A knowledge base exists to enable the reuse of past information. Storing facts, judgments, grounds, and outcomes can reduce the need to repeat the same research and accelerate the next decision. But in a knowledge base that handles human judgment, storing only conclusions can omit important conditions. The reasons a judgment was appropriate include not only the information available at the time, but also deadlines, the period for which options remained available, and the temporal situation of the people involved.
Suppose a record shows that an organization changed the way it provided a service within a short period. Looking only at the outcome, this might be reused as knowledge that “under these conditions, change should be prioritized.” But the meaning of that conclusion changes depending on whether the deadline for maintaining the old method was approaching or whether it was a temporary judgment intended to avoid effects on users. Knowledge that has lost its temporal conditions risks being overgeneralized in another situation.
Treating human finitude as knowledge does not mean storing the general proposition that “humans are finite.” What matters is preserving, as conditions of applicability, what could not be recovered in each judgment, which options had deadlines, and at what point a decision had to be made while some information remained unavailable. This allows us to examine differences from the present instead of simply transferring a past conclusion into the present.
In this design, the meaning of updating knowledge also changes. An update is needed not only when new facts are added, but also when we learn that a judgment once considered reversible actually carried a large recovery cost, or that a temporal condition previously given little weight affected the outcome. A change in our understanding of the conditions of judgment itself becomes knowledge that can be used in the next decision, not only the outcome.
At the same time, this does not mean that every personal circumstance should be stored. A knowledge base is constrained by reusability, necessity, privacy, and other considerations. What should be retained is therefore not a complete record of an individual’s life, but conditions that could change the validity of a judgment. We need to select information that can make a difference in the next judgment, such as the presence of deadlines, the possibility of change, the degree of recoverability from failure, and irreversible effects on others.
This is where the connection to Internal Efficiency Expression emerges. If an efficient internal representation means only reducing the amount of information, temporal context may be removed as redundancy. But if preserving differences that can change human judgment is included as a condition of efficiency, information related to finitude becomes necessary structure rather than a target for compression. What should be retained is determined not only by storage capacity, but by which differences will be needed for future judgment.
For Sai＊aus, knowledge is not only for fixing past answers in place. It is for returning to a present question what was known in the past, what was unknown, and under what conditions a decision was made. Precisely because the person in the present is not necessarily in the same situation as in the past, comparison needs to include conditions of applicability rather than conclusions alone. In this way, knowledge becomes a foundation for understanding differences in judgment rather than a shortcut around human judgment.
AI can refer to vast amounts of the past, but humans live the present only once at a time. Within this asymmetry, the role of a knowledge base is not to command the present with past optimal answers. It is to make visible the conditions under which past judgments were valid and how they differ from present conditions, allowing humans in the finite present to choose again. What knowledge should remember is not only a conclusion, but also the boundary of the time within which that conclusion had meaning.
Not answers. Understanding.