There is a particular kind of person on every team who answers fast. Ask them a question in the morning and you have an answer before lunch. Their reply is usually right, usually complete, and usually saves you an hour of reading.

Nobody ever decides to route work through this person. It happens the way water finds a channel. Each individual choice is correct — of course you ask the person who knows, of course you ask the person who answers — and the sum of those correct choices is a system with a single point of failure that nobody designed and nobody is tracking.

The Reward Is the Trap

Fast answers get rewarded, so the behavior intensifies. The fast responder gets more questions, which makes them better informed, which makes their answers better, which attracts more questions. Within a year they know things nobody else knows, not because they were given special access but because every ambiguous situation passed across their desk.

From the inside this feels like being useful. From the outside it looks like the team is doing fine. Neither view shows the actual state, which is that a growing share of the team’s throughput is gated on one person’s attention, and that person’s attention is the one resource that cannot be scaled by hiring.

The failure mode is not that they leave. Everyone plans for that badly but at least they plan for it. The failure mode is subtler: the fast responder becomes so busy answering that they stop doing the work that made their answers worth having. They are still the expert, but the expertise is aging. Six months of answering questions about a system is not the same as six months of working on it.

Interruption Is Not the Cost

The usual framing is about focus — interruptions are expensive, context switching costs twenty minutes, protect your maker time. That framing is real but it is small, and it puts the problem on the interrupted person to solve with calendar blocks and status indicators.

The larger cost is structural. Every question answered in a direct message is a piece of knowledge that transferred to exactly one person and left no trace. The asker got unblocked. Nobody else learned anything. The next person with the same question asks the same person, who gives the same answer, slightly faster and slightly more tersely each time.

Measure a team by how many of its questions are asked twice. A high number does not mean people are inattentive. It means the answers are landing somewhere unsearchable.

What Actually Helps

Not “stop asking.” The asking is fine; the asking is how work gets done. What helps is changing where the answer lands.

Answer in the channel, not the direct message. Same effort, and the answer becomes findable. If the question came privately, answer privately and then post the general form publicly.

When you notice you have answered something twice, stop and write it down before you answer the third time. The second occurrence is the signal. Waiting for the fifth means four people got a private answer and the fifth gets an irritated one.

Occasionally answer with a pointer instead of a solution — not to be difficult, but because a person who has been walked to the answer once can find it themselves next time, and a person who was handed the answer cannot.

And if you are the fast responder: notice when the questions stop being about things you currently know and start being about things you used to know. That is the moment the channel has outlived the water in it.