At User Experience Week, August 22-25 in Washington, D.C., Eric Costello will join us for a talk on the evolution of Flickr, the wildly successful photo sharing site that has become one of the most talked-about examples of the next generation of Web applications. Adaptive Path’s Jesse James Garrett recently talked with Eric about Flickr’s past, present, and future.
Jesse James Garrett: Give us a quick preview of what you’ll be talking about in D.C.
Eric Costello: I’m going to give a short history of Flickr, from its beginnings as a massively multiplayer online game called The Game Neverending, and slowly morphing over time into the Web app for photo sharing we now call Flickr. I think this will give some insight into the reason that Flickr works the way it does.
From there I’m going to talk a little bit about how our team operates, how we divide up duties in development and design, and how that leads to a good, collaborative process. I’ll also talk about how input from users has really influenced us, and how that all feeds into our process.
And then I’m going to talk a little bit about the UI features that we’re especially proud of, how they work, and why they work well. I’m going to present some challenges that some of the advanced UI features have presented, in terms of training users to understand what’s going on, and also talk a little bit about what the future holds for Flickr.
JJG: How old is Flickr now?
EC: We launched it in February of 2004, so it’s about a year and a
half old. But it’s radically different now than it was at that time.
JJG: It’s pretty unusual for an application to change so radically in
such a short period of time. What drove this rapid evolution of Flickr?
EC: That’s an interesting question. When we first launched Flickr, it
was a Flash application that was mainly just a chat environment with
real-time photo sharing. So it was quite limited in what you could
It wasn’t a photo sharing site, so much as it was a place where
you could go to chat and talk about photos. But none of that activity
was stored in any asynchronous way – there were no Web pages that
hosted the conversations people were having about photos, it was all
We wanted to extend that into the Web, so that
everything that could be done in the Flash chat environment could be
done on the site itself. User feedback also drove a lot of the
decisions about features. We had user forums very early on and people
told us what they wanted.
JJG: Can people still use that Flash application?
EC: It’s indefinitely retired. We’re not sure when it will come back,
but we hope to bring it back. It’ll probably be in a different
incarnation than what it originally was.
JJG: Do you feel like you had to sacrifice something to make the
transition from the Flash application to a more traditional Web
EC: I think we did, but really what motivated the changes was what we
had the most success with. Although the Flash application was, I
think, a really innovative interface, and really fun to use, and a
lot of people enjoyed it, I think it was a little too off the
beaten path to really get a wide audience.
As we started adding features to the site itself, like pages that hosted the photos so that people could visit them at a unique URL, we had a lot more
success with that. People responded to it, and the site began to
grow. So our energies tended to be dedicated toward enhancing that
aspect of the site.
We kept the Flash application, which we later came to call Flickr
Live, around for a while. But eventually we took it down because
there were some security issues with it, and we felt our development
time was better spent on other things rather than fixing those. So we
sacrificed something, but we ended up with something better than what
we had before with Flickr Live. So you lose some things, you gain
JJG: Were people upset about losing Flickr Live?
EC: Yeah, there were certainly some people who weren’t very happy. In
fact, just the other day in one of our support forums, someone
noticed that in our FAQ we make mention of Flickr Live, which you
can’t get to anymore, so they reported this as a bug that we should
remove that reference.
I think the first three responses from
other Flickr members were, “Instead of removing the reference, just
bring back Flickr Live, please!” So there are still people that
remember it fondly and want us to bring it back. Someday maybe.
JJG: You mentioned that the changes in Flickr were driven by the
areas where you saw the most user interest…
EC: Yeah, user interest, and also just comprehensibility. People understand a website full of photos better than they
understand an innovative chat interface with photo sharing. Power
users got what we were doing with Flickr Live and learned to swim
pretty quickly, but people like my mom weren’t quite as quick to
figure it out.
Also, there’s a huge advantage to the asynchronous nature of Web pages, where I can leave comments on a photo for someone to find at any later date. There’s a huge advantage to that over the synchronicity of the real-time chat, where
if you’re not there to enjoy the conversation about a photo at any
one specific time, you’ve missed it altogether.
JJG: How did you know what features were connecting with your audience?
EC: We have not historically been a very metric-driven company. We do
look at numbers, but really we just keep our ears open. We listen to
what people say to us on our forums. The bottom line is, when users
sign up and actively start using the site.
We had a lot of signups when we had Flickr Live, but relatively few of those signups turned into active users. Whereas when we started introducing traditional
Web-style features, we got a lot more signups, and more of
those people stuck around and continued to use the site.
JJG: Could you see how much people were using the new features you
were rolling out, or was it just that you were hearing people talking
about the features?
EC: It was mainly conversations that we listened to. People talking
to each other about the site, people talking directly to us about the
site. We didn’t really track feature usage.
JJG: In the transition from Flash to a Web application, what new
features were driven by user demand?
EC: Tags were not in the initial version of Flickr. Stewart
Butterfield wanted to add them. He liked the way they worked on
del.icio.us, the social bookmarking application. We added very
simple tagging functionality, so you could tag your photos, and then
look at all your photos with a particular tag, or any one person’s
photos with a particular tag.
Soon thereafter, users started telling us that what was really interesting about tagging was not just how you’ve tagged your photos, but how the whole Flickr community has been tagging photos. So we started seeing a lot of requests from
users to be able to see a global view of the tagscape.
JJG: That’s interesting. I would have thought that people would be
most interested in their own little world – themselves, their friends,
their family. But what happened was that once they got hooked on the
site on that small scale, people wanted to see that big picture. They
wanted to feel like participants in this larger community.
EC: Yeah, that was definitely a surprise to us. Flickr was really
envisioned initially as an organizational tool for an individual who
has this huge collection of photos. The social network was built in
just so that you could restrict access to your photos. But what has
really taken off with Flickr is that it’s turned out to be a
great platform for sharing with the masses, and not just with your
small collection of friends.
And people certainly use it in different ways. I primarily use it to
share photos with my friends and family, and most of my photos are
restricted so that only people I’ve said are my friends and family
can view them. But we found that it took off when we got some
excellent photographers who were interested in using Flickr
as a new kind of photo blog, so that the world could see
their pictures. And that, I think, is really the primary usage of
JJG: So you almost accidentally ended up in this position where you
find yourself competing with personal publishing tools
like Blogger and LiveJournal. How much of Flickr’s evolution do you
think was driven by this kind of accidental discovery?
EC: Accidental is probably not the right term – I’d call it
“fortuitous.” There are a lot of bright people on the Flickr team who
have great ideas that have influenced our direction. But we
also have a very agile development process. We deploy code to the
site maybe 10 times a day on a busy day. And we’re constantly adding
new features, small and large, even though lately it’s been
relatively small features, sadly.
But because we’re quick to develop and deploy new things, and because we have a talkative bunch of users and a lot of places for them to talk to us, we can quickly assimilate suggestions from the community. We can build a feature and deploy it
sometimes within a week of hearing a feature request.
So it’s not accidental, but most of Flickr has not undergone a lot of
extensive planning. We’re kind of rolling with the punches, which makes it fun. And I think that makes it fun for the users, too.
JJG: How does collaboration in the design and development process
work for the Flickr team?
EC: We have occasional meetings where we all get together and throw
stuff up on the board and hash out what we think are important
features to add. When we have a good list of stuff that we want to
act on immediately, George Oates, our UI designer, will do some
mockups. She’ll bring that to us, and we’ll iterate.
Based on her work we start implementation, starting with prototypes, then
iterating and getting feedback from the whole team. Then we release
it on the site and listen closely to what people like or dislike
about that feature, and make adjustments.
It’s gotten quite a bit harder to respond and act on user feedback
because we now have such a large user base. There are enough people
with any one opinion that we have to be careful what we listen to,
because sometimes a minority opinion can be a majority of voices on
the site. Ultimately, every decision comes back to what we think is
best for the most number of users.
JJG: Do you have any formal project management tools?
EC: No, we’re not very formal! That can be kind of frustrating for me
as an engineer who has to make the UI work. But it’s also what I
thrive on, which is solving UI problems. So although we rarely have
everything spec-ed out adequately, it’s part of the fun of doing the development, meeting challenges that arise as we start to actually build the UIs we’ve talked about.
JJG: You mentioned del.icio.us earlier as the inspiration for
Flickr’s use of tags. What are some of the other inspirations for the
interface and the user experience of Flickr?
EC: Before Stewart and I started working together on The Game
Neverending, he had done the 5K, a contest for Web designers to see
what they could do within the constraint of a 5K total file size.
That was inspirational to a lot of people on the Web. For the second
year of the competition, I helped him build the site. We wanted to
allow a popular vote for the contest in addition to the judging.
Netflix had this very innovative little interface where you could
rate movies on their site that was all done client-side. You clicked
on one of five stars to indicate your rating, and it talked back to
the server without reloading the page. We wanted to do the same thing
for the 5K site. We implemented that in kind of a hacky way, because
that was all you could do at the time in a cross-browser way.
So that sort of UI with immediate feedback, not requiring a page
refresh—which of course you later came to call Ajax—was
something we’d been playing around with since well before Flickr. So
it was natural as we started to build the Flickr site that we would
take that a little bit further. I think one of the first things we
did was make it so you could edit titles and descriptions on photo
pages by just clicking on the text and, through the magic of Ajax, no
page refresh, but your data has changed. So Netflix is an indirect
The Game Neverending
JJG: Tell us a little bit about The Game Neverending, because I think
there are a lot of people who aren’t familiar with that aspect of
EC: With The Game Neverending, we hoped to build a massively
multiplayer online game that was totally Web-based. You could play
the game from a browser wherever you were. It was a Flash app that
talked to a Java application server that we built. It’s still a great
idea. Someone still needs to do it.
It wasn’t an immersive environment at all. It had interfaces that
were really like Web interfaces or desktop application interfaces.
The mode of interaction between users was in IM windows. The way
Stewart talked about it early on was giving people an excuse to use
chat and IM. There are a lot of people using these tools but there
are a lot more who aren’t used to the idea of online chat, who think
of it as something kind of geek and weird.
The inspiration in large part for Game Neverending was actually
Neopets, which had tremendous success among young people as a way to
interact with others around the idea of this magical world of pets
that you cared for, and all sorts of things you could collect. The Game
Neverending was the same basic idea, but more real-time.
We did a couple of things in the UI that were kind of neat, I think.
You had IM windows where you could drag a person from your contacts
list into any chat window and it would invite them to join your
conversation. You could also drag game objects into an IM
conversation and it would send to all the other members of the chat
an image of the object. So it was a way that you could share the
things you found in this world with the people around you.
That feature was where the idea for Flickr came from. We thought,
what if instead of game objects, you could drag and drop other
digital objects into these conversations, like Word documents, or
PDFs? Photos were the natural thing to go with because they’re more
When we initially launched Flickr, it was just a stripped-down Game
Neverending interface, with photos instead of game objects. You had a
list we called the Shoebox at the bottom of the interface with all
the photos you had uploaded, and you could drag them to other people
to share them. You could drag them to an IM conversation too. That
was all straight out of The Game Neverending.
JJG: How much of The Game Neverending would you say is still present
in Flickr in its current state?
EC: I think the spirit of it is there, definitely. Someone once
described Flickr as “massively multiplayer online photo sharing.” I
think that’s a good description. There’s kind of a feeling of
exploration within Flickr. It feels like a world where you can move
around and find wonderful things – the wonderful things being the
great photographs that people upload.
And because it’s got the social network aspect of it, you can kind of
build neighborhoods within Flickr. The page in Flickr that shows you
all the photos from your friends and family is very much a space like
you might find in a game. It’s a place where you go and interact with
the people you know.
With the absence of Flickr Live, and with as many changes as have
gone into Flickr since that time, there’s not a lot of actual
functionality that’s the same as the game anymore. But certainly the
game’s there at its core.
JJG: What led to the decision to retire the game?
EC: Really, it was the success of Flickr. Flickr was kind of a side
project. We had tremendous success with it, and it kind of
overshadowed the game. We didn’t have enough man-hours to work on both.
The Future of Flickr
JJG: You mentioned that using Flickr feels like play to people. Would
you say that’s a deliberate part of the design, or an artifact of its
EC: I’d say it’s pretty deliberate. Our team, led by Stewart, is very
playful. We are always having a good time, whatever we’re doing. We
let that carry through into the way we describe things on the site,
the way we talk to our users. It’s all very playful, and that’s
intentional. We want it to be a playful place.
JJG: What do you think the difference is between Flickr and older,
more established photo-sharing sites?
EC: There are a couple of differences. With most of the other photo
sharing sites, they seem to be mainly about facilitating the ordering
of prints. Which is a fine goal, and soon enough in Flickr you’ll be
able to order prints, which is a much-requested feature from our
users. But with Flickr, it’s much more about doing what you want with
your photos. What we’ve found is that people really want to share
them with a wide audience.
And I think that’s the key difference between people who love Flickr
and people who love those other sites. People who come to Flickr want
an audience. We’re all about facilitating sharing however you want,
whereas the other sites are more about uploading your photos to a
place where you can easily print them. They’re not as much about
exploring and sharing with the masses.
JJG: How do you see the acquisition of Flickr by Yahoo! influencing
the direction of the site?
EC: One reason we were so excited about the Yahoo! acquisition is that
we had a number of places we wanted to go with the Flickr site, and
what we heard from Yahoo! was that they wanted the same things. So our
interests are aligned, and it’s wonderful to have Yahoo helping us
make reality all of the things that would have been pretty hard to do
as a small company.
JJG: How do you see Flickr evolving in the future?
EC: We’re going to be releasing some really cool features pretty
soon. We’re working on new features to increase the relevancy of tag
search results. It’s quite exciting because there are a lot of
fantastic photos on Flickr that people don’t always find. I think
it’s going to change the nature of Flickr a little bit. It’s pretty
JJG: Thanks Eric, and we look forward to hearing what you have to say
EC: My pleasure.