My Corner of the Web
As the former Metadata Program Manager for the Department of Homeland Security (DHS), I have significant experience in using metadata and data standards (like NIEM) to improve information discovery, access and sharing. This was our number one concern in reacting to 9/11: improve information sharing so that information “silos” don’t prevent us from connecting the dots and preventing the next 9/11. For the most part, we were successful and my book about the experience, “Information As Product: How to Deliver the Right Information, to the Right Person at the Right Time”[i] discusses those lessons learned in detail. Additionally, a previous book about the Semantic Web, detailed how to better infuse semantics into our data to turn it into “Smart Data”.
As has been widely reported, DHS is creating a “Disinformation Governance Board”[ii] and people are equating it to the “Ministry of Truth” in George Orwell’s 1984[iii]. Clearly, the concern is that those in power should not get to define “truth” because their political bias will taint the process. In that concern lies the seed of the answer. Ask yourself, “How do we determine whether an educational institution is qualified to teach a particular subject?” The answer is that we “accredit” their processes to insure they have the proper amount of rigor to consistently achieve the stated educational objectives in accordance with best practices in the field. Thus, what is needed to determine whether something is disinformation is that content creators have an accredited, open process for creating information and that a receiver of such information can determine, via secure pedigree that the information has not been tampered with in transit. Let’s break this down because care must be taken at each step of the information production and consumption process. First, let’s examine the Information Production process:
Figure 1 The Information Production Process
Figure 1 depicts information production (top workflow) as analogous to physical product production (bottom workflow). First, we need to state the obvious – any organization serious about their information must have a formal and trusted information production process. Does your organization have such a process that could stand up to the scrutiny of an accreditation audit? It should. The central insight into the creation of information is that information is not data, it is created from data and augmented to satisfy a particular purpose for a particular audience. In fact, you should not even get to call it information if it was not created specifically to satisfy a particular audience. We know how to do this and anything less is a dereliction of duty. Furthermore, if you do not have a formal process to create information, then you certainly cannot create trusted information. Creating trusted information is a critical component in understanding and preventing disinformation.
So, now that we understand that we need a formal process to create “accredited information” – what are the elements of such a process?
· Authority – every large organization has a plethora of data sources but there are typically a small set of prime sources at the point of collection. Those “primary data sources” are your “authoritative sources” because they collect the data in its raw form. Knowing the source of your data is critical to knowing whether it is authoritative. Secondly, knowing whether your source is trusted is to understand how it collects or creates data, whether it ensures the quality of the data and whether it protects access to the data. Let’s digress for a moment to consider secondary sources of information like news reports as these are important to what most people consider “disinformation”. The central question there is whether the primary sources of information that the reporter is relying on is trusted and transparent. Furthermore, does that reporter attempt to interpret the information or are they merely reporting it? So, secondary sources of information have the burden of proof to disclose their primary sources of information in their reporting and this brings us to the second characteristic of accredited information – pedigree.
· Pedigree – the Art world has always been concerned about providing the pedigree of a piece of art. Specifically, how do you know a “Rembrandt” was actually painted by Rembrandt? You know that by keeping careful records that follow a painting from its creation, through any number of owners, to its eventual destruction. This is following it from its cradle to its grave. In terms of information, a process that captures pedigree captures the creation, integration and modification of the information from its cradle to its grave paying special attention to insure you do not reduce or eliminate its authority. Do you have a pedigree process for your authoritative data? For your enterprise information? For your enterprise knowledge? If not, how do you know if it is still authoritative? There are ways to seal and sign data to protect it once you have deemed it authoritative. There are ways to track changes to data. You just have to set up a process to implement these known techniques.
· Quality – the final piece of creating accredited information is to ensure its quality over time. What is data quality? There are multiple facets but it begins with clarity and ends with cleanliness. A data governance process is required to insure that your information is not ambiguous, that you have standard definitions for the terms in your information (and sometimes even formal, logical ontologies for deep semantic precision), and that these standards are shared and promulgated throughout the organization. Ambiguity is your enemy here. Finally, your governance process must insure your data is cleansed and thereby free of common data-entry errors like transposition, spelling, grammar, faulty default values, and lazy inputs. High quality data is data that can be trusted from causing processing errors in any of the data consumers (especially intermediary IT systems).
How do we implement this so that lay persons can trust the information we see on the internet? One idea would be to institute a “Gold Check” program for Accredited information. As depicted in Figure 2, we could model gold checks after Twitter’s blue checks and a browser’s lock icon for secure websites. Specifically, a content creator would get a gold check for accredited information where you could view the information’s authority, pedigree, and quality metadata.
Figure 2 Gold Checks for Accredited Information and Accredited Information Producers