 Hi. So, my talk is titled Networks of Networks in the OpenMRS community. OpenMRS, as you'll learn in a bit, is an open source medical records community. And I'm Saptarasi Purkayasta. I am an assistant professor at Indiana University, Purdue University in Indianapolis in the U.S. and I'm a senior manager of education programs in OpenMRS. So, let's start with an experiment. I'll distribute two pieces of paper in the room. All of you have to do some effort in filling out these. So, I'll start, distribute one paper here with a pen. So, I'll tell you what to do with it. But we'll do it at two sides. I've done this in class a few times and it works. At least the theory works. But we'll see if it works here too. So, there are two sheets of paper that I've circulated in the room. One sheet of paper, you write your first letter of your last name and pass it on. So, pass it on to the next person. So, we'll go through the rows. And on the second sheet of paper, you'll write one letter, but with a goal in your mind that you're going to try to create a word out of it as a group project. So, you're going to have the word with at least three letters. And as soon as you find a word, so if the first person wrote A, the next person wrote N, the third person wrote D, and the AND is the word. And so, the third person would strike out the word and then get another extra word that they'll, a letter that they'll type in. So, say they write D again and they pass it on. So, this process continues and I want you to think about number of words the group will create, some random number you think this group can create together. So, we'll do those two sheets. We'll pass them around and we'll see what happens. So, what is OpenMRS? It's a community that works in a domain of health, medical informatics that builds a medical records platform for many years. We've been thinking of, or we've been thought of as a community that builds a product. But as you'll see through my slides, it's more than just a product. It's an infrastructure that's resulted in many different products being built. And so, over the last 11 years, the community started 11 years back. I've been associated with the community for the last 88 odd years. And so, we've seen community members build different things based on the need. And what do we do in our community of practice? We create, share, learn about and create a medical records platform. That's the product. We create, share and learn about different kinds of terminologies and health care, some standards around health care. That's kind of a byproduct because we don't necessarily do that as our focus area, but it happens based on the number of implementations that go on. And then we create, we've had different kinds of mobile applications, communities getting created on the side who started in our community, but then went on to other, to build their own communities. So, that was kind of a co-product that happened. We then had health informatics knowledge that was created, kind of a raw material that's shared between the community members. So, you build capacity and things like that. We have open source principles that we train people. So, we do GSOC for many years. We do code in Google code in. We do university programs and we train individuals in open source principles. So, that kind of is a side effect that happens because we are an open source community and because we've taken up the principles of health equity with the idea that they go hand in hand with open source production. And then we've created some innovative development processes where we use Scrum, but without contracts that you're going to do this the way that a Scrum runs. So, Scrum works very fine in companies that have contractual developers with them, but because this is an open source community, we have volunteers coming in and so we have a different Scrum model that we use in our community. So, these are different things that we've created and this is something that I like to quote from one of my ethnographic interviews. So, a group of doctors were having their 10th annual meeting. For these years, they had been coming together to hear how others were doing health information management and sharing knowledge about a platform that they felt was missing for their profession. So, only a few of the original members are still part of the group, but shared work, stories, artifacts created over time gave them meeting a sense of continuity and purpose. If you were a fly on the wall at one of these events, you would notice how new members were warmly welcomed into this family. How many people stepped up to take initiative or share their battles, battle stories of how they have ambitions to provide care to the underserved and also about how they practice health information management across the planet. And so, what the community of practice quote here talks about is how we've evolved over the years by involving new members becoming part of the family and yet we are still having the same ethos that we started with. So, what are the parts of community of practice? So, community of practice is coined in academia by Levin Berger and talks about three main aspects that domain the community and the practice. But one of the reasons why I say OpenMRS is rather than a community like many other communities in the open source world is because it's become an infrastructure where people build other stuff than what our core product is. So, we have a domain which is true, which is constant, which is medical informatics. We have a community of people who work together to improve healthcare around the world, but the practice is quite different and I'll get to how that is different. Because we follow what I like to call an extreme Bazaar model. When I say Bazaar, I'm talking about the ideas from Raymond's Cathedral and the Bazaar, which is Cathedral is the ivory tower kind of top down model. The Bazaar which is the more distributed, different ideas come together, production and distribution kind of model. So, why is this, why do I call OpenMRS not a real community of practice is because of the practice domain. The OpenMRS platform and the core modules are worked on by a central team. So, the central team is about, let's say, 8 to 10 developers, core developers who work on the product who have been hired by organizations. But there are a huge number of volunteer developers who come in, contribute something, go back. And so, there's a coordinated communication that happens for this core product and platform. But there are over 300 publicly available modules, some in our own repositories, but others as source code in different places. And this is not really organized. It's like multiple flowers blooming in a garden and they're done and they go on without much idea of coordination, but it still goes on and people build modules and distribute this and implement it in different healthcare settings. So, there are modules that have moved between maintainers also without really, again, the coordination. So, people haven't really said that I'm going to maintain this, but they continue maintaining it, contributing to it and goes on evolving. This is some metric that I'd like to showcase from our last annual report. So, there are about 174 committers, 6,000 commits, 117 core repositories. But this is only what I said was our core development that we coordinate, that we work on or we as an open source or project think of as our repositories. But there are many more other repositories, like I said, that go on and happen on their own. And this is kind of our architecture. I won't get into the details of it, but what you see in the dotted box is kind of our core platform and what we really work on in those 117 repositories. This probably another distribution where you see on the side, which is called the reference application on the top left there. And that reference application is another product that we build. So, what are OpenMRS distributions? We took the idea from the Linux distributions where the kernel is a core product and we have multiple distributions serving different kinds of purposes. And OpenMRS is a flexible multi-layered system. And you have different modules that people configure or write custom modules to implement a different process in different healthcare settings. So, to implement OpenMRS, you typically start with an openMRS distribution. You configure that distribution and you do an implementation of your use case. But depending on the distribution you choose and what kind of use cases you want to meet, configuring metadata is sometimes done in other places. You have to write custom add-on modules or you write a custom UI over the back end that does not have just a UI and not really core business logic. So, it depends on your use case, depends on the distribution you've started with and that's the kind of effort you need to do. Implementations with strong development teams may choose to build their own custom applications on top of the platform. Another kind of distribution that we kind of build is called the OpenMRS reference application. I know the name kind of says that it's an application but it is basically a showcase of the platform. And so, it's a sample electronic medical record system. That sample electronic medical record system is for just showcasing, it's not for a particular use case but just for showcasing what could be built on top of the platform. And so, what we want to do is we start a set of content that has basic functionality like registration, reporting and things like that. And then we use Groovy and Java and combination of those to do this. So, in the community lingo, we call this distribution a ref app but there have been many, many more distributions that have come up recently and we see that there are a few other ones that are a challenge. So, coming back to the topic of so, coming back to the topic of why the Bazaar model is difficult and why I call this and why I call this and why I call this a difficulty or challenge is because the Bazaar model extreme Bazaar which is how I call this is there's an under performance in terms of efficiency. You've seen other communities building similar kind of e-health applications have progressed more rapidly but they've not been able to gather a community around themselves because they have followed kind of a cathedral model of development because the Bazaar model has these four main ideas that are a performance challenge. So, you waste a lot of resources of people doing different places but they do the same kind of things. So, there's multiple duplication of effort that happens. So, we at least know six different kinds of registration modules they do very similar things but because they've been created by different organizations they've all been creating their own code. They've been at least three different attempts with different three modules that do rest web services for the same platform. So, it's wasted of resources. The next is the undiscovered resources. So, a lot of people do this on their own on the side a few implementations in say one country are doing something that people have already done in another country. So, there are these many undiscovered resources that probably has generalized features but people still don't are not able to use it. There's lack of motivation to merge back and so at least from what I know because I was helping the proprietary company. So, they use the platform in a very large proprietary EMR but there's no reason to really contribute back and their market research showed that they'll not probably get any benefit out of community participation and friction and competition is the last part where implementations have a fear of their competition overtaking them by taking their work their modules and building upon it or selling it. So, they do copy left licenses. So, it's more of a fear reaction. So, how do you manage the Bazaar and this is kind of our approach that we've thought of. So, we've seen marketplace models that have been successful think of Alibaba's and other companies that sell different kinds of products together from different vendors. The regulation model is also common where you have regulation anyone who clears that regulation policies is then able to sell something in the market and that's the crowd source model where everything is acceptable but the users who rate different people with different ratings and then that is more visible than the other ones. So, we are planning a combination of all these models by creating a certification process and we're going to certify these different implementations or implementing partners, these different developers, these different trainers and all of this is going to happen in this year in 2016 and so we want to create service providers who will be able to implement this. We have an annual worldwide summit in 2016 in December in Singapore. We had one last year and we're going to come back. So, if anyone's interested in training or certification wants to learn about OpenMRS we'll be doing that in this year but why do we certify? And a lot of companies have made considerable amount of money with certification but that's not our goal. What our goal is that we'll only be doing this because we want to take the advantage of understanding the Bazaar well and bringing more visibility to the Bazaar that's available. So, there are different theories that talk about how the whole is greater than the sum of the parts and the experiment was about that and we'll see what results from that experiment but being able to create the whole of the OpenMRS community we'll be able to hopefully bring more visibility to these various different products that are around that are not coordinated but are very independent of each other but we want to make them visible to our users because we don't want to waste those resources and the problems that you saw. We've also found the service provide a visibility to be one of the biggest entrances in countries or large organizations wanting to take up OpenMRS and that's the reason why we want to certify. Coming back to our experiment sheet again so we had a few things happening here on sheet one we didn't create any word because what we are doing is we are just trying to add alphabets and we never created a word. How many of you thought what was the wild guess to the number of words you thought was going to get created? Any numbers? Six? Three. From this side of the row because you guys I guess started with sheet two right? Sheet one so sheet two so this side would their guesses would probably be the most relevant to us because the other ones have seen it seen what the how much we created but what was your guess? Given that there are about 20-25 people in the room. Six? Okay and so we've created here about three words and so what I'm trying to show here is yeah so what I'm trying to show here is and this probably not worked as well as thought but what happens is when you're trying to create something with a goal of creating that thing the amount of material that you create you don't know at the beginning of the creation process and so you need some kind of visibility a different lens to be able to understand what has been created and so this lens kind of tells us that we've created these three words and then you can create different processes around providing more visibility and that's what we're doing with our certification process is that we're creating visibility for the different activities that are happening so that's how we conclude this. You can tweet me at sunny tweets and if you have any questions then please ask yeah. Thanks very much.