Yes, this is an excellent point to consider. A conspiracy seems sexier, but it comes down to the human potential for error and the bureaucratic fear of torpedoing your own career (or worse, someone else's life) just because of a mistake in a menial task. If you were the redactor and the redactor's supervising manager, if something is a gray area and your lawyers aren't giving this their full attention, why not redact?
Of course, any one who is interested enough to file for documents may already know what he/she wants and will call you out (with a lawsuit) if it seems you've redacted things unnecessarily. But for other situations, it seems unlikely that the requester will either not go through the trouble of suing or not even know if what was redacted is worth suing over.
An example of a high-profile amusing snafu occurred during the Rod Blagojevich trial when someone using Adobe Acrobat forgot to complete the redaction process, allowing redacted text to be copy-pasted into another text editor: http://capitolfax.com/2010/04/22/blagojevich-hurls-allegatio...
Yes, that was a technology-error rather than one of judgment...but technology errors are much easier to double-check for and yet it's not an uncommon error in the digital age of public documents.
Overuse of classification labels is a huge problem. The incentives are skewed as you suggest, so there is no real benefit to spending the time required to make decisions on each and every individual fact or statement. Entire swathes of boring information are classified for no real reason and a large number of things that someone might use as useful data are sequestered in random agencies. The cost of complying with FOIA is not insignificant and over-classification jacks up the cost. I don't know if there is a really effective way to "fix" the situation without somehow changing the incentives.
Of course, any one who is interested enough to file for documents may already know what he/she wants and will call you out (with a lawsuit) if it seems you've redacted things unnecessarily. But for other situations, it seems unlikely that the requester will either not go through the trouble of suing or not even know if what was redacted is worth suing over.
An example of a high-profile amusing snafu occurred during the Rod Blagojevich trial when someone using Adobe Acrobat forgot to complete the redaction process, allowing redacted text to be copy-pasted into another text editor: http://capitolfax.com/2010/04/22/blagojevich-hurls-allegatio...
Yes, that was a technology-error rather than one of judgment...but technology errors are much easier to double-check for and yet it's not an uncommon error in the digital age of public documents.