEARCH, one that makes the most of the growing digital collection and its assets - bibliographic metadata and digital images.
I want to explore what Generous Interfaces might do for library users.
Marking a total step change in how we were thinking about website development, What’s In The Library was born.
WITL was a 4-week rapid prototyping project held last summer. Our core objective was to expose Library collections - and information about the collections - visually.
To do this work we commissioned a brilliant group of creative technologists led by George Oates of Good Form & Spectacle.
WITL is also the name of the website, which you can find at whatsinthelibrary.com. Caveat - it’s all experimental code! You may experience application errors, slowness, etc. Be gentle!
OUR ORGANIZING PRINCIPLES
Our four weeks of work were based on the four problems we needed to solve. [describe 4 problems]
We operated as a a lightweight agile project
user stories (where stories were projected needs from library staff - NOT users at this stage), Very short sprints that resulted in working prototypes by the end of the week Daily standups : what did we do yesterday, what are we doing today, what are our blockers? End of sprint demos the following Monday; where library staff were invited to attend and bring us their comments and questions
WHAT WE MADE To start, we exported 960K MARC XML records from our system and dumped them into a PostgreSQL database
Now I’m going to show you the types of things we produced over the four weeks.
Ultimately, 3 out of the 4 prototypes became effectively a library catalogue flipped inside out.
It’s not a comprehensive view of everything, but what I hope to convey here is really, how important to the project our catalogue data was – and continues to be!
Week 1 - Scope of the Catalogue - one for the cataloguers, by far Visualizing in a graph the ‘fullness’ of MARC metadata, field by field This helped us to see where there are gaps in data, or which fields just aren’t being used at all. From a web production point of view, this prototype helped us to see where we could reliably extract reusable catalogue data, for example, Title field, Author field, Pub date, Subject, Genre.
Week 2 - Show the Thing
The goal this week was both to maximise use of our digital images, to show things by time period, and to display interesting groupings of things Range of visualizations
Graph of collection by year
Playfulness seeps in to the interface
These monkey years indicate dirty / incomplete data
During this week we also tried to show all digitised THINGS in SPECIFIC CATEGORIES
Here’s an example of our caricatures
While many people may already be accustomed to seeing this kind of view in your own digital library sites, what I thought was important to do here – because our own discovery layer can’t handle it – is to show the thumbnails. We have evidence from GA that our most used stuff on the site are those items that include a thumbnail.
Lists of Subjects (LCSH and MeSH headings) Lists of Genres
Displaying digitised content programmatically wherever we have it Fleshed out what an object/item page could look like
Week 3 - Context was about editorial development on the 18th century satirist James Gillray. While it may seem an anomaly alongside the other three weeks, this is where we wanted to poke around the edges of ‘outsourcing’ context - how much expertise can we bring to our collections, and how much already exists on the web. Here we were looking to Wikipedia to provide biographical and historical background. 1 week isn’t enough to do it justice, but we made a start!
Week 4 Was about browsing all of our metadata at scale, and is the culmination of everything we learned in the previous weeks. This prototype explores over 960K items, 233K people, 99K subjects From this prototype, we now have a page per Author/Creator, and a page for every subject, and a page for every item with a MARC record
Our design goal is to get people moving laterally through the collection, through contemporaries and co-occurring subjects. We wanted to show the connections between people and topics.
Here’s an example of a Person page – Bio from Wikipedia where possible Subjects written about Further down the page, a list of her things
And this is an example of what’s further down as you scroll….
You’ve joined me on a march through the week-by-week outputs - and you might be wondering what the response has been
In many ways, this 4 week design project has had positive effects.
Most positive impact= served as a catalyst for internal change - for changing how we design and build websites, for changing how we think about descriptive metadata across all of our collections.
We are already building the notion of ‘generous interface’ into our new vision for digital engagement.
But it’s been disruptive, too. Disruptive in the sense that it is forcing us to think creatively about our underlying systems, practices and ways of working.
WITL intentionally shifted the solidity of the catalogue silo and hand-crafted editorial content we were trying to make work.
Four Big Areas of Disruption:
By exposing our metadata in a visual way, it allows us to question our collection development policies. “I thought we had more on that topic”
BIBLIOGRAPHIC AND ARCHIVAL DESCRIPTION We’ve opened up a conversation about bibliographic data, too. How much data can we agree is for the user? How much is there for data exchange? For the system’s own fixed fields? Can cataloguers and editors work together to prioritize effort in describing the collections? How much rich descriptive data have we lost out on because of the sheer mass of backlogs, and pragmatic cataloguing policies that limit the amount of time spent on ‘local colour’?
SYSTEMS We are also questioning how we interact with closed, third party systems. We need a collections data API that allows us to build better interfaces. Will we get this from our vendor? Most likely not.
WEB DEV TEAMS
By hiring a creative agency to work with us, we shone a light on what we need in order to make an amazing library collections interface: we need bums on seats if we want to innovate. We can’t design without designers We can’t develop without developers.
This can’t not be disruptive!
There’s a lot of stuff that needs to happen inside the library to make it possible to liberate your collections from the tyranny of search!
What have we learned so far?
We’ve recently taken our week 4 prototype out into the open, as alpha.wellcomelibrary.org
In January, four of us spent a day doing interface testing in the library. We showed the site to 23 visitors of varying degrees of expertise and age ranges.
We wanted to test the ideas, rather than the execution, so the test was fairly open-ended, and discursive. Our test objective was to observe whether people would browse or explore the site unprompted, following their curiosity through the site.
Our main objective with Alpha is to offer ways to browse the collection by subject, person, genre and collection – and to show EVERYTHING, whether digitised, or merely a record for a physical thing.
What we heard:
One thing we heard in the interface testing is varying tolerances of a visual interface - it can throw people off if they were expecting boring, text-heavy interfaces. Which, let’s be honest, many research library users are accustomed to.
The challenge here is to design a visual and engaging collections interface that makes the job of research easier from the start.. But also a challenge for graphic designers everywhere! Colour palette, font choice, and language…
A publishing graph raised questions from savvier users - made apparent that there are gaps in our collections during certain time periods. But why? Was there a publishing lull? Are there things in the collections that are mis-dated? Was a librarian or curator particularly interested in a specific time in history? How do we make it apparent to researchers WHY the data is telling this story? Thinking about the interface, then: we need to add better signposting, possibly a function to suggest corrections to the data or suggest we acquire something else!
There’s more to do to find out whether offering a way to browse by subject, person, genre or collection does actually increase usage, and engagement with, the collections.
Over the coming months, we’ll also be focussing on what the data can tell us. Besides using Google Analytics to monitor usage, we are also experimenting with two other products: Hotjar - heatmaps, scroll tracking Optimizely - for A/B testing
I hope to have data to share this summer
CLOSING IT OFF
While we don’t yet have tons of user research, I can say without a doubt that this design approach has helped us to see into our collections better.
By visualizing our own data, we start to see how it can be repurposed for interfaces that don’t rely on search. We can also see what improvements we need to make to ensure the interface is reliable and trustworthy (as well as interesting and delightful!)
By taking a generous approach to the interface, and by offering ways for users to explore the collections without search, we might begin to see an Increase in usage, and therefore a better ROI, for our digital content Connect materials at scale, offering serendipitous paths through the collections
I want to encourage everyone, whether you can only dedicate a day to it, or a week...or even a month, to learn more about your collections data.
What hidden connections can be made?
What can your collections tell you, and your users, about themselves?
How can you introduce generosity to your library interfaces?
What's In The Library?: Prototyping the Future of Digital Collections
What’s In The Library?
Prototyping the future of digital collections
Designing for Digital 2016
Jenn Phillips-Bacher / @MrsAudiac
Web Manager @wellcomelibrary
“Keyword search is ungenerous: it
demands a query, discourages
exploration, and withholds more than it
- Mitchell Whitelaw @mtchl
The problem space
Week 1 : Scope
How much and what is in the library catalogue?
Week 2 : Showing the Thing
What do you have on Topic X?
Week 3 : Context
As editors, how can we mix our own editorial content with
another content provider, like Wikipedia?
Week 4 : Scalability
How can we provide context on a large scale?
What can your
collections tell you
What’s In the Library?
Mitchell Whitelaw. “Generous Interfaces for Digital Cultural
Collections”. Digital Humanities Quarterly (2015). Volume 9
Jenn Phillips-Bacher / @MrsAudiac / @wellcomelibrary