> frivolously marked "top secret" in order to increase the perceived importance of bureaucrats.
This is a real problem, but I suspect the reason is more laziness than increasing power. That's no score card on number of classified documents one has access to.
Rather, it is time consuming and risky to determine classification level. No one is going to get in trouble for mistakenly over classifying something, but theoretically risks jail and fines for under classifying a document, not to mention the real harm that could arise from such an error.
> it is time consuming and risky to determine classification level
This!
I once had the opportunity to work on a project with DoD folks to help them write software to speed up the declassification process.
I was astonished by the current manual process. Imagine a decision tree 50 pages long that you have to apply by hand to each sentence of a document to determine what can be unclassified or not.
These were such nice people and overwhelmed by the work and genuinely trying to make it better so more information could be declassified, faster.
I realized it was a hopeless task. The only real solution is to move toward a world where next to nothing needs to be classified. To do more in the open. That needs to be the vision. Laws need to be adjusted so that these crazy complicated rubrics don't need to be created at all.
[The end of the story is I opted not to join that project. LLMs provide at least some hope that they can make it somewhat better, lacking the changes at the legal and organizational level that I mentioned]
This is a real problem, but I suspect the reason is more laziness than increasing power. That's no score card on number of classified documents one has access to.
Rather, it is time consuming and risky to determine classification level. No one is going to get in trouble for mistakenly over classifying something, but theoretically risks jail and fines for under classifying a document, not to mention the real harm that could arise from such an error.