This blog is a part of International Justice Monitor’s technology for truth series, which focuses on the use of technology for evidence and features views from key proponents in the field.
Information and communications technologies coupled with new patterns of human behavior have created an unprecedented amount of data. From the sensors whipping around earth at tens of thousands of miles per hour, to the billions of devices in humanity’s collective pockets, never before has history been so granularly documented. As such, never before has evidence of criminal acts seized upon by the international community through tribunals, including the International Criminal Court (ICC), and Commissions of Inquiry been so plentiful for use in securing justice for the most heinous of crimes.
Yet, in keeping with the emerging theme of this series, there remain key impediments to the technology-for-truth imperative. From the perspective of a so-called “first responder” investigating international crimes, these impediments can be parsed into three issue areas.
The seminal challenge to leverage the rich, detailed, and expansive universe of data being produced every day to justice and accountability is one of awareness: how do we know a piece of information exists that addresses a question of fact? Traditionally, investigators have sought to uncover eyewitnesses, victims, and whistleblowers to advance investigations, but what if the “witness” is a mechanical sensor? In some ways this is preferred; sensors have perfect recall, are impervious to intimidation, and cannot be retraumatized. However, if the sensor is no one’s sister, neighbor, or friend, how do investigators become aware of this data?
To determine whether probative remote sensing data exists—in, say, northeastern Nigeria, North Korea, or Central African Republic—one need only query a database, meticulously maintained by commercial providers. Issues arise when the data is a piece of “user generated” content (UGC). The growth of citizen journalism and the new abilities for people of all strata of society to document the actions of others is changing power dynamics, and there is growing recognition that UGC will revolutionize efforts to secure accountability for international crimes. Investigator awareness of relevant materials, however, remains a challenge. Though open-source intelligence (OSINT) relies on public information, one should not equate “public” with “readily discoverable.”
When people generate content, we often do so in a way that allows others to discover. Maybe we use a hashtag on a tweet or note the date and time of a video or photo that we share. Maybe we consciously or unconsciously have geotagged our content. Investigators (and others) can gain awareness of such content through familiar use of search terms and automated lists, geo-fences, and traditional social networks.
The problem is that most of the data humanity produces is unstructured. Unstructured data cannot be readily queried. It is an amorphous blob of bytes with no set form and undefined substance. In a sense, structuring data is what human rights investigators do: they compile, sort, and annotate information in the context of legal framework that defines elements of crimes. Investigators, however, cannot know the evidentiary value of data that is so unstructured as to not meet the basic definition of ‘information.’
For all of the much-touted promise of bigger data available to investigators seeking linkage evidence, corroboration of testimony, or impeachment of alibis, the truth is that most of it remains inaccessible and every bit as buried as incriminating documents in a safe. First responders are working to address this challenge through innovative approaches to the structuring problem, from social computation and micro-tasking to artificial intelligence and algorithmic approaches—subjects for a future post.
Chain of Custody and Accessibility
The adage that materials on the internet are forever is false. Though many individuals take on great risk to share materials, uploaders of content may get cold feet or may have their accounts compromised. Often, content of most use to investigators violates terms of a content host. For as many reasons as there are to upload content in the first place, there are more reasons it disappears.
In this environment, the investigator is burdened with an obligation to secure materials, even those that are “public.” Putting aside the cost, technical, and security challenges — as well as the terms-of-use complications of securing and storing large amounts of material — the simple fact is that many first responders have only limited experience in the security of physical evidence. I would offer that these materials, which include UGC, are essentially physical, as evidenced by the fact that corruption of evidentiary containers — whether files or media — can render a piece of digitally-encoded evidence worthless. Within Amnesty International and across the field of first responders, investigators have had to develop new handling protocol and “duty of care” standards to ensure the integrity of the evidence increasingly forming the bases for findings of fact. The evolving obligation to secure these materials — as law enforcement would secure evidence in a locker — remains a significant logistical and coordinative challenge.
While it is clear that investigators require such an “evidence locker” to secure discovered materials, it is less clear how to do so. Often, a piece of OSINT or found material has no evidentiary value in isolation — it is only upon triangulation that such value becomes apparent. In my experience, other investigators — maybe operating under different mandates — have interest in the materials informing our research. Like any evidence locker, there must be a system of organization and classification that will allow others working toward justice to use it. This poses a considerable collective action problem, the solution for which requires collaboration within and beyond the community of first responders.
As mentioned in previous posts, leveraging technology for truth requires forensic standards and methodologies that simply do not exist for the vast majority of new data available to investigators. Of course, as new data types increasingly form evidentiary basis for determinations of fact, first responders and the ICC alike are investing in the creation of standardized methods, toolkits, and standard operating procedures, whether for UGC, satellite imagery, or found materials.
Importantly, assessing the veracity of a piece of data, such as a photo, tweet, or video, is only the first step in assessing evidentiary value. Investigators accustomed to tracking down witnesses and gathering testimony are increasingly coming into possession of materials that require specialization in content analysis, such as forensic pathology, ballistics, or geology, in addition to the specialization required in assessing the meta-content — the digital containers in which we increasingly find our evidence.
The sciences, of course, do have standards. The application of methods derived from best scientific or peer reviewed practice will hold up in judicial proceedings; but if the provenance of materials are not demonstrated, the forensic analysis of content is meaningless. In the case of remote sensing data, the task is easy enough, as anyone can access commercially available data. However, if the evidence in question is a video of — for instance — a military commander inspecting a group of child soldiers, the video must be validated as genuine, regardless of whether all experts agree that the children are obviously so.
As we would evaluate the testimony of a witness with further questions, first responders must often probe pieces of digital material, handicapped by the material’s ability to provide only a limited number of responses to potentially impeaching questions.
This challenge is with us for the foreseeable future, and it points to a silent arms-race. Simply put, the OSINT environment is hostile to truth. False information is widely disseminated, often intentionally and through coordination. At Amnesty, we have come into possession of materials — both OSINT and not — that have been altered, fabricated, and obfuscated. Although we’ve adopted methods to spot such fabrications, the conflict between those who seek to control the narratives of international criminal tribunals and those who seek to uncover truth in the pursuit of justice and accountability is an evolving one and always will be. This places significant implied future costs at the feet of first responders who must develop ever evolving detection tools to keep pace with the latest advances in propaganda and control.
Finally, in this adversarial environment, the transparency of methods used by investigators analyzing and assessing digital materials becomes a potential future liability, pitting the demands of fair and transparent judicial proceedings against the incremental and occasionally fleeting advantages investigators have over those who would conceal or obfuscate truth…a tension as old and universal as cops and robbers.
Scott Edwards is an adviser to Amnesty International’s Research Directorate and is Professorial Lecturer at George Washington University’s Elliot School of International Affairs.