Mycroft Holmes and the Knowledge Manager

Most organizations do not lack specialists. They lack a place where one specialist’s work can meet another’s before a deadline forces the introduction, usually at the least convenient possible moment.

Arthur Conan Doyle gives Mycroft Holmes a deliberately impossible job. Sherlock’s brother is described as the British government’s central exchange, a clearinghouse through which departmental conclusions pass, and Watson calls his specialism omniscience—which is a bad staffing plan even in fiction. What interests me is Mycroft’s position between domains: he does not replace the naval officer, foreign-affairs official, or financial specialist, but receives the conclusions each has made from a narrower view and makes a balance from their relation.

That is a useful way to describe knowledge management, especially when an organization has a digital asset management system, or DAM. A DAM is often introduced as a place to store and retrieve images, video, drawings, brand files, and other media. Storage matters. Permissions matter. Version history matters. None of those functions automatically produces the cross-department view that makes the collection useful.

Consider a project photograph. The design team may see an image that documents a material choice. Communications may see a possible case-study image. Facilities may see evidence of a maintenance condition. A client team may need it to explain a decision that no longer appears in the current drawing set. If the file is described only by project name, photographer, and date, each person must already know why to look for it. The system preserves the image and leaves the pattern invisible.

The knowledge manager’s question is different. Which work does this record touch, and who is likely to need it next? That question changes classifications, access routes, and the conversations around a collection. It may lead to a shared project vocabulary, links between a photo and a maintenance record, a route from a completed project to a design precedent, or a small review where people discover that identical evidence has been filed under incompatible terms. The point is not to label everything until the labels become their own weather system. The point is to make a relation usable.

Specialist silos are reasonable at the scale of daily work. A structural engineer should not be asked to keep the whole firm in mind while resolving a structural question, and a communications editor should not have to interpret every technical drawing before selecting an image. Trouble begins when the silo becomes the only retrieval path. Then a good file system can still produce repeated research, weak handoffs, and decisions made from partial evidence.

Mycroft’s clearinghouse offers a test. When two departments are working from the same underlying record, can either discover the other’s use without already knowing whom to ask? If the answer is no, the DAM may be functioning as a cabinet rather than a shared instrument. The file exists, but its possible meanings do not travel.

This does not require one central person to approve every connection. That would turn synthesis into a bottleneck with an impressive title. It requires a role, a practice, or a set of regular exchanges that can notice patterns crossing the normal boundary of a team. Someone needs to compare the terms, trace the repeated requests, inspect the gaps between collections, and ask whether a connection should become part of the system rather than remain private lore.

There is also a limit in the Mycroft comparison. A knowledge manager should not pretend to have perfect overview, and a DAM should not present an authoritative pattern where the records do not support one. The clearinghouse must preserve provenance: who made the record, what it described, when it was created, and what remains uncertain. Synthesis without those details can produce a convincing but false overview. A better system lets users travel from a cross-cutting insight back to the source material that supports it.

The practical work is modest and ongoing. Watch which assets are repeatedly requested across teams. Ask which terms make a file discoverable to more than its original owner. Notice which people serve as informal translators because the system has not yet learned their route. Then make one of those routes visible.

Mycroft is compelling because he stands where the departments meet. A knowledge manager earns the comparison by making those relations usable while keeping the underlying specialists, records, and limits in view; otherwise “omniscience” becomes a polished name for conclusions nobody can trace back to their source.