Every project takes time and energy both to create and maintain over time. As we evaluate we consider several factors: institutional goals, comparative engagement metrics across many projects, and a careful look at what’s going on within any given offering.

As of today, we are retiring the Brooklyn Museum Posse along with our tagging games, Tag! You’re It and Freeze Tag. The decision to pull these activities was difficult because we fully believe in how important tagging is to the health of our collection online. After all, one person’s “landscape” may be another person’s “tree,” and all of these terms help make our objects discoverable online. As invested as we were in the program…

• Engagement within the games and posse has been incredibly low in numbers, but high in yield.  Launched in 2008, our Posse only numbered 1100 users over all these years. While 1100 users many not seem like many, collectively their contribution was quite a lot given Posse members had contributed 230,186 tags. That’s a lot of SEO in our collection online and represents an incredible effort that cannot be ignored.
• Posse members were using both games and object pages to tag objects. In fact, more than half of the tags were being delivered directly via the object pages showing the games were not necessarily a more compelling option.
• Tagging has shifted to a more social language, not a descriptive one. For as much as we want the keywords, the notion of tags as keywords has changed considerably. We need to change along with our audience and recognize that our games are outdated conceptually.
• Over the years, tagging has decreased substantially.  Within the games, tagging contribution peaked in 2009 with 32,409 tags, but by 2014 we were logging 8,089 (ytd).

When we started seeing the above, we began asking ourselves who we were engaging.  If our institutional mission centers around community with the aim to engage a broad audience, are the Posse and our tagging games doing that effectively? No…

• I’ll preface this one by saying “we love you, we really do.”  Our core taggers were likely the very niche audience reading this blog.  When we looked carefully at all the bio statements people were giving us when they signed up for a Posse account, it was incredibly clear these activities were engaging museum professionals and museum studies students with a smattering of art history students. 30% of the tags were coming from Brooklyn Museum staff who used the games to contribute their own tags (and were encouraged to do this through accounts so we could track internal participation), 22% were coming from museum professionals, and 8% were coming from other accounts not identified as one of the other two buckets.
• Staff were the most consistently engaged with an average activity rate of 338 days.  Museum professionals were with us an average of 102 days.  Other accounts held on for an average of 97 days.  Even though the metrics between Museum professionals and other accounts were roughly similar, when we compare that with the tagging percentages it shows us that the people fitting into the “other” account bucket—the core audience we were hoping to engage—were far less engaged given the numbers of tags they were contributing.

Hiroko Okada (Japanese, born 1970). Future Plan #2, 2003. Chromogenic photograph, 54 13/16 x 35 1/8 in. (139.2 x 89.2 cm). Brooklyn Museum, Gift of the artist and Robert A. Levinson Fund, 2008.25. © Hiroko Okada

So, we faced a bit of a conundrum. We know tagging is incredibly valuable, but our statistics were showing that we had a small audience for it and, in addition, that audience was more one of insiders than the general public.  If tagging is meant to democratize collections by applying everyday words instead of specialized ones, you have to wonder how much traction we were getting if the majority of tags were coming from specialized voices. That insider aspect is pretty interesting…

• In looking at our analytics we can see the majority of the search terms for the collection are actually specific. People are looking for artists, movements, types, specific cultures and fewer people seem to be searching by general themes. (It’s important to note that we are not looking in depth at the analytics, just glancing for trends.) The biggest exception to this in the last year was 169 searches on the term “seaweed,” likely owing to this tumblr post.
• Analytics is also showing users browsing by tags and this represents 4% of our collection traffic. At a glance, the majority of these terms are specific (egypt, tissot, sculpture, painting, watercolor, amarna) and fewer are thematic (woman, erotic, mask, glass, bird).
• In presentations, I’ve always cited Hiroko Okada’s Future Plan #2 as a key example to the benefits of tagging. The object metadata tells us practically nothing about this image.  If you didn’t know the name of the artist or the title of the work, you’d likely never find it.  Tagging allows the image to be now be searched on the term “pregnant.” This is clearly an isolated example, but when only two searches cropped up “pregnant” in the last year it makes you take a close look at where time and energy is being spent.

At this point it was pretty clear that tagging wasn’t working on many levels, but why not keep these activities around in the hopes that some data is better than none? Well, tagging isn’t gone from our site totally and you can still add and delete tags from any object page.  What’s gone is the technical overhead that is required for signing in, creating a profile that attributes your tags to your identity, and the games. We decided we needed to eliminate the games because we have to allocate the limited resources of our staff carefully. We simply had to acknowledge this was not working well enough to keep the staff time going.

This was not a decision we took lightly especially given this a program that we hold dearly and are known for; it took us months of wrangling before concluding this was the route. The path, however, comes with the learning there’s a better way for our community to contribute to our web presence and this is something you’ll be hearing about very soon.

This is an interactive tagging scavenger hunt that (we hope) motivates people to find and look at the work in the gallery. In addition to the find and look idea, there are a few other internal motivations that we’ve incorporated into the design.

Roam! One of our institutional aims is to get visitors looking across collections and that’s always a challenge in this very large building.  To encourage players to cross boundaries in the building, they gain more points if they tag objects on different floors.

Crossover! One of the big issues we’ve seen with BklynMuse is that it’s chock full of information and various paths to take and that can be an overwhelming amount of choice.  The recent simplifications are going to help, but we want to implement different ways to get people into that content.  As players use Gallery Tag!, there are links that crossover into BklynMuse.

Convergence!  All of the tags created go right back into the online collection, bridging the physical and virtual.

We’re going to do a launch meetup at this month’s Target First Saturday and I’ll post more details on that next week!  In the meantime, if you are coming here with your device hit m.brooklynmuseum.org to get started!

We designed our system to accept tags from users who might not want an account and that’s been both valuable and a bit of trouble.  On one hand, the 7,657 tags by anonymous contribution are nothing to sneeze at, but we’ve had to keep a close eye on those submissions and have deleted roughly 6% of them due to complete inaccuracy.  We could eliminate the capability to add tags anonymously, but 94% of those contributions are of great value and, more importantly we want our online collection to be welcoming to anyone with or without an account.  That said, there are plenty of people testing us just for fun and when the tags “how long will it take you to delete this tag” and “are you going to block me” showed up on the scene, there was only so long it was going to take an overworked Technology department to do something about it.  We knew the Brooklyn Museum Posse would have a lot to do with the solution.

The hunt for the Governor gang of bushrangers. A posse of mounted police, aboriginal trackers and district volunteers. Jimmy & Joe Governor were sighted at Stewarts Brook on 12 September 1900 – Stewarts Brook, NSW / by A C Jackson (via State Library of New South Wales on the Flickr Commons).

Today we are introducing a new game called Freeze Tag! which puts control of the tags back into the hands of our most valued community members.  If you are a member of our Posse, you can delete tags from object pages−this is new, previously we were not allowing tag deletion except by system admins. For any tag that is deleted, it takes another two pairs of Posse eyes to “agree” within Freeze Tag! before that tag’s fate is sealed.  On the other hand, if three Posse members within the game think the tag should be saved, it will be restored.  After a short stint on the live site, all tags created anonymously will automatically be “challenged” and moved into the game for vetting by Posse.  Freeze Tag! is designed with all that great Wisdom of Crowds mentality−influence is minimized by each Posse member coming to their own decision independently, then we aggregate into a collective decision to determine if a tag should stay or go.  After all, why should one person decide the worth of a tag, when a collective decision may be more accurate?  It will be interesting to see the results of this and we’ll report back as we see what happens.

To start Freeze Tag! off with a bang, we’ve populated it with all the anonymous tags to date and, in addition, thrown in all those auto-generated tags that need a bit of human review.  This may sound complicated, but I think when you play Freeze Tag!, you’ll agree that all the complicated goings on behind-the-scenes is bundled up in a pretty simple package that, we hope, is fun to play.  No spoilers or anything, but be on the lookout for cameos from our own on-site security posse.

Rock on, Posse—thank you for all your incredible work to date and we hope you continue to have fun with us as we move forward with our collection online!

So, today, we are introducing “Change My Settings.” From within Tag! You’re It!, you can now change your game preferences by selecting the areas of our collection that you’d like to see:

As Mike mentioned earlier, joining the Posse is now easier than ever with our Google account integration and we are excited enough about these enhancements, that Bob (along with a few willing participants) made us a really awesome video to celebrate.  Have fun tagging and many thanks for all your enthusiasm (50,000+ tags!!) thus far!

The re-launch (above, right) allows this resource the same features as our Collection including a more visual layout, community-driven comments and Posse favs—yay! I get to favorite Click!. Some records may seem a little sparse at times—especially for the older exhibitions dating back to the 1840’s where we don’t always have photos or complete dates, but we will be working with Angie Park in our Archives Department to scan and OCR historical press releases. Over the next several months, we hope to add these to exhibition records to fill-out information and give some additional context.

In addition to the exhibitions, we have a few new features going into the collection this week. Related audio and video can now be embedded right into the collection interface (example). We’ve just implemented a way to add related links to display on any object record (example). Lastly, we’ve established a method to display gallery label copy on records (example), which provides some nice context for objects (thank you Francesca & Erin). While none of these additions are earth-shattering, all three allow us to maximize the time and effort that went into existing content and integrate all of these elements we’ve been working on for a long time.

Now for a quick update on the state of tagging (a.k.a., our Posse is awesome). The collection has been live for roughly two months and in that time we’ve seen a lot of participation. In looking at the 3,772 records with images, 76% of those objects have been tagged by Posse, while another 19% have been tagged anonymously. At this point, only 5% of our records online (again, those with images) have no tags. In terms of tagging sources, we’ve seen:

• 7,498 tags contributed by Posse using the collection interface
• 12,623 tags contributed by Posse using Tag! You’re It
• 3,821 tags contributed by anonymous taggers using the collection interface

The quality of the tagging has been impressive—I’d suggest browsing the top tagger profiles to see what I’m talking about. We’ve received some great feedback worth noting, so take a look at the comments on this record and this one for examples of various conversations surrounding tagging issues.

We’ve just launched our collection online and now we need some help tagging, so visitors can better find objects within it. Of course, we were thinking, why does tagging have to be such work? Why not try and do something fun with it? So, Tag! You’re it!

We’ve all seen Google’s Image Labeler and it’s a really fun tagging game. Google pairs you with someone live via the web and you both tag the same image independent of each other. At the end of the timed session on that image, the “matches” are revealed and you gain points if your tagging terms match your partner’s. Google gets two things here: tags and relevance. If two people independent of each other tag the same image with the same terms, Google can establish that term with that image as more relevant when searching. Pretty cool. Alas, Google has traffic we all wish we had and it’s easy for them to match partners up on the spot. Chances are there will always be at least two people in Google’s universe who will play Image Labeler at the same time. We don’t have that kind of circumstance, so we had to re-envision our game to work in a slightly different way.

We may not have Google-like traffic, but we do have a Posse (now sortable by most recent, most active, top taggers), so why not design the game so the Posse can play against each other, but perhaps not in real time? Zing!

When Posse members login to play the game (yup, we are requiring an account for this) they tag away at objects that appear on screen. When a Posse peep tags he/she can’t see what other taggers have termed an object. [Think The Wisdom of Crowds where we know independence and minimizing influence is key.] As a Posse member tags, they move up the ‘tag-o-meter’ scale and pass other posse members in the standings. The ‘tag-o-meter’ has few surprises in store as you move up the scale—no way am I giving anything up, so you’ll have to go play for yourself to find out what I’m talking about. If you get to the top tagger position, there’s an even bigger surprise and you get the option to get an email notification if someone out-tags you from the top spot. When you’re tired of playing and want to end, we’ll show you your term matches with other Posse members (matched terms then become more relevant when visitors search our collection) and you’ll also see your standings (both for that session and overall, from the start of your Posse account).

That’s it: A simple and fun activity for the Posse which, in turn, establishes better relevance for the visitor trying to search our collection. Sweet!

Why no screenshots in this post? Go play Tag! You’re it! and see for yourself.

It has been an incredibly long haul. To get an idea of what we went through, check out this seriously-funny video a friend sent me. We started more than a year ago, had to find and implement a Digital Asset Management System (DAMS) and then develop a way the DAMS could talk to TMS (The Museum System, a.k.a. what we use for collection management internally) and then port everything to web. Deborah Wythe is going to be writing soon about the ins and outs and Paul Beaudoin (our fabulous programmer who made this possible) is going to open-source the code by the end of Summer if you want to take a look at how we did this.

There are currently 5,168 records online and this will continue to grow over time. The curatorial staff felt is was important to only release works with vetted data. While there are all kinds of arguments both for and against this kind of thinking, we felt it was important to honor their wishes. Records will move out more slowly, but it also means the data will be in good shape when it does and that’s a good thing.

We had three goals to accomplish in this round of development. First, provide the collection online for researchers and scholars. Second, provide a way a casual user could just jump in and start to visually navigate throughout. Third, we wanted to ensure putting the collection online would be in keeping with our mission and our community-oriented goals.

All of these factors are in careful balance. The strict data is there in clearly formatted areas and we provide an easy way to print this kind of information. We’ve implemented a very visual “related” column to promote browsing and accidental discoveries (serendipity is key). We’ve created a social component where visitors can create accounts and then anything they favorite, tag or comment on will be attributed to them both in the collection area and on their profiles—here’s mine.

In terms of the social component, the biggest thing we did was look at established tagging models and sort of reverse them. Sure, we’ve made it easy—if visitors want to tag they can do so without logging in, but I really wanted to re-think this and put the “social” back into tagging. When I tag another person’s photos on Flickr, I know the owner is going to see my contribution coming from me and even though that exchange is private, it is distinctly social. Even in The Commons on Flickr, as the manager of the Museum’s account, I know the taggers—Flickr lets me see their contribution and I get to know them as individuals. We took this same idea and made that a public exchange in our Collection. So, if you create an account and start tagging—you are rewarded for your effort because it displays right there on the page and we get to know you (or, at least, what you decide to share with us). Check out this record or see below for an example.

The reality is, there’s an amazing amount of work to do to make all of these areas (research, navigation, social) a lot more rich, but this is a start and we’ll be publishing more about this in the blog as we go along. Needless to say, we have plenty on the to do list.