Interpret the causal drawing and vector comparison separately
The drawing above fixes B's local Lamport value at 5 before it receives A's message stamped 2. B's receive therefore has label 6. The arrow establishes the send-to-receive dependency. C's independent event has label 3, but there is no arrow path from A's send to that event or from it to B's receive. Numerical placement between 2 and 6 does not insert C into the message chain. Horizontal positions are schematic. They do not establish synchronized physical times or quantify how long the message spent traveling.
The nearby evidence table introduces a vector, a list of counters with one slot per tracked actor. Unlike one scalar label, it preserves which actor's progress is known. In these two-actor examples, the first component always belongs to A and the second always belongs to B. Relative to [2,0], vector [2,1] contains at least as much information in both slots and strictly more in B's slot. Vector [1,1] has less A information and more B information, so neither vector dominates the other. Adding their components would discard which actor supplied the information. The sums of [2,0] and [1,1] both equal two even though the histories differ.
Derive those relations with a separate two-actor paper trace. Begin at [0,0] on both actors. A has a local event and becomes [1,0]. B independently has a local event and becomes [0,1]. A then sends a message as its next event and stamps [2,0]. B merges that incoming vector with [0,1] by taking the maximum in each slot, obtaining [2,1], and increments its own B slot for the receive, obtaining [2,2]. The send precedes the receive because every send component is no greater and B's component increases.
Notice the intermediate [2,1] in that calculation. It is the merge result before the receive increment, not the receive event's final timestamp under this convention. This distinction prevents mixing conventions when checking a worked trace. Implementations that timestamp versions rather than every program event may increment at different semantic boundaries. Their metadata needs its own definition. The static table's vectors illustrate comparison; they do not assert that every listed vector is a timestamp from one particular complete execution of the drawing.
To explore the comparison rule, copy [2,0] and [1,1] onto paper. Raising only the second vector's A component to 2 produces [2,1], which now follows [2,0]. Raising only the first vector's A component to 3 leaves it incomparable with [1,1]. Neither larger sums nor larger first components decide the relationship alone. The page's drawing and table do not let you enter arbitrary events or animate these changes. These additional cases are manual calculations using the stated rules, not hidden capabilities of an interactive clock simulator.