Excited to let you know that the work I’ve done at G,F&S has made its way into academia.

Abstract

George Oates is a web designer, producer and developer who has been working with digital collections since 2003. She was one the founding team of Flickr, and the originator of Flickr Commons. She currently heads heritage design studio Good, Form & Spectacle, working with institutions including the British Museum, the Wellcome Library and the V&A. In this interview with Mitchell Whitelaw, Oates discusses her developing practice in making and remaking digital collections, spanning web-scale services, unsolicited interfaces, messy metadata and gentle interventions in institutional practice.

Oates, G. & Whitelaw, M., (2018). George Oates: Making and Remaking Collections Online. Open Library of Humanities. 4(1), p.32. DOI: http://doi.org/10.16995/olh.325

Advertisements

After a fun three-ish years in fancy Bloomsbury, we’re off further east, to Hoxton. It’s been lovely being in the middle of everything, and down the street from the British Museum (and that excellent currywurst place), but our landlord, in all their wisdom, is ending our lease because they want to refurbish. We keep wondering whether they’ll uncover any of Hans Sloane’s collection buried in the garden!

We’ll be at:
Unit 9, The Energy Centre
Bowling Green Walk
London, N1 6AL

During our Bloomsbury tenure, I took a bunch of panoramas of the office in whatever state it was in, for the record.

IMG_4561.JPG

We’re happy to announce our latest project, for the newly-opened Postal Museum in Clerkenwell! The new museum opened late July, and included a new Discovery Room, a brilliant public space where you can access archival objects directly, upon request. We were tasked with making the first instalment of an interactive table to help people browse around the collections, and hopefully be prompted to ask an archivist a question, or delve a little deeper.

Discovery Room

IMG_4618.JPG

Part of the new building is this new public space you can visit, and ask to see original materials (as well as browse the digital collection and use our swanky new touch table). You need to allow a bit of time for things to be retrieved from the stores, but, why not give it a try? It’s upstairs on the first floor, and free!

Two Collections to Start

The Postal Museum has a huge trove of all sorts of visual materials, from the obvious philately collections, to letters, to maps, to posters, to photography. For this first stage, they wanted us to show the core collection of the museum, the R. M. Phillips Collection (which tracks the history of the UK postage stamp in unparalleled and thorough detail), and a selection of Post Office: Photograph Library, mostly around post and war, but with some vehicles and interiors too. We’ve tried to keep in mind that it would be nice to incorporate more collections over time too, once the new building has settled in a bit.

We used a metadata export from the museum’s CALM database, served up images using IIIF, served with Cantaloupe, through a Ruby app.

I’ve done a bad job of getting nice images of the actual UI, so will endeavour to update this post with some once I have my act together!

Seeing Things Big

We knew we wanted to take full advantage of the affordances of a lovely big screen and simple touch gestures. Rather than crowding the interface with lots of niggly controls, we wanted the brilliant imagery to take centre stage. Even as we were getting to know the R. M. Phillips Collection, we just kept zooming in all the time to see the lovely details, so wanted to help others do that too.

2017-05-17 10.56.50.jpgScreen Shot 2017-05-17 at 18.12.04.pngIMG_4566.JPG

This is where we ended up, the view when you’re looking at a single thing, really zoomed in:

Screenshot from 2017-07-27 16:26:45

Collaborators

This was our first collaboration with Buckley Williams, and we sincerely hope it wasn’t the last! Thank you, Nat & Dan.

2017-05-18 16.18.00

A favourite part of the collaboration (apart from the consistently snappy outfits and extreme talent) was our placeholder image, which we used on the table so we could proceed with the UI as we loaded all the whopper images… it’s one of those ones that you know won’t slip through.

DDVgoTmWAAAMwi9

DateRanger

As you may be aware, we try to always look for small utilities we think other people might be able to use, like the marcxml-to-postgres slurper, or the dimension-drawer, and we’ve found another one, called DateRanger. Nat has posted separately about this, so if you’re interested in why the heck dates are so hard in cultural collections, maybe head over there for a look.

This is a post by Nat Buckley, of Buckley Williams, our recent collaborators on the new Postal Museum Touch Table project.

As humans we find it easy to quickly read a variety of date formats and almost instantly understand what they represent. We can glance at dates written as “23rd Jan 1894”, “Jan-Mar 1856”, “c. 1960” or “1945-1949” and we know how big the date range is and how accurate it might be. These are just a few examples of dates formats we found in the Postal Museum collections.

To work with those dates effectively in our software we needed to find a way to parse them and represent them in a single format. There are many software libraries designed for translating between standardised date formats (eg. ones used by different countries), but parsing formats commonly used in archives is a slightly less popular problem to solve. Archives tend not to have an enforced, fixed way of writing down dates, so there can be a surprising variety of notations. This isn’t a bad thing — it gives the archivists the flexibility they need to represent their knowledge about the objects under their care. Each collection might have its own quirks.

Some smaller software libraries do take on a challenge of parsing dates from more natural, human-readable formats, but we decided to devise our own way. We had a very specific set of formats and couldn’t find an existing solution that could deal with all of them easily.

I wrote DateRanger, a Ruby library which takes in those formats and translates them into a data structure which represents the start and end of the date range. It makes it straightforward to understand the accuracy of the date — the wider the range, the less accurate or specific the date was to begin with. We’d love to see contributions from anyone interested in expanding how many formats DateRanger can work with.

I used automated tests to build up the code in stages, starting from parsing really simple dates, and culminating at testing even obscure format combinations that we didn’t quite encounter in our data sample. The tests-first approach meant I managed to notice and catch some pretty confusing bugs really early on.

We used DateRanger on the Postal Museum touch table, to help us determine where on the timeline to place the collection records. We did, however, use the original date formats from the archive to display to the viewer. After all, those are already perfectly human-readable.

Photograph of Rousseau installed at MoMA

Installation photograph from The Museum Collection of Painting and Sculpture exhibition, held 20 June 1945 to 13 February 1946

On September 7, 2016, The Museum of Modern Art (MoMA) in New York cracked open a new repository on the code repository service, Github. In addition to the release of their popular collection dataset, the museum also made the unprecedented decision to share the exhibition history of the museum (from 1929 to 1989), which also included a ton of archival material like press releases, installation photographs, and catalogues. You can see the official version on moma.org.

MoMA asked Good, Form & Spectacle to make a spelunker to showcase this data, and we were thrilled to oblige.

Voila!   http://spelunker.moma.org

moma-screenshot-home

First Steps

As with all our previous spelunkers, the first step we usually take is to map out a basic plan for the information and how you’ll move around in it. This is often a basic List -> Item pattern, where List can be made up from any of the main data elements. In the case of MoMA, that was Exhibitions, Roles, People/Orgs, and Departments. From each list view, you can move to a single instance of that type of thing, and from that single instance, back out into others, like from an exhibition to the year it happened, or to an artist in that exhibition. (We also get a Python project running on Heroku, and pop our own first commits on Github.)

The main reason we enjoyed making the MoMA Exhibition Spelunker so much was because the data was of a different nature to a collection dataset. As well as listing each exhibition over that period, the data also shows all the people who were involved, both the artists, but also, interestingly, museum staff and other collaborators. We’ve looked a little at how metadata can represent institutional dynamics with Week 1 of the What’s in the Library? project with Wellcome Library, but this was altogether more fine-grained. Who are the actors in this institution? Can  you see their influence?

First Impressions

So, once we have the basic List -> Item scaffold in place, it’s really a matter of trying to answer the questions that come up for us as we poke around all the corners of the data. One of the first simple visuals we added as the basic timeline, to show quickly and clearly what happened, and when.

moma-screenshot-exhibition-graph

Straight away, even a simple graphic like this asks questions. What’s that peak in World War II? Why the dip in the 50s? Why was 1978 such a full year?

It’s important to say at this point that a lot of what I’ll write next about what I discovered in the MoMA Spelunker is pure conjecture and presumption, based on looking at this data quite a lot. It may not be true, at all. (See Open Data, Assumptions and Naïvety below.) 

The Museum & People Dynamics

We don’t often get to see or know the dynamics and politics that go on within the walls of a museum. It can be such a designed space, all about the art or objects, that often the people who made an exhibition come true are almost entirely invisible. It was a real pleasure to work on this data from MoMA specifically because it isn’t about the art, but about the people. When we first began the project, I was keen to be able to show the art, but that desire was quickly supplanted by interest in, and display of, who was doing what.

One of the visualisations we made was about directors of departments, how long they were directors, and in some cases, showing where people went as they moved around the museum, in some cases in careers spanning 40 years. You can see in this screenshot what we can highlight a single individual as they move around too, so you can see who skipped where and when, like John Elderfield

moma-screenshot-department-map

It was bordering on titillating to imagine who might have worked together and how such long term tenures really did shape the museum as it is today. But, it’s also crystal clear that a little map like this can reveal nothing about why people moved around. You’ll need to look in the Archives to learn those stories.

MoMA Characters Emerge

I was enjoying the way that showing this kind of data over time helps you spot blobs and trends and gaps in the data. I was interested to try to uncover whether we could show who the key staff were that really got MoMA off the ground in the early years. By taking a role – Curator – and drawing it over time, you can see quickly see who was kicking things off, in this case, the founding director, Alfred H. Barr, Jr., curated 36 exhibitions, and worked there for about 40 years.

moma-screenshot-first-curators

Something interesting happens when you change how that list is sorted. We can show the same data, but order by who curated the most exhibitions, to get a really different picture that shows the curators who’ve worked on the most exhibitions, and how much they made in any given year. It’s there that you start to notice the effort of curators like William S. Lieberman, who led three different departments over his career, or Dorothy C. Miller, who worked at the museum for about 30 years, and was head of the Department of Painting and Sculpture for a relatively short time, too. Did they work together? Did they like each other?

moma-screenshot-prolific-curators

We liked this “most appearances” view much more, so set that as the default.

This list view is cool too, when you’re looking at artists. There’s a lot of info squished into that single list view, and the overall impression is mostly that MoMA has an incredible collection full of heavyweights (and very good relationships with collectors and other museums), and shares it with the world a lot! Here are the most exhibited artists in the 1929-1989 data… could you infer Jasper Johns blasted on to the scene at some point there?

moma-screenshot-artists-top10

One small design thing I introduced was to show a small  symbol in big lists of people, so you could spot women quickly. This revealed a little gap in the data, where gender isn’t always noted. Since perfect is the enemy of good, rather than remove the feature, I’m trying to help by slowly making additions/corrections to a copy of the data, which MoMA is welcome to. (Want to help?)

World War II and The Responsive Museum

defenseflyer008Something that piqued my curiosity almost immediately was about how the museum operated in World War II, 1939-1945. They were exhibiting a lot, and browsing 1942 and 1943 in particular, there were clearly lots of exhibitions related to the war.

I first noticed the annual exhibitions called Useful Objects of American Design under $10.00 that opened in 1939, and repeated for a few years thereafter. I found myself wondering if this was austerity-related, but then realised I didn’t know the equivalent of $10 in today’s money! Nevertheless…

Then I noticed several others that were much more specific to war, starting to emerge in 1940 and many more in 1942/1943, and beyond, like War Comes to the People: A Story Written With The LensNational Defense Poster CompetitionArt in War: OEM Purchases from a National CompetitionWartime HousingRoad to VictoryCamouflage for Civilian DefenseUnited Hemisphere Poster CompetitionThe Museum and the WarNational War Poster CompetitionArt Education in WartimeWar Caricatures by Hoffmeister and PeelAirways to PeaceMagazine Cover Competition: Women in Necessary Civilian Employment, and more.

cri_000000350367.jpg

Airways to Peace (and a cool interactive dymaxion globe!)

With a little further digging beyond our dataset, I quickly discovered that “numerous exhibitions at The Museum of Modern Art were produced in collaboration with the United States government,” as it continued to exhibit the very best in modern art, including Starry Night, which MoMA acquired in 1941, and exhibited soon after.

w1siziisijezmza3nsjdlfsiccisimnvbnzlcnqilcitcmvzaxplideyodb4mti4mfx1mdazzsjdxq

I simply don’t know how data like ours could possibly show that Victor D’Amico chaired the Committee on Art in American Education and Society, which established as “art education’s answer to Fascism and its contempt for creative art. We hope to mobilise the art educators and students of America, combining all their art efforts, large and small, throughout the nation to work for victory.” It’s just one or two steps away from his basic data, and you always have to stop making new columns in data at some point, don’t you?

moma-screenshot-pressrelease

The committee worked throughout the year to keep art on the educational agenda. There’s a lot more background about the museum and the war available in this archival finding aid online: The Museum and the War Effort: Artistic Freedom and Reporting for “The Cause”. (It’s worth a read.)

I have to say it makes me wonder how and if museums in the USA (and elsewhere) are mobilising to produce a “vast program of art activity” in this way to combat the 45th president. Ahem. In any case, I actually removed a bunch of other “this was interesting” and “I enjoyed this” links and stuff. You can find your own way!

What did the critics say?

NATURE and social satire are the themes of two current shows here. Both are important and well worth seeing.

Jacob Deschin, on Elliott Erwitt: Improbable Photographs

Remember mashups? When publicly accessible code-level interfaces (or APIs) became a thing back in the day, the fantasy was that all manner of mashups could be made to combine and recombine data from all over into compelling new presentations.

One of our early ideas for this spelunker was to try to bring in content from the fabulous New York Times archive, a treasure trove of history around New York City and its surrounds from 1851 to the present. We knew that The New York Times regularly reviews events and exhibitions at MoMA, so it was simple step to try to combine that with the MoMA exhibition data. Luckily, too, the MoMA team had already done a lot of the work to connect exhibitions to articles, which was hugely helpful.

Could we show what The New York Times critics said about MoMA through the twentieth century? Yes! Critics such as Edward Alden Jewell, who was watching from the start in 1929, or Jacob Deschin, who often wrote about photography exhibitions from the late 1940s to the late 1960s, also form part of the fabric of the exhibition history at MoMA.

(Note that you can’t see full articles unless you’re a subscriber. We can show first paragraphs, which is a start.)

Open Data, Assumptions & Naïvety

This work for MoMA has been interestingly different from previous spelunkers we’ve made. In other projects, we’ve made exploratory interfaces into object-level metadata, which is (arguably) simply factual, representing objects and their attributes. While there may be errors or omissions in this kind of metadata, each object is as well-described as possible. Sometimes, viewing this data in the aggregate can bring insights — like seeing instantly that prints form the largest group of things by type at the Victoria & Albert Museum — but there’s really not much ‘colour’ to it.

Part of the provocation of the spelunker concept is to challenge the notion that people know what they’re looking for when they encounter a new museum’s collection, and I’m wondering if that could be extended to museum datasets. It seems to me that “drawing” this data makes it easier  to hypothesise about and ask questions of than examining a big .CSV file. You follow your instincts or an image that appeals or a person you recognise or a theme you’re into, and, I think, start to form your own opinions pretty quickly. The challenge is that this metadata can be quite rich, but, at least in our experience so far, also pretty superficial, so the picture that’s drawn for you is just a surface view. Perhaps though, it’s like a physical exhibition where you don’t read any labels (or there aren’t any), and you’re left to make your own judgements, some of which may be wrong, but all are personal.

As we were fleshing out the interface to the MoMA spelunker, I found myself making all sorts of assumptions about the institutional dynamics at the museum related to who was working when, and why they might have moved around, their areas of interest or speciality, and things like that. I’d written about some of that sort of stuff on the About page, but the kind folk at MoMA archives were good enough to let me know that some of those assumptions were just plain wrong! Maybe it was more that the data we were able to display only gives you a tiny glimpse into the actual dynamics, and it’s simply a must to explore more deeply with experts or other source material. I don’t want to draw wrong conclusions, and experts in house, who live with this information day to day, surely don’t want to express things that are wrong. Of course not. I think what I’m coming around to is that these sorts of explorers help orient viewers towards questions they’d like to answer, once they’re acclimatised to the terrain. That seems good!

More broadly speaking, we’ve now made six spelunkers at G,F&S, and it’s probably about time I had a proper think about how well they’ve worked and whether they’re useful. More on that later…

The team for this project was George Oates (design, project lead), and Phil Gyford (engineering). Thanks, Phil!

We’re very pleased to have some lovely new shelves made and installed by Scott Stannard join us in our Bloomsbury studio. They’re fantastic, and now we can get all the crap off the floor and every other surface!

Here’s the pre-installation shot:

Slack for iOS Upload (1).jpg

And the evening post-installation, with weird Charlie head, and Phil working hard on our new Explorer project (more on that later).

Slack for iOS Upload (2).jpg

 

It’s like it’s a real office or something.

This is a blog post by Phil Gyford, who helped me make a thing. I’ve been curious about a tool called sheetsee.js for ages, and we made a site to show you the museums we visit, and it’s driven directly from a Google spreadsheet! I find myself wondering if small museums might be able to use this simple tech.


hunterian.png
PHIL SAID:
George has been keeping track of the museums she’s been visiting in a Google spreadsheet and wanted to make it a little more visible, useful and attractive.

Sheetsee.js looked like it might be just the thing to help. It makes it relatively easy to use data from a Google Spreadsheet to make pages containing tables, maps and charts. One of the Sheetsee.js examples, Hack Spots, was along the lines of what we wanted, which was reassuring.

Within a day we’d got a quick single-page site up on Heroku, letting us list all the museums, clicking one to display a museum and its details, including a map. Pretty good! We spent a second day refining it, making nice URLs for each visit, and filtering the table to show only the museums built, founded or opened in a particular year.

The site has barely any back-end; there’s a single PHP file and a `.htaccess` file to make URLs like /visit/23 load that one file.

Sheetsee.js uses Tabletop.js to fetch the Google Spreadsheet’s data. We run through that data and tidy up each row a little: we make some fields more readable; and add fields to show (for example) whether or not that museum has any external URLs, which helps with displaying its details. Sheetsee.js then handles displayig the table, paging the data, making it sortable, and making the filter form work.

The JavaScript listens for clicks on the museum names and then displays its details. When the data includes latitude and longitude we use Leaflet to display a Mapbox map.

Making the page work more nicely — changing the URLs for each museum and keeping the browser back/forward buttons working — involved more custom coding, which got me in a bit of a tangle, given I’m not used to making single-page, solely-JavaScript-powered sites. It seems to work, thanks in part to History.js.

Displaying the museums that were only built, founded and opened in particular years also involved going round in circles a few times, and involved more URL-fixing shenanigans and manually filtering the data before handing it over to Sheetsee.js.

All-in-all Sheetsee.js was lovely to use and it would be a great tool for creating nice views of modest amounts of data held in a Google Spreadsheet, without the complication of a database. Going beyond what it can do by default can be trickier. For example, we wanted to have the table’s filter only filter based on the contents of a particular column, but that’s not easily possible. But, otherwise, two thumbs up!