Inside a New Museum (Part 1): Digging Into Data

George Oates
M+ Labs
Published in
6 min readAug 6, 2019


In part one of ‘Inside a New Museum’, designer George Oates writes about her two-week data residency at M+ along with developer Dan Catt. Read part two here.

Lara Day and I first met in 2015 at Museums and the Web Asia, just after she’d joined the M+ crew as the founding member of the museum’s digital team. It was lovely to hear from her again when she was visiting London a few months back, and we got together to discuss me doing a residency at M+ through my agency Good, Form & Spectacle (G,F&S). We decided on a two-week placement in the M+ offices, in the fantastically named Cyberport in Hong Kong. I was also joined by Dan Catt, a lovely engineer colleague and friend from our days together at Flickr.

The M+ digital team — which has since grown in size — are making all the right moves as they look towards the opening of the new museum building in Hong Kong: working in public, releasing their catalogue metadata under a CC0 license, providing a public API to access it, running hackathons, making great online content, and generally being birds of a feather with other museums around the world doing forward-thinking digital work. Our residency with the team culminated in Dan and I coaching attendees of the museum’s second Hackathon, celebrating the second version of the public API and more data therein.

G,F&S specialises in these sorts of commissions, where we are brought into a cultural institution to investigate its metadata quickly, productively, and roughly. We are interested in the formation of metadata to describe collections, and in representing that knowledge visually and interactively. Through this work, we’ve found that new insights into the structure and organisation of a museum’s collection becomes possible. We’re able to show the collection as an interactive whole, instead of it being locked up inside a hard software interface that expects either record-by-record enquiry, or a report-style interrogation — neither of which allows you to wander through the collection, and discover its shape. The basic outcome of this work is called a spelunker, a functional but rough web offering to allow exploration and wandering, and hopefully seeing the metadata with fresh eyes.

A New Museum for the World!

M+ is a rare bird. It was created in 2012, has a growing collection of twentieth and twenty-first century visual culture, and is currently building its new home on Victoria Harbour in the heart of the West Kowloon Cultural District. It will literally be a bright beacon on the harbour.

This situation presented some early questions for me about how a new, contemporary museum could shape itself: what twenty-first century practices could the museum be already adopting to ensure a strong digital foundation? Is there an opportunity to realise a new model of collection — one that specifically cultivates digital distribution of contemporary visual culture?

Expectations are high for this slick but young museum. While I was there, I heard talk of M+ being compared with MoMA in its scope and scale. (For comparison, MoMA was founded in 1929, while M+ has been operating as a museum since 2012.) How might digital tools and resources help its work go smoother as it heads towards the launch of its building?

How We Work

At G,F&S, we have honed a few simple techniques to make these short-burst appointments as productive as possible:

  1. Fast software work
    Much more can be done at a cultural organisation when you have software builders under its roof. We like to demonstrate the possible savings of developing digital building chops — just like you would curatorial, finance, content, or educational.
  2. Sketch, don’t codify
    We swerve in to museums from software-land, where we’ve learned to move so fast that it’s hard to keep up. We make progress by sketching and showing and asking questions. Consensus is not a requirement.
  3. Show everything
    It’s often near-impossible to use one of the centralised software tools you find in museums to see the whole collection. We combat this by representing collections as a whole; a landscape. As soon as you do this, people can point at things, and see where the gaps are.

During our ten-day engagement, we started each morning by sketching out our daily plan, and aimed to produce whatever we’d sketched digitally that same day. We also gave two internal presentations to staff, and kept a Work Log in Keynote. Here are some of our sketches:

Daily design sketches.

And here’s a snapshot of the Work Log in Keynote:

Capturing our work as it happens.

Things We Made

‘DNA’ to Show the Whole Catalogue

On the first day, we made visualisations of the whole catalogue — about 11,000 records. (There are about 5,000 of those that have been prepared and made available in the public API. This dataset will continue to grow.)

In the below visualisation, one row represents one record and a column represents a field in the data. We colour the visualisation black when there is data in a field; white means that there’s no data in that field.

Zooming in, you start to see different patterns:

Flecks of metadata across the records.
Some parts of the catalogue are very regular. One staff member reckoned that this block represented her start at the museum, when she started standardising the data required for new acquisitions.
You can see this block has a very different ‘completion pattern’ compared to the top one. Why?

We printed it out and stuck it on a wall in the office. It was exciting to see people wander up, have a look, and discuss their impressions and reactions.

Printing things out is fun and useful! (Here, we coloured the records in the public API in green.)

Work to Do, Work in Progress

One of the useful results of this sort of visualisation is that is shows, objectively, where there might be gaps in the metadata. Now that we’ve worked with different museum catalogues around the world, we know that these gaps are the status quo. It’s great when staff members who see this work use it to direct their attention, in particular to fix an error or omission they’re able to see for the first time.

DNA at the Object Level

In the second week of the residency, we had been refining the spelunker to streamline the catalogue exploration, and thought it might be interesting to break the giant DNA graphic up and show it at the object level. That way, for any record in the database, you could glance at its metadata to see the gaps.

Example object-level DNA for a small set of records.

You can see that we’ve also used colour to quickly indicate usage rights for an object’s digital image/reproduction. In the above example, M+ staff can use reproductions for all their needs, indicated by the green blocks. You can see in the example below that the artist(s) have said that their works can be shown at exhibitions (the green blocks), but would like further discussion for any other use (the orange blocks).

Image rights glanceable settings.

I’m chuffed that we were able to make this existing G,F&S ‘DNA’ technique we’ve been developing more useful. (You can also see it in action in the What’s In The Library project with the Wellcome Trust.) It’s starting to feel like a tool, and I’m looking for a way to make a web-based service that can do this for any collection.

These are the basics of what we did during our M+ residency. Read part two to continue following our journey with the museum, including a look at how contemporary museums must navigate copyright issues when releasing their data.



George Oates
M+ Labs

Designer interested in what Museum means. Loitering with intent. Director of @goodformand, CEO of @_museuminabox, thinking out loud at @thesmallmuseum.