An unmaintained knowledge system is a degraded asset, not a cognitive extension — active use is the criterion

Why This Is a Rule

Andy Clark and David Chalmers' extended mind thesis argues that external structures (notebooks, digital tools, knowledge bases) can function as genuine cognitive extensions — but only when they meet specific criteria: they must be readily accessible, automatically endorsed, and actively integrated into reasoning processes. A knowledge graph that you traverse daily, maintain weekly, and consult during decisions meets these criteria. The same graph, untouched for three months, does not.

The danger is the illusion of cognitive extension. Knowing that your knowledge graph exists gives you a feeling of expanded capacity — "I don't need to remember that, it's in my system." But if you haven't traversed the graph recently, the connections have gone stale in your mental model, new notes haven't been linked, and your actual retrieval probability approaches that of someone without the system. You have the comfort of extended cognition without the function.

This is link rot applied to the mind-tool interface. Just as hyperlinks break when pages move, the cognitive coupling between you and your knowledge system breaks when you stop actively maintaining it.

When This Fires

• When you realize you haven't opened your knowledge system in weeks but still feel confident about "knowing where things are"
• During knowledge system audits when evaluating whether the system is actually extending your cognition
• When you catch yourself saying "it's in my notes" but couldn't navigate to it without a search
• When deciding whether to invest in maintaining an existing system versus starting fresh

Common Failure Mode

Conflating possession with extension: "I have 2,000 notes in my system, so I have extended cognition." Possession is necessary but not sufficient. The question is: does the system actively participate in your reasoning processes? Do you traverse it, update it, and consult it as naturally as you access your own memory? If the answer is "sometimes" or "not recently," the system has degraded from extension to archive.

The Protocol

(1) Assess: When did you last actively traverse (not just search) your knowledge system? When did you last maintain it? When did you last consult it during a real decision? (2) If any answer is "more than two weeks ago" → the system has degraded. It's an archive, not an extension. (3) To restore extension status: commit to daily traversal (even 5 minutes of backlink browsing — see When opening a hub note, check backlinks for 2 minutes — surface forgotten connections), weekly maintenance (link new notes, verify hub accuracy — see Invest quarterly maintenance in your top 5% most-connected notes — they are your graph infrastructure), and active consultation during decisions. (4) If restoration isn't worth the investment → honestly downgrade the system's status in your self-model. Don't count it as cognitive capacity you can deploy.