| |
|
|
|||||||||||||||||
| |
|
|
Martus development, marketing, and implementation plans are based on extensive research — in the lab, and in the field.
What is Martus?What is the Martus™ mission?Martus is a technology tool to assist the human rights and social justice sector in the collection, safeguarding, organization and dissemination of information about human rights violations. Martus brings the power of technology to the field by placing a simple yet powerful tool into the hands of human rights defenders around the globe. With the Martus system, documentation of human rights violations will be safeguarded and disseminated, accelerating response to violations and in some cases preventing additional abuses. Who is building, marketing and implementing Martus?The Martus Project is a major initiative of Benetech, Silicon Valley's leading high tech nonprofit. We combine the technology enterprise model with the social mission and heart of the nonprofit sector. As such, we see the grassroots organizations in the human rights sector as Martus customers and seek to build tools that address their specific needs. Who uses Martus?The Martus user community is made up of three distinct audiences: information producers, trusted intermediaries and information consumers.
How does the Martus team know what these customers need?Before and during product development, the Martus team met with numerous international human rights organizations such as Human Rights Watch, Lawyer's Committee for Human Rights, Open Society Institute, Open Society Archive, United Nations and AAAS, as well as grassroots organizations from more than a dozen countries. In addition, we conducted market research in Russia, Sri Lanka, Cambodia and Guatemala, incorporating the recommendations of human rights activists into the product design. After initial development, Martus entered an extended test period with field-testing in Sri Lanka, Guatemala, Russia and Eastern Europe. By working closely with grassroots and international human rights groups during the conception and development of Martus and following up with these and other human rights groups post-development, Benetech has created a tool that addresses the needs of many organizations in the field. How much funding did Martus require?Martus required an initial investment of approximately $650,000 for development of the software and initial outreach and training campaigns. This has been provided by major grants from Aspiration/Open Society Institute and high technology leaders, as well as significant funding from Benetech. Additional funds are required over the next two years to continue enhancing the technology and provide outreach, training and support to greater numbers of users. Consequently, Martus will be funded by revenues generated from providing customization tools for larger NGOs and from providing ASP and training services for server operations. These revenues will cover the costs of continued open source development on a limited scale. Does Martus accept government funding?The U.S. State Department, which acknowledges the importance of strong encryption to protect human rights data, has provided funding for the development of an additional Martus security feature that erases data from a users local computer in the event of an adverse security incident. The U.S. State Department also provided funding for initial Martus translations in French and Arabic and Martus trainings in Kenya, Egypt and Nigeria. Regardless of the source of funding, Benetech controls and manages all Martus software development which is always undertaken with open source software tools that support source code review. Martus funders and Benetech staff do not have the ability to view human rights data encrypted by Martus or stored on secure Martus servers. Martus users are free to use any of the publicly available worldwide Martus servers to store their uploaded information, or develop their own backup solution for the bulletins. Benetech encourages Martus supporters to conduct code reviews of the open source Martus software to ensure trust and transparency. How do we know that Martus software is safe for us to use?Our decision to make Martus open source, free software was driven by the desire to ensure transparency. If anyone wants to know exactly what our software does, s/he can review the source code and compile the program for him or herself. What is the central product of Martus?Most grassroots organizations focus on the development of one central asset: information about the human rights violations they are organized to combat. Like any industry sector where data is the product, human rights monitors and activists must have dependable, secure systems in place for the collection, storage and retrieval of raw data. Once this data is available, information consumers such as activists, prosecutors, the media and academics, can analyze the data, study it and publish reports about it or take direct action through channels such as prosecutions or investigations. To provide an end-to-end system for human rights abuse information processing and output, Martus will ultimately consist of three key software components:
What kinds of human rights violation reports are we talking about?Martus will be used to:
Is Martus an activist organization?No, Martus is not itself an activist organization or an urgent action network, but rather a tool addressing the specific technological needs of the human rights community. The Martus Project is a nonpartisan initiative that acts solely to provide a technological service to help human rights information organizations. Martus is not affiliated with any government and is independent of the nongovernmental organizations that it serves. Martus Technology and SecurityHow does technology help human rights?With Martus, human rights NGOs will have access to the information technology tools they need to manage their most important asset: information about human rights abuses. By carefully making the tools usable to someone with basic experience with electronic mail, the technology currently available to more powerful entities will be made accessible to the grassroots activist community. NGOs need the ability to 1) create files using a consistent model that can be formatted to satisfy their reporting needs, 2) store and secure reports and 3) access and retrieve files efficiently and securely. The need for secure storage and backup to counter possible attacks on saved information is a vital component of the Martus solution. During field research, we found that threats to information can vary from equipment theft to user error to environmental factors. At one human rights organization in Sri Lanka, termites ate seven years of collected violation information that was stored in paper form! As another example, Guatemalan NGOs are currently experiencing a string of computer thefts. Any solution must take these real life examples and more into consideration. Dissemination of information, both locally and globally, is vital to human rights NGOs. Groups need to share within their own organizations, both locally and throughout the world, as well as with the media, international NGOs and other information consumers. These other consumers include emigrants, activists, foreign ministries and citizens of the country where the group is active. Martus will make all of this possible. Is Martus easy to use?Yes. Because potential Martus users have varied levels of technical ability, Martus tools are built on very basic platforms and systems. The software design elements of Martus technology are aimed at the user who has basic computer skills, such as the ability to use a simple email program. What does free/open source have to do with Martus?What is open source software, and why is it important? Open source software can be loosely defined as software building blocks that are distributed free of charge as either source code or in compiled form. The basic idea behind open source is very simple: when programmers can read, redistribute and modify the source code for a piece of software, the software is more trustworthy and evolves more quickly than in proprietary software development. Learn more about more about free software here, and about open source here. Martus tools are designed as free/open source software programs, built on top of open source tools. Martus customers see three distinct advantages to the open source implementation:
What is encryption, and why is it important?Encryption is a central component in the preservation of human rights violation information. Martus™ software – on the client side – will incorporate encryption capabilities to maintain the integrity and security of the bulletins created and reported by NGOs. This encryption takes place at the initial stage of bulletin creation, before it is transmitted over the Internet. Human rights groups have occasionally employed cryptography on a limited basis, but currently available technology such as PGP (Pretty Good Privacy) has not become widely successful because of its complexity. Martus provides the simplest possible interface for encryption, while using strong technology inside its tools. The goal of using this encryption technology is to offer the NGOs various means of protecting the information collected through best-of-breed information technology practices. The benefits Martus provides go well beyond peace of mind and include secure storage, offsite replication and information sharing technologies – all with transparent encryption technology. Using Martus DataCan Martus be used to do statistics?Perhaps the first question to ask is What do you want to count? For example, do you want to know the the number of bulletins you have recorded? The number of victims? The number of violations those victims suffered? You can use the Sorting and Search functionalities in Martus to count the number of bulletins you have. You can break the total into various "disaggregated" subtotals, such as how many each author has generated, how many bulletins have specific keywords, or how many are tagged as coming from each specific location. Martus searches the entire text ofyour bulletins to find your search terms, but you need to be careful to use standard sets of terms in the bulletins to ensure that you will find the bulletins related to your subject of interest. For example, if a bulletin contains the word "jail," and you search on "prison," the bulletin will not be found. You can use the Martus keywords field to enter synonyms to help in this kind of searching. In future releases of Martus, we may add the ability to automatically search using synonyms. Searching in this way is very much like doing an Internet search: some of the bulletins you find are not the ones you are searching for, but others that you are searching for might not appear the first time you search. We are currently building reporting tools for Martus that will enable you to further reorganize the qualitative, textual information captured in your bulletins to create reports with different formats. For example, you may want to create a tabular report showing a summary subset of bulletin fields for all bulletins created by a particular author during a certain date range. You could then count the bulletins you find in your report. This is counting bulletins, which is a useful measure of what your organization is doing. However, it is *not* statistical analysis of the human rights violations documented in the text of the bulletins. Beyond the number of bulletins, counting violations and victims gets much more complicated. We strongly recommend that you do not attempt to use Martus for this task. Instead, see "Different tools for different purposes," below, to learn about a tool that can be used for human rights statistics. In order to accurately quantify data about human rights violations, it is necessary to manage the following complexities: duplicate reporting by multiple sources, representing the structural complexity of human rights violations, and consistency in meaning and counting, described below. When you collect narrative statements about human rights situations, many different narratives may describe the same events. That is, you may have had the same killing reported to you by five different witnesses or sources. When trying to count the total number of abuses, it is critical to distinguish where reports overlap so that violations or victims are not over-counted. Detecting duplicate reporting can be very difficult when you have more than few hundred cases. It is important to go through a process called matching (sometimes called "record linkage") to identify repeated victims and violations, and then manage the information from the multiple (sometimes contradictory) sources. This is a core challenge in generating human rights statistics. To learn more, see Source and Judgment. It is important not to lose or delete duplicate reports. In fact, information for overlapping reports can be extremely valuable. At the Benetech Human Rights Program, we have pioneered the application of a statistical technique called Multiple Systems Estimation (MSE) for human rights data analysis. MSE uses the pattern of overlap between data-gathering projects, or systems, to make inferences about how many deaths were never reported to any project. We have employed MSE in human rights data analysis projects in Guatemala, Kosovo, Peru and most recently in East Timor. Representing the complexity of human rights violations In addition, there is a considerable amount of complexity that must be managed when counting victims and violations:
There are other complicated relationships among elements in a human rights narrative, but we call the connections listed here the "who did what to whom" data model to maintain the integrity of the data elements. For more details on this kind of reasoning, see "Who Did What to Whom? Planning and Implementing a Large-Scale Human Rights Data Project." It is critical to capture the complex relationships between the different roles (a person can be a witness, victim and/or perpetrator within a sequence of events) and events to be able to accurately reconstruct which victims suffered which violations committed by which perpetrators - simplifying these points leads to distorted statistical results. Consistency in meaning and counting Before an information stream can be used for data analysis, it has to be in a reasonably regular format or structure, as described above. But even more importantly, the information needs to have a regular meaning. To convert qualitative information to data, we have to classify the qualitative information in consistent, repeatable definitions. This process is called "coding." For example, what distinguishes "rape" from "sexual abuse"? The two categories must be defined very clearly so that all the people doing coding in the team apply the definitions in a standard way. That is, the definition must be so clear that if you take the same narrative statement and give it to all the members in your group, they would classify it in precisely the same ways. We also refer to this as defining a controlled vocabulary. This is hard work. There are other hard questions to be asked. If a narrative describes an event in which someone dumps one barrel of toxic waste in a farmer's field, we might say that this is one act of "environmental destruction" What if two barrels are dumped? Is that one or two acts? How do you define a standard unit of analysis of "environmental destruction"? The rules that define when one violation has occurred, and when we find a second violation, are called "counting rules." Different tools for different purposes We designed Martus to be an easy-to-use tool for gathering and securely storing sensitive field information about human rights abuses. The information saved in Martus can be sorted and searched. Martus can also be extensively customized to organize information in any format a group may require. At the Human Rights Program at Benetech, we have developed another tool called Analyzer that is specifically designed to organize human rights data for statistical purposes. Analyzer facilitates managing the challenges involved in structuring and quantifying human rights data described above. From our experience, statistical projects that would benefit from using a tool like Analyzer require a considerable investment in technical development and thinking about exactly what the organization intends to analyze. Martus and Analyzer are complementary tools. We recently completed the initial steps on a Martus/Analyzer "bridge" to integrate them. The bridge enables a project to securely import and display data collected in Martus bulletins into a format ready to be coded into Analyzer. Please see I have lots of Martus bulletins: now what can I do? for more details. It is hard to do statistical analysis that will withstand serious criticism. The Human Rights Program at Benetech helps groups to design and build human rights projects that produce statistical knowledge about violations. We have experience advising statistical projects at seven truth commissions, several UN missions, two tribunals, and many NGOs. From our experience, these types of projects require a considerable investment in technical development and thinking about exactly what your organization intends to analyze. If your organization has collected several thousand Martus bulletins and would like to work on statistical analysis, we would like to help. Please contact us at info@hrdag.org. Can Martus make graphs of my data?Graphs are visual representations of quantitative data. Creating a graph means first creating statistically rigorous data. Therefore, the challenges in the creation of statistical materials (Can Martus be used to do statistics?) apply in order to prepare qualitative information stored in Martus for quantification and subsequent visual representation. As described in Can Martus be used to do statistics?, Analyzer can be used to facilitate structuring and quantifying human rights data. Once you have coded your data, based on a controlled vocabulary, and matched duplicate victims and violations (please see Duplicate reporting in the previous section, Can Martus be used to do statistics?). Analyzer can export your dataset carefully with the proper shape and definition to use in other tools to create statistical or visual analysis. We mostly use statistical software, like Stata and R, to create graphs. Other examples include ATT Labs' graphviz and ArcView. (It would also be possible to export the processed and matched data from Analyzer to a spreadsheet to create graphs. We recommend against using spreadsheets because spreadsheets are hard to audit and don't scale well beyond a small handful of graphs or tables; we find that in most of our projects, we create hundreds and sometimes thousands of graphs as we explore the data.) I want to do GIS (Geographic Information Systems) mapping with my bulletins. Can Martus help?You can use Martus to capture GIS information in the body of bulletins or with a custom field. This data could then be exported from Martus to XML and you could create a custom application to do mapping. It is important to consider carefully *what* it is that you are mapping. Like the previous question, GIS mapping usually implies underlying quantitative data. Without properly processing the data in the ways described above, aggregating GIS information, like any other type of information, could compound underlying data problems, such as duplication. In addition, you would be mapping what was reported to you, or "what you know," not "really what happened." That is, Martus does not provide a way to make a projection (or an inference) from the reported information to the true patterns in the population. The result could be an "artifact," patterns that seem interesting though the underlying reality could in fact be different or exactly the opposite. There would be no way to assess the relationship between the data and the real world. At the Human Rights Program at Benetech, we have developed ways to apply statistical techniques to human rights data to make projections about the total number of human rights violations, reported and unreported. Please see Can Martus be used to do statistics? for more details. GIS mapping could be done in the context of statistical analysis project. The Martus/Analyzer bridge can be used to securely import and display Martus bulletins in the human rights database, Analyzer. The data from the Martus bulletins, including GIS information, can then be processed in Analyzer in preparation for statistical analysis. See Can Martus be used to do statistics? for more information on the steps required to accurately quantify human right data. Custom GIS tools could then be used to map the results. I have lots of Martus belletins: now what can I do?What is "lots" of bulletins - 20 or 20,000? When thinking about undertaking an analysis project, it is important to consider whether you have enough data to make the effort required to prepare it for statistical analysis worthwhile. A small amount of data is not likely to provide much coverage of the events you are trying to measure. Most of the projects we have worked with have started from several thousand stories or testimonies. As mentioned in Can Martus be used to do statistics?, you can use the sorting and Search functionalities in Martus to count the number of bulletins you have with various characteristics such as author, keywords, etc. Martus searches the entire text of your bulletins to find your search terms, but you need to be careful to use standard sets of terms in the bulletins to ensure that you will find the bulletins related to your subject of interest (e.g. If a bulletin contains the word "jail" and you search on "prison," the bulletin will not be found. You can use the Martus keywords field to enter synonyms to help in this kind of searching.) In future releases of Martus, we may add the ability to automatically search using synonyms. Upcoming reporting tools for Martus will enable you to further reorganize the qualitative, textual information captured in your bulletins to create reports with different formats. For example, you may want to create a tabular report showing a summary subset of bulletin fields for all bulletins created by a particular author in a certain date range. Once you have accumulated several thousand bulletins, you might want to consider doing statistical analysis. The Martus/Analyzer bridge can be used to securely import and display Martus bulletins in Analyzer. The bridge also has the ability to pull customized formats from Martus bulletins and display them in Analyzer to facilitate coding. The data from the Martus bulletins can then be processed in Analyzer in preparation for statistical analysis. See Can Martus be used to do statistics? for more information on the steps required to accurately quantify human right data. We have worked (and are currently working) on projects in various
countries where information from various sources has been brought
together for analysis. Please see our Following its launch in January 2003, Martus
is being introduced to the human rights community through outreach and
training over a three-year
period. We are pursuing media coverage to help ensure that Martus is
introduced to and used by the widest possible range of organizations.
Martus outreach, customer support and training teams represent Martus
at human rights conferences internationally and engage with NGOs that
express interest in Martus. We also approach NGOs individually to help
them adopt Martus as an information management application. In addition,
we will collaborate
with human rights
information users and aggregators to demonstrate how the Martus Search Engine
will assist them in researching, analyzing and distributing violation
information. The Martus™ software will be disseminated
throughout the human rights community free of charge. During the initial
rollout stages, Martus staff has been offering training and installation
services at no charge for the initial adopter groups
unable to pay for these services. Martus needs to achieve critical
mass to have the widest impact, and providing these direct services
to beta
and early users is essential to building the core user base. Benetech
leverages the expertise of highly qualified trainers through the "train-the-trainer" model.
Our longer-term plan is for Martus technology to become part of the core
toolkit of NGO technology
staff and to phase out Benetech training efforts over time. At that
point, the community of Martus users will grow rapidly each year and
the reliance on professional outreach by Martus staff will decrease. Martus software is complete, and we began formally introducing
the software to the public on January 15, 2003. The Martus Search Engine,
which allows groups to publish the public portions of bulletins to the
Internet in searchable formats, became available in the third calendar
quarter 2004. |
|
|||||||||||||||
| |
|
|
|||||||||||||||||