Why We Run Three Rounds
To state the conclusion first: one round is not enough.
After completing a single three-stage consultation and addressing the issues raised, you might ask, "Is this sufficient?" In practice, it often is not. When you respond to the points raised in round one, new questions surface. Fixing one part can break the consistency between that part and another, or the correction itself can introduce something that needs to be checked again.
For example, suppose the first half of a document says "clarify the division of roles," and round one feedback leads us to rewrite that passage. If the second half still contains a line like "in cases where roles remain unclear, proceed as follows..." the document ends up contradicting itself. That kind of inconsistency is easy to miss while you are in the middle of making corrections. That is why another round becomes necessary.
The reason we do not stop at one round comes down to a simple fact: the more rounds you run, the fewer blind spots remain.
"Three Stages" and "Three Rounds" Are Different Concepts
A brief clarification on the terminology is useful here.
Three stages (the number of steps within one round) refers to the sequence inside a single round. One round consists of three stages: the external auditor's initial observations, parallel consultation with the seven internal subagents, and the external auditor's follow-up review.
Three rounds (how many times that sequence repeats) refers to how many times the entire sequence is run. In this series, the maximum is three repetitions.
Putting "three stages" and "three rounds" side by side can be confusing, so it helps to keep them separate in your mind. Three stages is about what happens inside one round; three rounds is about how many times you stack those rounds on top of each other. To use a cooking analogy: the three stages are like "wash, cut, saute" -- the steps within a single attempt; three rounds is how many times you run that whole attempt as a trial. Even though the phrasing sounds similar, the two concepts refer to different things.
What Changes With Each Round
After round one, we incorporate the feedback and move into round two.
Round two brings a different quality of questions. Round one tends to focus on broad checks: "Is this structure sound in the first place?" Round two shifts toward tighter checks: "Has the correction introduced any contradictions?" and "Does this align with the surrounding chapters?" The focus moves from a rough pass over the whole to a closer look at the revised details.
By round three, the number of issues flagged drops further. The major gaps have already been filled in rounds one and two. What remains are pinpoint observations: minor inconsistencies in phrasing, or subtle mismatches that a reader might notice. Comments at this stage sound like "this transition feels hard to follow" or "swapping the order of these two paragraphs would improve the flow."
We call this process bulletproofing (building decisions that hold up under pressure -- making them harder to knock down). Rather than trying to produce a flawless design in one pass, the act of finding weak points and reinforcing them in cycles is what builds real structural strength.
Why "Up to Three Rounds"
The word "up to" is there because three is the ceiling, not the default.
Running more than three rounds causes the level of detail being checked to become so fine-grained that decision-making stalls. The loop of correct, review, correct never closes. That state is not productive.
What we observe in practice is that three rounds is enough to settle the overall judgment. Round one surfaces the rough problems. Round two tightens things up. Round three catches the remaining details. After three cycles, the work is in a state where we can move to the next action.
Cases where a fourth round or beyond becomes necessary are cases where round one or two uncovered a fundamental need to redesign the underlying structure. In that situation, we do not restart from round one of the same version. Instead, we begin round one fresh with the revised design. In other words, something that has changed at its foundation is treated as a separate matter entirely.
Making "Never Stop at One Round" a Habit
Not stopping at one round is something we do consciously at first. Over time, a natural sense develops: "anything that passed round one still needs another look."
When round one returns a GO decision, that decision carries the weight of one round only. It is not until rounds two and three are complete that the decision is genuinely supported.
Once "never stop at one round" becomes a habit, the speed of decision-making stays the same while the quality of each decision rises. The log built up across three rounds also becomes a foundation for the next decision. By recording what was flagged and what was changed in each round, the same category of problem becomes less likely to carry over into the next design.