Earlier this year, the Wikimedia Foundation asked designer George Oates, who has worked for Flickr and the Internet Archive‘s Open Library, among others, to conduct a deep dive into Wikimedia Commons, the free media repository that provides many of the images used on Wikipedia. We wanted a fresh pair of eyeballs, and we were particularly interested in Oates’ observations about possible areas of improvement.
Oates soon realized navigating Wikimedia Commons and finding interesting materials is challenging. She noted how the category system used to organize and tag media files on Commons is confusing and hides—not shows—the richness of content there. Other areas of improvement she mentioned include making both contributors and contributions more prominent and inviting, improving the user experience for new users, and actively recruiting more diverse gender participation in the project. (Oates noted that the group of administrators on Wikimedia Commons is internationally diverse, but not so much in terms of gender; this affects, for instance, the selection of featured images.)
Oates also identified several ways in which Wikimedia Commons can work more fruitfully with GLAMs (Galleries, Libraries, Archives and Museums). This is partially due to her background in the field: over the last decade, she as worked on a series of projects that have reimagined how the world engages with the collections of libraries, archives, and museums on the web.
She created the Commons on Flickr, which to this day stands as one of the most popular shared platforms for the world’s visual archives. She worked with the Internet Archive to lead the Open Library, a universal and openly editable catalog of the world’s books, connected to a digital lending library. More recently, she has worked with a variety of individual institutions to devise new ways of exploring their collections, like this experimental interface to the Wellcome Library collection in the UK.
As we contemplate the future of cultural partnerships and the sharing of open collections in the Wikimedia movement, the Foundation asked George to do a short assignment: look at Wikimedia Commons through the lens of your background and experience, tell us candidly what you see, and suggest areas for improvement. We asked her to focus on the experience of new users, of cultural institutions, and to share with us some ideas of how Commons might evolve in the years ahead. Ben Vershbow (Director of Community Programs) and Sandra Fauconnier (Program Officer for GLAM and Structured Data) sat down with George to speak with her about her findings.
Ben: Please tell us about yourself!
George: I’ve worked on the web since 1996, mostly in design-centered roles. I designed a lot of the user interface of the photosharing service, Flickr, for the first few years, wrote its Community Guidelines (with Heather Champ), and particularly enjoyed creating the privacy and safety tools that helped people coexist. Since 2008, I’ve specialised in the cultural heritage sector, and today, I’m based in London, building an edtech company called Museum in a Box.
Sandra: What were your first impressions of Wikimedia Commons, looking at it as a ‘well-informed newbie’ and with your background of having worked on Flickr, and Flickr Commons?
Turns out, Commons is huge, dense, variegated, and alive.
It’s a system that’s about the size of Hong Kong (7.3 million people), but it’s only about 14 years old. Hong Kong has been settled for centuries. Just like the city, Commons is full of nooks and crannies, mazes, rules, signs, and people doing their thing.
Here’s a map of Hong Kong drawn using locations where photos were taken. “Blue pictures are by locals. Red pictures are by tourists. Yellow pictures might be by either.”
For me, this Locals and Tourists project symbolises the way people move around software systems. Software can be learned like a place. You can move around it how you see fit, and inhabit corners you like. You can see other people in it, and follow, copy, like, ignore, and meet them.
This sort of socialising and movement happens on Commons too, but there’s an important difference between Commons, and, say, Facebook, and that’s that the locals are building the city. The ‘old school’ admins know the back alleys and the best noodles, and the Commons newbies head for the tourist hotspots.
So, in terms of my first impressions, I faced three primary challenges to get to a position where I could present ideas, not just observe:
- I am a newbie to the platform, unfamiliar with its rules, rituals and traditions
- There’s a major usability challenge: Information vs Interface
- The rules of engagement are not clear
But, it also feels important to say that maybe there’s nothing wrong with it. The question in itself reveals a quiet tension. The Wikimedia Foundation’s team who support the Commons is in a hard, interesting position. They’re a crew of professional, philosophically aligned staff who find it difficult to help improve Commons. They meet resistance from the old guard, the community of volunteer admins and other long-term Commons folk who are there every day, have probably been there longer than anyone who works at the Foundation, feel deep ownership of the system, and can exercise deep control. They’ve patiently and visibly negotiated the dynamics of the system for the last fourteen years.
So, the question becomes what is it OK for the Wikimedia Foundation to do?
Ben: Yes, you’ve put your finger on a very delicate aspect of our work. Fortunately, we are finding ways to channel great ideas from the community into well-defined efforts to improve the platform. The prime example right now is our work to integrate Commons more tightly with Wikidata to improve discovery and to offer contributors and partners better ways of describing and organizing content (see here and here).
Before we go deeper into your impressions of Commons, can you say a few words about your design philosophy and how you applied it during this assignment?
George: There are two main themes in my work that come to bear here:
Challenging the dominance of search as primary digital cultural interface
Many of the web projects I’ve made recently focus on explorability of gigantic cultural collections. I use the prompt “what happens if there’s no search box?” to help you find things. How can you show the shape and contour of a collection to allow people to follow their nose and sniff out the things that interest them?
Importance of copywriting in interface design
Wiki systems are always a battle of Information vs Interface. The more functional pages you need to use the thing also take the time to explain and define themselves. It can be so informative it’s overwhelming.
Wikis are also written by many. At Flickr, I wrote a lot of the interface copy; the “system voice” was largely mine, and therefore, consistent. On Commons, you don’t know who is telling you what. This opacity can be confusing if you’re expected to follow certain rules, but don’t know who’s setting them.
Ben: So what did you learn as you delved deeper into Commons?
George: It’s huge and arcane. And successful. “Anyone can contribute” means that all sorts do, and there are now about 50 million things there. That’s astonishing.
But, there’s very little sense of the scale or depth of the thing. It’s one thing to say there are all those things that anyone can use, but it’s another to show them all. Newbie explorers are confronted with a vast textual category system including an obscure interface element to show contents of each category… can you tell what it means?
Categories are a beautiful disaster
The various Wikipedias work so beautifully because they embrace the networked nature of information. Pages are stronger (and accepted) when they interconnect across the system. Wikipedians joke about getting “lost down the rabbit hole”, where you’re at one place at one minute then you look up an hour later and you’re somewhere completely different. This is because all of the entries are deliberately and delicately interconnected.
Commons is not like that. Even though there are about 6 million “multi-hierarchichal” categories, the majority of files are only in one category, as this 2017 research shows:
If you select Images at the top of the homepage, you see a list of categories of types of image, not types of things in images. That’s tricky if you’re looking for a photo of a flower and not Images by Resolution. There are also Topics on the homepage, which operate a little more like you’d expect. You can click on Mathematics, then explore Chirality in Mathematics or Sets of Mathematical Objects. I was able to find what might be the best category in the history of categories though: Potato truncated from cube to predicate lattice. Does that sound like a category to you?
Showing the actors
I was surprised the thousands of people participating aren’t more obvious. Who’s editing what? Which people like which categories? Who is uploading a lot? Who’s helping other people? I became interested in who the admins were, how/when rules were established, and how decisions are made. If we think in terms of locals and tourists, who are the actors who know what’s happening, who I could ask for advice, or imitate?
When you get to a media file, like this fabulous The approach of the spirits, there’s nowhere to jump across to; no way to see images like the one you’ve just discovered. Its categories are listed as PD-Art (PD-US-1923) and Illustrations. Nothing like myth, stories, night, spirits, stars, or the night sky.
Not being able to move sideways is a huge weakness of any hierarchical structure. It can also be difficult to navigate a hierarchy unless you know its whole structure. You’re constantly going into a branch then back out again, instead of skipping across an interconnected collection. That said, the new work on interconnected Commons infoboxes looks really promising, e.g. Category:St. Paul’s Cathedral.
There’s no limit to how many categories a digital object could belong to. How could the system be improved to allow more and different descriptions? As with many traditional cultural collections, an object’s metadata is revisited rarely, so instead of a live database of media like Flickr, where objects enjoy a very social existence, a thinly described object in a sea of 50 million is practically invisible.
Rules of engagement
“At the start of my wiki-journey, I was left to fend for myself. It felt akin to being surrounded by hungry beasts and wondering why you’re still alive…” –Pyfan
While it’s also easy or almost trite to say that Wikis are transparent democracies where anyone can participate, I also read quite a lot that people feel really intimidated when they first “come to the party”. The category structure might also have evolved into what’s there today because the rules of engagement are dense, and the interface when you try to add or enhance things is obtuse.
As I did my research, I asked a few questions in the IRC channel, or on the Village Pump (which is really well run, by the way). Answers would be presented, and be useful, but, they referenced pages that I bet would have taken me forever to find, if I could find them at all. They often displayed old “votes” or decisions made by a very few, and not revisited, or with no recourse to revisit.
Sandra: Do you have any tips and direct feedback for the Wikimedia Commons community?
Like I said, I remain interested in the administration of the thing. In this self-governing system, how are rules created? How do those rules percolate to the newbies? How visible are they? And most importantly, who is making them?
There are about 225 admins for 7.2 million accounts on Commons. It’s a media library built by 7 million, but not all are editing and adding things all the time. In fact, the Wikimedia Foundation suggests about 32,000 folks a month are active, but even so, the system is administered by just 0.00003% of the population.
Diversity in administration / representation / arbitration
I looked into who the Commons admins are, with the data to hand. Here’s a graph that shows how many admins were brought on each year, and how many from each year “survived” as admins.
I also looked at each admin’s page, and noted whenever I could see a declared gender (using names or photos etc). Here’s the breakdown of admin community demographics:
The geographical and language spread of the admins seems pretty good. There are people from different countries, who often also participate in their language Wikipedia and/or the EN Wikipedia, and sometimes even work for, or have worked for, the Foundation. But, even in a best case scenario (where the “unspecified” category from the pie chart was actually about 50/50 women and men, that would still be 66% men and 33% women as declared administrators. There also aren’t many new folks. This might be because the rules of engagement take a while to take in and operate at a sophisticated level. Perhaps it’s also a place to inject diversity, particularly around the things that require some kind of vote or consensus.
Also interesting to note at this point that Hong Kong has a police force of 34,000.
The Geena Davis Research Institute was founded in 2004 to study and shift the balance of representation of women and girls in media. Their tagline is “if she can see it, she can be it,” and the institute has done groundbreaking research about how to measure the presence of women and men in films, mostly, to demonstrate the presence of unconscious bias in our media. The stats are telling.
One of the very simple ways the institute suggests to check when you’re making a movie is that all the crowd scenes contain 50/50 women and men. I really like this. It’s easy to check and simple to do. If we apply this idea to Commons, one place to look for an “equal crowd” is the various Highlights areas, like Photo of the Day, or Meet our photographers. Today, those lists are mostly of men, and even some of the older admins.
Those public lists are a place to look for fostering more diverse representation in the community. How could the procedure for creating those lists be realigned with a positive and welcomed diversity agenda?
During my research, I was shocked to see a photograph of a naked woman show up on the homepage of Commons, marked as Photo of the Day. I went to the Help Desk to ask “Why is OK to have a picture of a nude woman on the homepage?” I wasn’t crying OMG P0RN!, but suggesting that the image may be an alienating first image for some to encounter. It’s worth having a read (and a mark of the strength of a wiki’s nature that I can retrieve the conversation). For me, it was telling that the rebuttals to my query ranged from tired responses around “pornography or art”, or that “galleries are full of naked women”, to more more resigned responses like “it’s already been decided”, or, “But this is the Internet, where the opinions of Western young men dominate, so good luck getting any sort of grown-up discussion about that.”
The fact that there the system is full of old decisions that are no longer questioned is a problem. You can see how it’s now handed out as a “it’s the way we’ve always done it” response here, as another helpful person chipped in with support for my query. You can also see the original voting process in action, in November 2013, which pushed the photo of the naked woman into the picture of the day category. (Warning: you’ll also see the naked woman on that page.) Of the 18 votes cast, at least 17 of them were by men, 3 of whom were admins, many of whom have been on Commons 5+ years, and a handful of whom are also featured photographers.
I realise that this file and its life on Commons may be an easy target, but, it’s also indicative of the oligarchal participation and representation on Commons.
What if featured photographs votes require a more representative voter group? What if you, as a user, could only have 10 featured photographs? What if the “Featured” lists were removed entirely?
Sandra: Do you have any suggestions on how to improve the design of Wikimedia Commons in general, to make it easier to use?
George: If we’re talking about making it easier to use for individuals using the web interface, then yes, I do.
There are just five user pages and one email in the UI that could be improved to help new folks get grounded. Making them crisp and instructional instead of information vs interface would have a huge effect. It’s about decluttering the copy and trying to help the new person figure out what to do first. The pages are User, (and welcome email), User talk, Watchlist, User contributions, and File list.
For example, here’s the current Welcome email, the first point of contact from the Commons to an individual:
Hi there [Username],
Welcome to Wikimedia Commons! Someone (probably you) from IP address 22.214.171.124, has registered an account “[Username]” with your email address here. We’re glad you decided to join us.
What next? First you should confirm your account.
To confirm that this account really does belong to you and activate email features on Wikimedia Commons, open this link in your browser:
After that, you’ll see ideas on how to get started and links to help you learn about Wikimedia Commons.
From all of us here at Wikimedia Commons, welcome aboard!
If you did *not* register the account, follow this link to cancel the email address confirmation:
This confirmation code will expire at 13:53, 13 June 2018.
This asks more questions than it answers, and even casts doubt on me being a legitimate user! “Someone (probably you) / “to confirm this account really does belong to you…” come across as defensive and doubtful. Hardly a friendly welcome for someone new.
Have a look at the difference if you make it more concise:
Welcome to Wikimedia Commons, the world’s free media library!
Your first step is to confirm your email address, please:
Once that’s done, there are a few ways we recommend getting started:
• Read the Policies and Guidelines of contributing to Wikimedia Commons
• Take a Photo Challenge or Upload your first media file
• Help improve information about Commons resources
• Have a look around your account settings
Good luck! And if you get stuck, please visit the Community Help Centre.
It’s also worth noting that I have absolutely no idea how I—an interested interface designer—could independently contribute effort towards making this part of the system better, just like I can edit regular entries.
Seeing each other
One of the reasons I think Flickr and other social systems are so successful is because we can see each other when we use them. They show people their own stuff, other people’s stuff, allow them to follow/connect/gather, show the activity happening—particularly on your stuff because we all love that—and often, give a short list of things to try next, to show the way through. We learn from the way other people act and operate in the physical world, and that’s the same on a software platform.
The foundation is there on Commons to enhance representations of scale, activity, and actors to help people see each other more clearly. Improving those five core screens with these themes in mind would help locate new people in the system, and give them only as much information as they need, when they need it.
Ben: And what about suggestions on how to invite more and better contributions to Commons?
George: I would try to make it much more obvious that viewers can improve metadata; provide much clearer calls to action to interested people. I mean, it’s somewhat implied, given that it’s a Wikimedia system, but, as far as I can tell, the call to action to improve the metadata about a file is a small Edit link next to the Summary heading on the item’s page.
Here’s the Pied-winged swallow, Picture of the Day on 3 September:
If you have the gumption to click “Edit”, you get this:
So, even if you wanted to add information about the image — e.g. it’s taken outside, there’s blue sky, it’s a bird, etc — you have to figure out the editing code/pseudo code and its UI. I tried to add “tag=bird” for example, and it threw an error. It’s very different from having a UI designed specifically to encourage contributions and conversation. It’s not simple. (I realise that “special” images, like Picture of the Day, have protected edit capacity to prevent vandalism.)
There’s a ton of uploading that happens to Commons that’s not through the UI, but with lots of other tools written to streamline the process. (That’s another strength of the system, and was at Flickr for that matter—that people can contribute to it through all sorts of interfaces.) I found myself wondering if another way to improve contributions to Commons might be to focus on the connective tissue between the various Wikimedia platforms, in the UI. If I upload an image to Commons, I could be actively prompted to create or enhance a Wikipedia page…
Sandra: Do you have suggestions on how to make Wikimedia Commons a more interesting and useful platform for external organizations (e.g. GLAMs) to work with our communities and make their knowledge available?
George: I would look at three main areas around usefulness for institutions:
Demonstrating use of collections, especially in a digital context, is especially difficult for institutions. If their stuff is nestled cosily amongst about 50 million other things, how and when can they know if the effort they’ve put in to sharing their treasures in the Commons is effective? Can any use of materials be reported on? Can that be established as new, legitimate usage of an institution’s materials? How can it be more refined, or indeed, more accurate than the mysterious “page view”. How can Commons help institutions (and individuals) see when their stuff is used? Something like the old school Dopplr Annual Report, perhaps?
Presentation and presence
As I understand it, institutions either create a category or use a totally different system to see all their media in one place. Can it provide a good destination for institutions to share around? What could their “User” page be like?
If you look at the Wikipedia page for a contributing institution, e.g. Nationalmuseum Stockholm, you see this little box at the bottom:
Even just rephrasing that to say “Explore the Nationalmuseum Stockholm’s 2,400 contributions to Wikimedia Commons” or perhaps creating another box that does that might be a start. That link also takes me to Category:Nationalmuseum Stockholm, but it’s not immediately clear which items are from the collection versus about the institution.
There’s obviously a question about whether institutions should be treated differently in the Commons context. I’m not sure about that. Perhaps it would be better to look for a way to enhance anyone’s presence equally. If part of the claimed benefit of participating in Commons is that metadata may be improved, visibility into the material also needs to be improved. What about a link on the home page to a list that shows all of the institutions?
Supporting people doing great stuff
When I was doing research, I attended a Wiki-a-thon at the Wellcome Collection in London. It was a day dedicated to improving the presence of women involved in medicine on Wikipedia. There were about eight people in the room, including the Wellcome Wikimedian in Residence, Alice. It was brilliant, and I could not have created my article on Clara Stone without Alice in the room. The simple act of having a living human who could explain the rules and secret doors (like, you need to make 10 edits before you can make a page) instantly made the whole process more approachable.
Human Wikimedian presence in cultural institutions can really help bridge the gap that’s sometimes perceived between authority and the crowd, which can be difficult for institutions to conquer on their own. This mediation is hugely valuable, and something that the Foundation could continue to support directly, and more. What if there was a Wikimedian in Residence at every national library in the world? (And lots of the smaller ones too?!?)
I was also surprised to discover that lots of the updates to Commons are done en masse, through scripts and other programmes developed by volunteers. What sorts of other support could the Foundation provide for these folks? How can it support this developer community even more than it does now? What new efforts and resources can be amplified to reach more potential users (both outside and within GLAMs)?
Ben: Any last thoughts on what a community-driven media repository like Wikimedia Commons could become in the future?
George: I was a bit surprised that this work turned out to be about representation in the end. Yes, there are simple interface changes that could be made to continue to improve usability, but there’s a huge challenge for Commons (and other huge collections online like Europeana and DPLA) to improve description of their millions of things so they’re easier to see.
These vast online collections will eventually see a return of the power and delicacy of curation. Only machines can consume 50 million things, and even then they might not be sure exactly what those things are, and if they find something interesting they won’t know who to tell. Computers “scan everything and hear nothing,” so we need all kinds of humans from all over the world to help gather our histories into meaningful units, but right now, Commons is a pretty closed system.
Imagine if Commons could become like one of the great old cities of the world, full of all kinds of people from all kinds of places instead of a big city run by a tiny group of people who basically look the same?
Interview by Ben Vershbow, Director, Community Programs
Sandra Fauconnier, Program Officer, Community Programs
 The admin data consists of two views: (1) The list on the right side of the Commons:Administrators page, and (2) those users listed on the Commons:Administrators/Archive/Successful requests for adminship, which I edited manually to separate users who’d ever been admins from those who remain admins. Note that the total count of admins in those two views differs: (1) 225 admins, and (2) 472. I decided (2) is the more useful, but it should be stated that that’s a manually maintained list, and possibly not accurate.