APS8: Cornell's Digital Well: A social networking repository for marketing information

Dirk Swart 
Assistant Director of Administrative Computing, CALS, Cornell University

The audio for this podcast can be downloaded at http://2011.highedweb.org/presentations/APS8.mp3

Announcer: This is one in a series of podcasts from the HighEdWeb Conference in Austin, 2011.

Dirk Swart: Good morning folks, thank you. I hope you’re all exhausted, if you’re not, you’re doing it wrong.


Dirk Swart: So today, I’m going to tell you a little bit about Cornell’s Digital Well. And there are going to be two things, which I hope you’re going to take away. The first one is obviously, the product, which is a product working with legal to open source, so you might want to use. But second of all, to share a little bit of a full process that we went through to do this.

But before I start my presentation, I’m going to tell…ask you a couple of questions and then tell you a little story - a little story about Alice the marketing executive. First of all how many people here are involved in Web marketing? So you’re writing stories, you’re creating images, you think some of it is no, it’s not IT stuff, but were marketing side of things.


So the reverse, how many of you are involved in the IT side of things? OK. So it’s about a 50/50 split. I’m going to try and use that to gauge of how much to talk about each thing. I’m going to talk to you, and tell you a little story, so Alice and Bob is her boss. So Alice is the overworked marketing executive. So she...and I’m going to use the well to tell you the story.

So she gets a call from - she works for College of Agriculture and Life Sciences, which is where I work, and she gets a call from her boss, he said, “You have to come up with some news abut the dean.” The dean happens to be Catherine Bush. She’s the dean of Cornell University.

Now, Alice is new, so she doesn’t really know what goes on at the university, and she sort of knows who the dean is, but she never met the dean. She’s really sort of a person who just wants to get along and get stuff done and there’s just a lot of things to do. So she needs to get this done quickly.

So what she does, she goes to the well and she searches for - let me see if I can pop up a new one.


So she searches on this and she searches for the dean’s name. And she comes up with a whole bunch of stories all about the dean. So here’s one, she addressed at the Colombian government and she has an image. So she looks and she says, “That image is no use on its own. Let me see the providence of the image. Let’s see what’s it about. Let’s see who it is and what it is.”

And she sees an article where she can look at the article, she can download the article if she wants to - I’m not going to open it - she can open it in her document, she can edit it, she can add tags - so she can look at. She was this, oh, it’s got five stars. Maybe it’s a good, maybe people have rated this as a good article. So I’ll take a look at that, and you know what, I’m actually going to take this picture and I’m going to associate it with this article.


So I’m relating this to...I’m going to say that together. They’re both about Catherine and I can look at this article now and I can see that it has a related image. So I can say this was about her trip and this is about the article that somebody wrote about this trip. And I take these things and I can add them to...I need to share them with my boss. I need to share this information that I found.

So what I’m going to do is I’m going to drag them across to a project box and add it to the project box, which I can then share with other people. So this is an easy way to really get in touch with a whole ton of information. What I’m going to explain to you today is a little bit about this and a little bit how none of this information that you see is actually stored in this Digital Well.


It’s actually stored somewhere else because one of the first rules of adding or creating a repository for content is plenty for somebody else to have another thing they have to add stuff to, and who the heck has got time to edit to the database that they already have, right? So that is the story of Alice and how she finds information in a hurry for her boss.

So let me go back and tell you what I’m going to talk about. So that was the introduction. I’m going to talk about three problems; the three problems that we encounter. I’m going to talk about the state of the art, a little bit of attachment on repository. Does anybody know what a repository is? How many people know what a repository is? About half of you know what a repository is. And how it works and some lessons learned that we can take home.


So this is what it looks like when most people get us to get marketing information and say, “Hey, roll up, and by the way, you’ve got this job to do.” And in addition to all that, he has the stuff you have to do. Alice have to go and get an article while she’s doing other things. So what she does is she runs away to find people to help her get the article. Let’s go and tap into our group of people to help her. So she goes and looks for the people to help her.

And they were all busy with their jobs too. So she doesn’t always find people to help her. You expect her to sort of make your own way and do your own job, so it’s difficult. So we need a place, we need a central place for somebody who’s new, knows to go to find information. I mean, we have - Cornell has about two photo repositories, we have the news articles, we have like a list of Twitter feed, we have a ton of stuff.

And if you knew or if you’re in a job where these things change a lot, you might not even know to look in those place. You might not even know they exist. You might not even know how to go to the repository of information to find articles and press releases, or to know to go to university photography - we have a photography branch - and look in their repository.


And even if you do go there you might not know how to find the stuff. You might not have a login. I mean you have to get a login to like these eight or nine different repositories. So it would be so much easier if there was one place you could go to, and one place gives you access to all those things, gives you a reach into those organizations and that’s what the Well is.

So let’s talk a little bit about the three problems that we encountered when people were trying to find information. The first one is, the information is blurry. From September, from CNN, mishandling of jobs speech timing blurs White House message. OK. I guess the white house is getting a lot of blurred messages these days. They’re facing a lot of pressure from other parts of the political spectrum.

You know, there are people trying to blur their message. But even if they’re not, blurring of a message happens a lot, and I’ll give an example. Our president of our university made two announcements and within about a week of each other.


The first announcement is like Cornell is going to spend top dollar make no, spend no expense to recruit top faculty. That was the first message. The second message was that there’re going to be a significant lay offs in the staff. Now, those two messages were sent on three different times. First, the message from the lay offs was sent and a week later, the message for recruiting top faculty was sent.

Unfortunately, their top faculty recruitment message was sent by the president. He stood up and spoke about it. So that message propagated off into the world very, very quickly and it overtook the other message. So some people heard the message in the other order right next to each other. Oh, by the way, you stop maybe working out in a part of Cornell. We’re going to make love to you when we’re spending lots of money somewhere else. That’s the message.


It didn’t really work. It was a blurring of a message. And I’ll explain later how that immediate propagation rates is kind of an important factor in how you do business as a marketing person. And that blurs messages, it creates problems.

The second problem that you get is inaccuracy. Now this is a physical inaccuracy but as messages get transferred from one organization to another organization, to another organization, get reported on by the central organization, they lose accuracy. And if you’re an academic, you’ve got a very strong vested interest in your...they read these words very precisely. You can’t just substitute it with another word because it sounds nice. They really have very specific meanings a lot of the time.

And the other thing is, things are hard to find. Do you even know what that is? That’s an ivory-billed woodpecker. There was a huge announcement; oh, we’ve found the ivory-billed woodpecker. Well, nobody found the ivory-billed woodpecker. People were out in the swamps in Virginia. I think it was in North Carolina for most of the time.


So that wasn’t interesting. So the information is hard to find. You don’t even know what is out there for you to reach. And because there’s no central organization, there’s no person saying, these are the things you can connect to, to find the information you want. Are these problem other people experience? I’m just kind of curious. I’m seeing a lot of nodding. OK, so I’m not completely making this stuff up.

So now, I want to talk a little bit about three concepts: message for each town, which I’ve touched on; content reuse; and location disinterest. Message for each town. People are getting used to it. Imagine a big like the world-is-flat kind of organization - big flat organization. We’re the central organization a lot of time, we’re the university administration, and we’re putting out the news.

We want to put out a coherent message to the world about what the university is saying and about the research the university is doing. But the people who make the news: the scientists, and the academics, and the researchers, they’re up on the periphery, right? And they sort of do something news-worthy; we have to go and find what they tell us and comes back, and there’s a central app, there’s this huge gap.


It’s not clearly easy to transition that gap. The cool stuff happens here but the ability to deal with the cool stuff happens here and that...I mean, it’s a very common dichotomy, right? You get this, the ability to deal with something is on the one end of the spectrum but the thing is happening on the other end of the spectrum. And it’s the same if you’re solving problems generally, problems to solve locally, but solutions to problems are organized centrally. So that’s a bit of a problem we have to deal with.

And the next one is content reuse. We’re expected at Cornell, if you’re in marketing, to reach out or to compete, if you like, on all forms of media all the time. That’s the message that we received. And now, we don’t have time to write content for each one of those things or that is impossible.


You can’t create tailored content for all those things all the time. So you have to come up with a solution. You have to come up with content that is these things so that if you were a politician you will be talking points. You have to come up with content that can be repurposed easily and if the content can’t be repurposed easily, you have to write.

I mean, that’s you have to write a new piece of content, you risk blurring, you risk reinterpreting the message which might be wrong. So it’s very difficult to do. But we’re expected to be competitive. We’re expected to reach out to people and give them a news in the form they want. If they do Twitter, we give them a way to do Twitter. If they do this on the radio, we need to do a way to do that.


So the third problem is - it’s not completely related to the slide - is what they call location disinterest. So here’s a trick question, does anybody know where on the hard drive their email is stored? At what sector of your hard drive your email is stored? You know why you don’t know? You don’t care. Who cares? It doesn’t freaking matter. It doesn’t make any difference.

And the same thing is with information, with photographs, with contents, I mean, who cares where it’s stored, right? It does not make darn difference. The only thing that matters is, can I find it, most importantly can I find it again. I found it the first time, it was like the Google search, again, you find the Google search, that’s great.

You try and search the same thing in your host. If it’s a big piece of news, you can easily find it, if it’s not, it’s very, very difficult to do. Finding information again is tough. So what matters is can you reach out and get that information? So in the Digital Well, we did not strive to create a tool that is a repository that you put information into.


What we did was connect to the information that already exist out there, and it was based on a realization that the mechanism that you use to access information, usually is the same mechanism that you use to put information and use the same sort of interface. That doesn’t have to be the case. It’s in fact, unnecessary to have the interface be the same. You can put information in through one interface, usually the native interface or whatever app you’re using, your photograph app, or your news app, whatever it is, and you can retrieve it through a completely different interface.

An interface is optimized for retrieval and it’s interesting that most people who design interface has designed them as if those two sort of things are somewhat symmetrical. Hey, you have to add information, you will retrieve information, but in fact, people retrieve information thousands of times more frequently than they add information. So as an example, Wikipedia, millions of people use Wikipedia. More than 90% of Wikipedia articles are contributed by less than 10,000 people.


Very few people contribute to Wikipedia. And as we saw, Wikipedia is kind of optimized to get information out and it’s pretty well done, it’s easy to add as well, but it’s very easy to find information. So I think that’s an important concept to get. They don’t have to be the same.

So let’s talk about repositories. So repository, very simply, is a place to store information that where the metadata for the information is tightly bound or associated with the actual information itself. If you’re a librarian, please don’t shoot me, that’s a very rough definition, but it will work for our purposes. The most important point is that the metadata is very important. You have to associate metadata with the data itself.

And we found that some of our repositories didn’t do that. I mean, they associated some metadata, like who took the photograph, but they didn’t associate the metadata that come. So, an example that we found when people came up to find information was they want to get a photograph, they needed a photograph, and they had to get a photo credit, OK. Who is in the photograph, who is in the photo? What are they doing? So that they can write little by line under the photo and do they have permission to use the photograph?

It took a week to find that information to go, to get a release for the photo, to find out who the people are. Sometimes it's listed, sometimes it's not, what are they doing? Can I use the photo in this context? It's not that easy, it took some effort, it's crazy. You should be able to have a document stored in one place that you don't care about which has that information, but a narrative about the photograph. And you should have to have the photograph and those two things should be associated with each other like in the demo this morning, I associated two pieces of information.

So, the person who's a news person working said to me, they don't even know, I mean we're a big university. They don't even know these people are half the time. So, that they can go and look at that photo and say, Oh, that's Catherine Bush she's talking about this, she's in Columbia, she's doing these. I can write a byline now, and I can add a byline.


And actually get people to write the byline in the description. So, what we're doing is tying into existing repositories, tying into those adding metadata if it doesn't exist. Harvesting the metadata they produced. And now the important design consideration we look at is, 'do not ask people to enter information that you can find out yourself. OK.

If you have a few coming up to tax time, right. You're filling your tax form you have to write your social security number at the top of every page. And it gets annoying, why write the same information five times? It turns out they have a good reason, OK, so if they lose one page. But for the most part asking people to enter information more than once is just annoying.

I mean, we're all busy people here. And so, things like who the person is adding the information? Who's uploading it? What time is it uploaded? What department of the university did they worked for? What's their job title? You can get all that information by looking at the university directly.


Do not ask people to enter, it's just annoying. So, what you can find -find, be intelligent about how you find information so that people don't have to do this. OK. We look at two existing repositories, just talking a little bit about the design process that went through.

Even heard of these ads from Duracom? You've heard them? A couple of people. These are actually fantastic products. I'm definitely not going to diss them, but they are really nice product. The design out of the library, the library's best. So, for IT people it's kind of interesting that a lot of stuff that we're sort of wresting with as we deal with the Internet. Librarians though of 20 years ago.

You know I thought about it with card catalog, but it's thought a lot. They have solutions to the stuff. So we should go ask the librarians. And those two products are very nice products. They use it in the places where too hot. Simply too difficult to use.

See a lot of people nodding their heads. I mean -the rule for us was if you cannot muddled your way through the interface, if you constantly look at it, and kind of like mess around a little bit, and kind of fine your way, it's too hard.


Of course it doesn't have a user interface. They expect you to sort of get some other one that you can kind of download and add on yourself. That's why it's kind of too hard. These base was too hard for us because it was optimized for academics.

It's optimized for academic publications. It has a lot of -it has this kind of old school stuff you know, when you go to the screen, you type in the screen. You type in all the stuff, you press submit, you upload it, it's really painstaking. We have drag and drop, we have all those features, we have an interface that the librarian who uses information everyday knows how to use, but it's not an interface that somebody who's familiar with the web is searching for information, knows how to use.

So, when I say too hot, I want to be a little bit careful because I mean I should this is great product and Cornell uses this space.


And they also use Fedora but they use have been very specific context. And this broad like appealing to marketing people who are busy who needs to get some , it's just not possible, it's not easy to do.

Also, from a technical point of view Fedora is very good at retrieving large records like books extremely poor at retrieving lots of small request like little snippets of information, little sentences. It's designed to retrieve manuscripts. Do proper searches on some manuscripts.

So, now I feel navigating most of repositories. That's one of the most watched videos on YouTube. It's to the tune of something from Pulp Fiction. So, if you're going to look for the cat you'll find that it's just -going to love it.

But they're just too difficult, they're just too hard to use. And for someone like me, I mean like, how many times you got it right? I mean , I hope.


And if I'm finding it difficult, it's not because I'm stupid, it's because I don't wish to apply my brain to that. You know my brain is like sort of valuable and I want to use them to solve my problems, I don't want to use them to solve my problems because you didn't think about it, no.

So, here's what we did, I want to take a little from segue, step out of the one conversation that I'm having to tell you the narrative of. Well, and tell you a little bit briefly about what we did. First thing we did is we didn't ask people what they want to do. OK.

We don't care. They don't know what they want. And the reason they don't know what they want is we find it very difficult for people to envision a future that is not the current future. How, to envision a future that is different from today is hard.

So, it's not a surprise that people find that difficult. So, what we did was we pose the problem to people, we said, this is a problem tell me a little story about this problem. Tell me a story about the problem.


And we drew up these character studies,OK. So, Joe is a director or communications and engineering college. He needs to have access to many communications with you. These are the things that are important to Joe. We wrote up this little stories.

He was asking people it was just a complete waste of time, because they don't know what they want. So, we asked communicators six communicators around Cornell to draw a piece of the stories. I think they did a great job, and you know, just tells the story of Joe coming here from like small university, he's adjusting to the fact that a lot of communication is carried around campus or generally by admission directors who are not aware of his interest or his views.

A little story about that and it listed the features that impress him and the characteristics of a successful solution. What is a successful solution looked like? Not what buttons do I press, what can I do with it? Tell me what I can do with it. And we'll work out the button. OK.


Then we characterized this, we had this. So this are names of people right up here that we invented. And we looked at all the features that people listed. We extracted the features. We didn't ask the people to extract the features, we extracted the features.

These are the features that in that narrative, when we read the narrative this is what we came across as a feature. Clearly define print and web resources, somebody wanted that, OK.

Again we're not asking them, we're not listing to say, check if you want this. And that's important so psychological trick, right. Because people are like, "Oh I want that" , you don't even know what it is, what do you mean you want it? We haven't even invented it yet.

So, we extracted this ourselves from the narrative. It was more work, no question, but in my opinion it was a much better way to go about solving the problem than to say to people he has to check this to stuff. However, but ease of use will triumph everything, OK.


If it's not easy to use, you're wasting your time. Put at least half your effort in ease of use. In ease of use is defined not by what you develop as one, it's defined by what your audience want. OK. Your audience tell you what's easy to use, if that say it's hard to use, don't say, " No, no, you just have to go here and do this." No. They've made their point, right?

You don't have to go and explain to them because when there somebody is using it and sitting it, they don't have somebody to explain to them, you know. You might be able to explain to one person, you can't explain to hundreds or thousands. So, a little lesson, lesson about ease of use, OK, it has to be simple.

That's the remote I want. My wife does the TV controlling in our house, because I just don't know how to do it, it's just too hard. OK. I do want to talk to you a little bit about the importance of good design as well. Say, we have a good design, that's a bad design. And the problem with bad designs is they stick around.


Bad designs stick around. So, this is to me the greatest example. I'm a foreigner, I'm not from here, this is a stove top kettle. This is an electric kettle, OK. In every other part of the world, people use electric kettles, OK. There are many better, more efficient, and safer. OK.

No question, electric kettles are better more efficient and safer. And switch off automatically. OK. This kettle, stove top kettles don't do that. OK. And they are used with gas or ranges which more electricity, OK. Most people in the United States use stove top kettles.

They don't use electric kettles. Canadians use electric kettles. Pardon? Some people do use one. They are penetrating we see enough in the 10 years, that I mean they have penetrated a little bit. But it is interesting -the point I want to make here is that designs that are many better, that are absolutely proven to be better on almost every metric. The only metric this is not better is counter space, OK.


There's a little bit more counter space to have an electric kettle. In every other metric, an electric kettle is better. People continue to not use it. It's hard to change people's preferences. It's really really difficult.

And if something like this that you can physically feel it is difficult, imagine how much difficult it is to change people's preferences when you're using a web product that then may have to use for the job or something. It's just hard. And I wanted to sort of caution you as I'm sure you know, but caution you again.

I know you think it's hard, but it's even harder than you think it is to get people to change, to change people's preferences. So, that's the one -if you take home one nugget, one little nugget from this conversation, it's even harder to change people's preferences than you think it is. It's really crazy.


And I mean, this is an example, this is just what it is. So, don't necessarily even try, I mean just work with people who had works well. So, now I want to talk a little bit about the well, this is a complicated diagram. You don't have to look at it.

And I want to go back and show you the product again, show you the demo of the product. But the important points here, the take home point is decoupling. Let's talk about decoupling, I'll talk about decoupling -separating the adding the information place from the retrieving , great. That's another we'll take up. OK. Talk about decoupling.

Go in and find things expecting to people to add information to another place, it can happen. I'm telling you that, it can happen. That wouldn't even added to the first place. They wouldn't even upload, they work to one central place, they certainly not going to upload it to two. It just ain't going to happen. So, you have to find ways to go and fix your content.


Your system need to go and fair up that content itself. So, what we do if we have the chronicles -the newspaper, the Cornell Chronicle, it goes, it fixes the articles. This is a streaming of the videos, it goes videos, goes every night, goes and looks, "Hey, what videos have been uploaded?"

That's going to add those videos automatically index them basically and make them findable. You know, let's go and look in the press releases. And the diagram here just talk about having an API that needs to way for programs to talk to software as well. And there's a very great interesting article that should came out about - the publication came out about a week ago by this guy called Steve Yegge, who works for Amazon.

Anyone ever heard of this? You can't Google Steve Yegge's rent. He talks about the difference between Amazon and Google. He says, "Amazon sucks at everything." But it's one thing they didn't suck at. and make information findable.


And that makes up for all the other they suck at. So, making information findable programmatically is really important. So, let's go back to little presentation here. So, the first thing we did when we deign this is we had -I have sort of a design principle that form follows function.

Whoever heard of that? Form follows function have you heard of that? So, form doesn't follow function, what does it follow? If form doesn't follow function, form follows something else, right? Form follows fashion, OK. It's whatever fashionable. We are not that interested in what is fashionable, we are interested in what is functional. Then form follows function.

And what we have is this tool sort of a realization that we came to is that going from one page navigating from one page to another page to do something that sort of "Oh there's an add screen, and then edit screen." It's kind of old hat.


You don't have to do that. This entire piece of software is presented to somebody on one page. This is it. This is the entire application, it's all here. And very simply what it is is little message board at the top here, but it's a list of -I can find people. A list of icons is infinite in length, OK.

It indexes the entire repository as you scroll down, you can add more items on the bottom, and more items on the bottom, more items on the button. So, and what it does is it ranks every single one of this items as ranks. It's ranked individually for every single person using the product. Looks at what have they ranked as important, what have they tagged.


What if they start, you know. What have they done that ranks all the pieces of content that it knows about individually for that person all the way to the bottom. So the highest ranking staff is on top and the lowest ranking staff is on the bottom. This is because we tend cross this problem which of course is a product commonly known about now but you know two years ago wasn't.

So do you know the stuffs good? Well, you know stuff's good when people tell it's good. That's how you know stuff's good. Reputation with it. When everyone gone eat last night, most of the time, you had a conversation with your buddies about where you eat. We went to this Iron Tech. Who went to Iron Tech this last month?. Hey! I feel the Iron Tech is just pretty cool.

But there was some sort of I mean some of you also where people were having this very single negotiation going on in the front of the hotel. And if you're a sort of student of human behavior that was kind of fun actually to watch that going in. And did you ask your friend? You ask your friends about stuff. Well, one of the ways you can ask your friends, you can look at ratings. OK.



Friends rate stuff and that helps you, you know, see what it is, that's pretty obvious, right? You can look at tags and if I scroll over here, you will be able to see, here we are. OK. So it pops up with how many items of tag for each of these things. What turns out when you work like 200,000 pieces of content in this, well you've a lot of stuff tag with the same text.

So, you can make your own tags. You can make your own ratings. Very quickly you can look at people. So that 189 people using this. The important thing to know is that it bothers me. People as we expect to use this product. It's a product for marketing professions. It's not available for free.

Anybody can't go and look at this and see what's in here. Anyone can't go and look at this and see what's in here. This is half baked stuff. All right. You don't want the whole world peering to your marketing process, you know. You wanted to see there our finished stuff at the end.



So you know you have to go to log in, I think there are about 220 people with the communications job title at the point, somewhere on that, when if anyone at the point knows a bit the number from there. So then the 189 people who've been used it, again, they didn't have to go and log in. When I went to this, then they're talking their credentials. For the first time the message, "Oh! You knew or edgy." They didn't have to do anything explicit to add it. They just had to visit it.

So it looks at those people. Also, I can follow them. If I'm the manager or let's say I'm a person who is writing news articles, one of the problems that you might find is how do you find stuff that doesn't exist yet. I mean it sounds ridiculous but I mean people do it all the time.

Findings, how you find something that doesn't exist? Well, if you have a photographer, your favorite photographer, you know he's going to produce a photo shoot for you. Next week, what you are going to do is follow the photographer. Follow means make more important. You can follow this person. So when they do produce the photos and upload them, you'll see them at the top of your feed, right?



Because you know that coming. So you can follow somebody and track what they're doing. You know if you're a boss and you want to see what your employees are doing, I guess, you can do that too. So the units. There's 78 units who use this. Again, nobody enter this information. This is automatically gathered by the system.

So that gives you where to do it and there's also, I don't know if this kind of interesting to you, but there's a whole lot of reports that you can do which I don't want to go into. I don't want to spend too much time on this. I want to spend more time on the lessons. Is that a good place to spend time? Rather than this? Great! OK. So let's go back to the power point.


I do want leave some time for questions, if there are any. Thank you.



So, I do want to quickly go into the lessons learned. I've told you what I think are the most important lessons but here some lessons. You know what this is, right? Just keep it simple. All right. So, that's the most important thing. I mean, simplicity is everything, right? Most of the interfaces that we see that are complicated whether it's to car dashboard or whether it's Amazon, follow the rule of that every good complicated idea started off as a good simple idea.

No good complicated idea started out as a good complicated idea because that's the bad ideas that turns on. They got progressively more complicated as people became use to them. And it's interesting when you see people not familiar with the internet. Look at it now, the pages are so rich, they just don't know how to use it. You get some old people who don't use the internet. They can't start. So, simplicity is everything.



The other one is, you know, you can't keep all the information at one page. You can with A-Jack. You can with all the other presentations you're going to. You can put everything on one page. And if you can't, you're designing it wrong. You know, you need to go back and think about how to design it. And the other sort of subpart for that is often new users don't know what they're talking about. I mean they think what they talking about but they really don't.

And this, I mean, if there's one thing that Apple, which is not my favorite company, but if there's one thing that Apple did right, the first thing they do is to Steve Jobs can be a fine UI expert. We just got rid of them and said I don't know what you're talking about. And Fine Lines was right which is interesting.

And the other is thing is changing the way people work is extremely difficult. It is so difficult that even if you produce some sort of, I don't know, you know, a perfect product, some sort of utilitarian perfect product, it's very hard to change the way people work. You have to do it for evangelism. It's just a challenge. I don't actually, I mean if anyone has a solution there.



You know, bring it on because I'm not sure I have one. So that is pretty much all that I have to tell you. And the other thing is some things are predictable. This is kind of a fun slide. If you can't believe what kind have done to most of presentation but I thought this was a fun slide to put up. This is a Google search frequency on what is Yim Kippur. Well guess when Yim Kippur forwards in that in a year.

A lot of things that you think are not notable are actually very easy to find out. They're very easy to know. I've have one final one. One final thing that I did which I think really works for me. It works really to get funding and nothing else. I gave my VP, my boss talking points if you like. I gave him every couple of weeks. I gave him like a print art and I said this is what I want you to say about the world.



The digital world is a project that we're doing that nobody seen. This is what I want you to say and I gave him a talking point and I said, repeat these words. Then he has had laughed off that time. Yeah. But he repeated them. It kind of work. It was great and he's got much more.

He's got 10 times the evangelizing power that I've brought. So him saying the words and keeping on message really made a difference and I sort have learned from politics. You're going to Sunday morning talk shows and every single person is saying, not similar words, exactly the same words. I mean they've obviously prepared it.

Well, we can learn from that. So I gave my boss and this applies to any project you're doing, I don't know about with your marketing organizations attention span is but my marketing organization attention span was 2 to 3 months. Next. That's about how long. Any loss o that, yeah, yeah whatever. You know, go to somewhere else. You know some other crises has happened. And so to keep people on track was very hard.



So I gave these talking points. This little sort of printed art but actually printed art little four or five points and it really worked. So that was kind of fun. That's pretty much all that I have for you today. So I think if you have any questions, please go ahead unless thank you very much and again, one year, you know, if you're not exhausted, don't drink and drive.


Thank you. One. Two.


Dirk Swart: The time, it was quite a long time. You know it's probably been, I don't know, a year since this started that. I was taking a lead for the first two versions of this product and I stopped being taking a lead so I'm probably not the person they can fully answer that question.



But we have a culture, I think the pretty good culture panel of sharing information, wasn't problem for us to get access to this information from a commission's point of view. The problem was technically how you get access to it and that to me, you know that just the resource issue. Does that answer your question?


Dirk Swart: In the beginning we had one development and then outsource this to a third party company. And so we wrote the specification to a third party company and one of the things I find in a very successful of keeping the price down with the companies that I work with who I outsource things to is that every detail specification.

So we have a sort of mark up and we said works. This kind of doesn't want, but you need to fix it, you know, doing the chronic cases otherwise when people coming at work, they all like make up some outrageous figure because they're want to insulate themselves as well.



So I'm having a very detail scrub and they're right so far is to write, I don't know, there's a long document which when a person clicks this button and this happens then this is what's going to happen. This is the test you going to do, sort of test to the developers. This is the test you're going to do. This is a pause of the test and it was mind numbing to write that stuff which is mostly I wrote but it worked. So, what we did was the detailed spec and we outsource and pay the company and we doesn't save any grand in the beginning to do this, like 70 thousand bucks I think.


Dirk Swart:That's a good question. So, I thought of some sort of just asking people. How can you open it up and that it didn't go anywhere so now I'm working with the legal department to get it open sourced. Cornell is kind of nervous about that sort of thing. For reasons that I don't really understand, I must admit. As a sort of a tech person that's kind of a weirdness to me why that's the case but I mean it's not going to happen in the next month I would say.



Because that's just how fast Cornell moves which is not very fast, but the goal is to get it open sourced and get it out there. Particularly, we want to get it. We have a neighbor in college across the hill, pulled it into college and I really want to get them to use it so I think there's a lot of value for them.

They have been exposed to a lot of interest in it. It is a tool, I think, it works you know and it solves a need. It solves just one in each need and you need lots of other tools as well but it does solve that need, you know. Any other questions? Yes, sir. One. Two.


Dirk Swart: So it's made to supplement. Actually, what I wanted to is use it to replace some things, because it does have a repository function in it, like the repository of last resort if you like.



So you can add stuff directly if you want to. And actually, I mean as a domain specific knowledge base, I think it's actually kind of a cool tool. Aid specific domain doesn't have to be marketing. Any specific domain we want to collect information. We have more information and we find ability of that information is a problem.

And you want people to be able to comment on it and you know socially comment on the articles and things like that and said this is good and that's bad. I mean it could easily be used to be a paste then for code, right? You can just add snippets of code to it and say this is good code, this is bad code. So I mean I would like you to have more a wide use but you know it's just a pinch.




Dirk Swart: Users cannot create a profile function but you, as an owner of repository can create a waiting function. So basically what that's doing is the system mathematical weighting function and assist doing a very simple ECSO created code is written. I should came up with a really nice weighting function that I thought it's awesome. It turns out there's a lot of school is going into that. A lot of algorithms exist for really cool weighting.

So it sort of pulled out so you can replace into whatever function you want in there but for now it's just a straight linear weighting function. And it's probably sufficient to have a linear weighting function, actually. If it gets too complicated, people kind of get confused. They want to sort of trying to work out how it works.

So it maybe some of the bad thing that it wasn't just that simple. So, no. You can't individually change the weights of things. If you rate something to five, you can't change how important giving it at that five is. And if you comment on something or you relate two items together, you can't change the weight to that but if you are an administrator, you can change the weight to that.



But you do it for everyone in the system. And another important thing to comment, everything in the system is visible to everyone. Then we did that. Deliberately, we didn't want people to say, "Hey! I want to put stuff in the web and I didn't know any way to put it. I'll just use, you know, your resource to do but I don't want it show to these people." No, we don't like that free storage place for people in today. Yes, ma'am.


Dirk Swart: It is not drawing in content from socially, not from Twitter. It's only drawing content from the repositories that we have access to. I mean it would be kind of cool to do that but I don't think that' where those marketing organization is going to take us. I think they're going not to take it there.

I mean, it could easily do that. I mean talking to Twitter is trivial. I mean it's really, really easy to talk to Twitter about this but I don't think that's what's going to happen. Anything else? OK. Thank you very much folks.