Why Run a Service?

The debate over when to build, buy, or use is one that rages in higher education information technology units all the time. I am constantly asked why we'd run that service versus just relying on someone else to host it for us. I sit in meetings where the debate over taking something off the shelf for our use is weighed against our desire to build it. It never ends and I don't expect to ever really have a solid answer.

Not too long ago, I was sitting and talking to Brad Kozlek about our choice to run our own blogging platform. I go through these massive swings about the topic -- usually settling somewhere around, "why not just lean on wordpress.com and focus on training and adoption." That argument works on lots of levels. On this particular day we came to another conclusion about why it is so important that we are running our own service -- the potential for community.

Several weeks ago I was lucky enough to spend time talking and presenting with Dr. Abdur Chowdhury, Chief Scientist at Twitter ... I wrote about it then, but have been thinking about it nearly nonstop. What became incredibly clear to me was that Twitter is sitting on an Ocean of data. Data that they are working really hard to turn into meaningful content. If you go to the Twitter Search page you'll see that they are making sense out of this data and showing us how clearly the social web is plugged into what is happening. They have their "Trending topics" displayed right below their search field and it shows you what we are all talking about 140 characters at a time. I'm sure many of you have heard the story about how reports in Mumbai were first broadcast via Twitter and the first picture of the plane landing in the Hudson River came through the same channel -- its obvious that what is wrong with big media is the same thing that is so very right with the social web -- connections building community that is, in the case of Abdur and Twitter, predicting the future as it happens.

So back to the Blogs at Penn State ... as Brad and I sat there we realized we are sitting on a river of data that is built entirely on people right here at PSU. Now that we are reaching the 10,000 user milestone with the service we are seeing an explosion in the understanding and use of tags for filtering content. Courses are using them to aggregate student posts together, students are using them to mark portfolio entries, departments are using them to pull information/knowledge about initiatives into focus, and so on. Once we realized that we started to realize that we could begin to act a little bit like Twitter and use our data to see trends and ultimately predict the future as it unfolds. With this in mind we're working on a few new and interesting ways to not only tap into the community but also ways to let them move the state of the University around a bit.

A simple example is something I'm loosely calling, "PSU Voices." Essentially we would hand out a tag each month (or perhaps week) related to topic we'd like to see the community explore. Imagine during April (when Earth Day is) asking the student body to write, or post pictures, videos about "ideas to make PSU a more green campus?" We'd ask that question, provide a tag, and watch as the aggregate posts of that month's conversation came into focus. If we took a simple advertisement out in the student newspaper, The Daily Collegian, to get people to participate I wonder if they would? If they did I think the results would be amazing.

We've already started to pull out some trending data based on the popular tags and we are seeing some really interesting things. It was clear last week that lots of students were working on their portfolios. One of the next steps is to build an interface between the tag and content search to see what people are talking about in mass ... I can't even imagine how interesting that could look when we have 20,000 or 30,000 people writing regularly around PSU. I'm not ready to share the pages yet, but I am hoping that in the next couple of weeks we'll start to see the unintended results of running our own service -- the ability to not create community, but to coalesce it. Anyone have thoughts related to these ideas and others?

Comments

[Cole Camplese]

@Peter Linehan I'm glad you found the Tag Search helpful! Its something we really think has lots of value. We are working on something the pushes this to the next level -- we worked on a prototype today and it has real potential. Imagine doing this plus being able to automatically pull content in from across the social web that is related to the same tag. Could be great for a course, and event, or anything that you want to gather additional intelligence about. Mashes up our own ideas with the aggregation approaches James is referencing.

[Peter Linehan]

Your article touches on something that I have started to realize, that blogs are multi-dimensional rather than flat like a single document. For example, at a recent local forestry conference I assigned each student to write about one speaker and to include a link to a similar topic. I game them a unique tag for their blog entries that I could use to string their reports together. When we discussed the meeting in class everyone had something to say about at least one topic. And they could see what everyone else had written. I think it would be great to try something like this on a larger scale, as you suggested.

[Peter Linehan]

@ Cole Camplese
Here's a link to the aggregation:
link

The spelling and grammar is rough in some of the entries, but that wasn't the point of the exercise. I found this technique in the help site for blogs.psu.edu. It is easier than setting up all the students in an RSS reader. I would be happy to talk to some about this.

[Cole Camplese]

Hi Peter! That is exactly the kind of stuff we are really interested in hearing about. Is there a way that I could have someone come and talk to you about your experiences? The stuff you are talking about is really all about good practice -- finding interesting ways to let students prepare themselves for real discussions is ideal. The thing the blog adds is the ability to do it openly. Through aggregation and commenting there is the potential for a rich pre-class discussion to break out. Very cool.

I looked at the comments James is linking to and I have to say I think the thing that is emerging is that we have a standard place to send faculty, staff, and students to. Even though we can use simple aggregation, having our own platform leads us to place where we can tap into the data at the source. Maybe it won't provide us anything more than just using external spaces and aggregation, but we're going to see.

[James E. Robinson, III]

some commentary... in friendfeed:

http://friendfeed.com/e/9004929f-1473-3f05-78db-e71bd5f81486/Why-Run-a-Service/

[Patrick Murray-John]

Completely agree about need to be able to tinker with and explore all the various data mashup and analysis possibilities. Coincidentally, yesterday D'Arcy Norman and I ended up doing exactly that. See blog posting patterns for UCalgaryBlogs and UMWBlogs

[Cole Camplese]

@ Patrick Murray-John I read both of your posts with great interest ... amazing to see the patterns associated with institutional participation. I recall hearing someone from AOL speak a number of years ago and he was showing off the usage stats ... amazing how much they mirror the patterns you are seeing. Without access to our own data we can't understand our services completely.

[Brett Bixler]

Why not ad some game-like elements to this? Make is a challenge with some sort of reward - add Pligg or something like Pligg so the community can vote up the best posts?

This would be a great way to encourage early participation. And the reward doesn't have to be much - maybe a custom caricature by Dave to be used as a personal pic?

[Cole Camplese]

@ Brett Bixler Great ideas! I wonder how we could mix more elements of game design into the Blogs at PSU? As you can see there are tons of killer posts floating around PSU -- from all sorts of sources. I remember Henry Jenkins talking about a game he's seen played with Wikipedia where students are given a random starting topic and a random ending topic and they have to see how many clicks through articles it takes to get from A to B ... an example might be, "get from the topic iPod to Kennedy Space Center." I wonder if we could design something like that uses tag aggregation as a pass through?

The pligg/digg voting is absolutely on our minds. Baby steps in between the real projects ;-) to get there!

[Brett Bixler]

The # of clicks game has been around for at least a decade - I remember the one where you started on a congressional site and had to navigate to a porn site in 5 clicks or less - not as hard as it might seem!

[Patrick Murray-John]

I s'pose we could implement our own -- that might make it easier to resolve some issues I see developing (more below). On the other hand, it might be rebuilding something unnecessarily.

On the issues...first one I see is directed vs. undirected relationships. This twists up what I said above about the sequence. So, if I've tagged two pages with "linkgame:ipod2jfk1" to signal the first step in connecting the concepts, so far I don't have info about which is 1st. And additional "a" and "b" at the end could do that job. Or, the delicious feed reports the timestamp, so if players bookmark the start first and the end second, we could derive the direction from that. Different paths could partly be differentiated by player (id'd by delicious username,also in the delicious data) But problems if there are two paths from the same person. Maybe a second tag to differentiate paths in that case? Or, build our own thing that asks for the direction, distinguishes paths, etc.

For the common tag applications/games/activities, just the usual use of delicious tags would probably suffice.

[Cole Camplese]

@ Patrick Murray-John What if we started to think about this not as a game, but as new way to direct students towards research as one would bang through posts/sites? My main goal may not be to push from point a to point b on the web, but to connect concepts and let them easily track that progress. So if they are trying to learn something about globalization, we might start them with a set of related sites (in a Times People style bar) and let them add new resources into that "space" as they learn more.

Do you see where I am heading? I am think of this as a directed starting point, but letting the students add new knowledge into that pool as they collect their own evidence. It could be further enhanced if they leave comments on the sites they are tagging along the way -- or perhaps better yet, using some sort of a trackback to indicate that they both linked and posted a meta reflection of that site.

Does that make any sense?

[Patrick Murray-John]

I love the clicks game idea, and I think there might be two directions to it. First is the idea of using/mining the links within the posts themselves. I've been a bit disappointed by how rarely students are linking between blogs in UMWBlogs (maybe things are different at PSU?) The game might be a way to help that. Second, maybe the game could involve anyone creating the links that _they_ see between posts. Wouldn't be part of the HTML structure, but would be easy with RDF. Maybe implemented as a FF sidebar that lets you tick items from the browser history and add a studentLink relationship between them?

BTW--don't know if it is still up, but there used to be a "Wikipedia Relationship Finder" based on DBpedia data that would mine out the links between two concepts A and B.

[Cole Camplese]

@ Patrick Murray-John Interesting concept ... I wonder if we could do something a little like the NYT Times People app that could follow people around and allow them to easily mark sites for linkage/sharing purposes. I like the idea of encouraging linking ... long term it could make for some serious new connections and ideas.

[Patrick Murray-John]

Oooh! I like it! Doubly so, because it might be really easy to implement with delicious and machine tags. Instead of "for:", use e.g. "linkgame:tag" where 'tag' is whatever you want. Participants register their delicious feed with the app for the game. Then, players apply the tag to relate whatever concepts, and the app processes the relationships between the URLs from the delicious data.

For a sequence, maybe using tags like

"linkgame:ipod2jfk1", "linkgame:ipod2jfk2", etc.

If all those longish tags can be made easier with a sidebar app, all the better!

Whaddya think?

[Brett Bixler]

This is heading somewhere! I'd love to automate a living, dynamic concept map built by this - one that expands as long as the "game" continues. So we'd need something that labeled the connections between concepts/sites.

[Cole Camplese]

@ Patrick Murray-John I think we may have found a cross institutional project! I have also been talking to a colleague in our Emerging Technologies team and he is keen on something like this. Let me draw him in and see what happens. I'll also bring in Brad Kozlek who is in the center of our Blogs project.

Really smart idea of using the delicious tags ... just a quick question, could we implement (not sure why) our own open source social bookmarking app specifically for this purpose or is delicious/google bookmarks the way to go?

[Patrick Murray-John]

Hmm...this is starting to sound a lot like Twine. Have you checked that service out?

[Brad K]

In regards to this link relationship builder: why not let the existing blog software do this? A blogger makes a post with links to various other posts and tags it. The tag cloud is made richer and pointers to more content are added.

I think it needs to be simple and flexible. I don't want to specialize it too much. Stick with the basics: blogs, posts, tags, comments. You can do a lot with these. I think the community is what needs development. We need to jump start communities that are reading, writing, linking, participating.

[Patrick Murray-John]

Brad,

I very much agree--any way we can encourage those better practices is good. I think that's why the game idea appealed.

On trickiness with the existing tools is the different varieties of links. So, let's say that there is a set of Cole's starting points about globalization, all posts tagged 'globalization'.

Then a student finds a new resource they want to add. They can link to it, and tag their post 'globalization', but they can't tag that resource itself. That's where delicious would come in. Something is needed to be able to bring in the resources wherever they are found.

It'd also be nice to be able to signal the relationship between a reflection and the thing that it is a reflection on . That would call for distinguishing between the link to the new resource, and any other links that also appear in the reflection.

And I'm always worried about disambiguating tags.