Transforming Classification: An Introduction
Welcome to Transforming Classification, a blog sponsored by the Public Interest Declassification Board. President Obama has charged the Board with designing a more fundamental transformation of the security classification system. In response to his request, we are proposing new solutions that address the shortcomings of the current system and tackle the challenges of digital records. By reducing inefficiencies and increasing public access, our proposals aim to improve the classification/declassification’s system capacity to protect and serve the American people.
Every other Wednesday over the next eight weeks, we will post either two or three “white paper” synopses to the blog describing an element of our proposed transformation. White papers will address the following topics:
Paper 1: Using Technology to Improve Classification and Declassification
- Employ predictive analytics and context accumulation technology to encourage consistency in classification and efficiency in declassification
Paper 2: Reconsidering Information Management in the Electronic Environment
- Implement a universal metadata standard for classified electronic (e-) records and integrate information and records management functions with archival processing
Paper 3: Regularizing the Declassification Review of Classified Congressional Records
- Systematize and prioritize the declassification review of Legislative records at the NDC
Paper 4: Discretionary Declassification and Release of Contemporary National Security Information
- Encourage policymakers to consider the advantages of not classifying certain categories of information or declassifying before prescribed deadlines
Paper 5: Simplifying the Declassification Review Process for Historical Records
- Transfer agency declassification authority to the National Declassification Center (NDC) after declassification exemption timelines have lapsed
Paper 6: Stewardship of Our Classified History
- Preposition topics of historical significance for declassification and underscore the importance of preserving agency contributions to national security history
Paper 7: Information Security and Access in the Electronic Environment
- The role metadata and access controls can play in information security
Paper 8: A Half-Life for Historical Formerly Restricted Data (FRD)
- Converting certain categories of FRD to national security classified information after 25 years and allowing for its review for declassification
We encourage you to post your comments on these white papers under their respective threads and comment on the posts of others. Your thoughts and suggestions on these topics will be of great assistance to us as we finalize our proposals to the President.
Comment on “About Transforming Classification”
Thank you for conducting this public conversation. While I welcome the opportunity to participate, I would like to make several opening observations.
The initial scope of the White Papers hardly seems transformational. They are geared toward making the existing classification system more efficient and effective, not transforming it. That approach skips over several important questions:
Can today’s hierarchical classification system reasonably be adapted to a networked world? Or do we need something altogether new and different? (If we are unable to articulate what that something new is, we should at least recognize the need for it.)
Why haven’t previous recommendations to fix classification and declassification policy (including some of those of the PIDB itself) been put into practice? What are the real obstacles to change? How can we confront them directly, and overcome them?
What if the classification system is transforming our system of government faster than we can transform the classification system? Full accountability in national security affairs is now a matter for historians. Public access to the official policy record of the past few decades is closed in significant respects, and seems likely to remain that way if current trends are allowed to continue.
It would be unfair to expect the PIDB process to provide a full solution to what ails classification and declassification policy. But I hope that the PIDB will help to express the concern felt by many that the classification system today is hurting our country, that it has lost the confidence of a sizable fraction of the public, and that it is poised to get worse if it cannot be fixed.
This is a great idea. Unfortunately, the reef upon which all ‘simplification’ of classification systems founder is the sheer size of the problem.
It may be more practical to break the information into ‘easy’ and ‘hard’ pieces. Easy being things that decay rapidly in value with time or have no value without context (table entries), hard being things nobody has ever agreed on anyway. If a substantial percentage of the data can’t be put into the ‘easy’ category, then the problem is intractable in any practial sense.
Another part of the problem is that the category boundaries are hard – it either is or it isn’t – when the real world is a continuum and not that neatly bounded. Things are non sensitive originally, then as factors are added, they become more and more sensitive until they qualify for the special rules. However, once in a category, adding more factors doesn’t necesarily do anything, untill you reach the boundary of the next level. This is why removing some factors doesn’t necessarily de-sensitize something – you have to remove the right ones, and be sure to get all of them.
The solution may be finer cuts. Rather than set only two categories is set a ‘range’ of sensitivity that would be susceptible to computerized ranking or grading – and assign a score. If the scale was 1 to 1000, one would need to show that something was *significantly* more sensitive than something else to warrant a certain grade of protection. Applying the same safeguards to data of widely varying sensitivity forces classifiers to put things in the ‘safe’ bucket because there’s no graduated scale of protections that can be applied.
Alternatively, as suggest in the prior posting, the problem may not be in classification itself but rather the impact inherent in insisting that two or three categories be used to signify the safeguards to be applied to a myriad of possible combinations. Set the safeguards to be as variable as (and proportional to) the definitions and there may be some hope – computers excel at chopping data into infintely small chunks, we just need to chop the safeguards into matching slices. However, please let us not fall into the trap of hyper complexity (e.g., the Common Criteria) that end up with a rule set or solution that is more complex than the problem – and that takes half a lifetime of study to understand
The problem is complex but the soution must be simple or it just won’t happen, no matter how great it is.
Transformation based on efforts to simplify categories and new technology make sense in conference but fail mathematical underpinings of information granularity. What is shown here will be an exercise in futility.