Precision Is Not Accuracy
Precision and accuracy get used interchangeably in casual conversation, and that habit causes real trouble in GNSS work, where the two can point in completely different directions at once.
Precision describes how tightly repeated measurements cluster together. Accuracy describes how close those measurements sit to the truth. A solution can have excellent precision and poor accuracy at the same time, and RTK FIX guarantees only the first of those two properties.
A Familiar Illustration
Picture a target shooter placing five shots within a single, tight, one-inch cluster, dead in the target's upper corner, a full six inches from the bullseye. That group is highly precise: the shots agree with each other almost perfectly. It is not accurate: the group does not agree with the target's actual center. RTK FIX behaves the same way. It guarantees a tight cluster. It says nothing about where that cluster sits relative to the truth.
A receiver's internal precision estimate comes from the statistics of its own least-squares solution, which assume the remaining error is random noise scattered around the truth. That is exactly why the math has no way to flag a systematic bias: a consistent, repeating error and a genuinely correct one produce identical internal statistics, since nothing in that calculation compares against anything outside itself.
How This Happens in an RTK Solution
An RTK rover measures its position relative to a base station with extraordinary precision. If that base station's own coordinate is wrong, every rover position computed from it inherits the same error, consistently, repeatably, and invisibly. The rover will return to within a centimeter of the same wrong answer every time it occupies the same point, because the carrier-phase solution genuinely is that precise. The precision is real. The accuracy depends entirely on something the carrier-phase solution has no way to check: the correctness of the reference it was measured against.
A rover that repeatedly returns to the same coordinate has not proven that coordinate is correct. It has proven the measurement is repeatable. A shared, systematic error, in the base coordinate, the datum, or the reference frame, will reproduce itself with exactly the same apparent confidence every time.
The Question That Actually Distinguishes Them
Precision is measured by comparing a solution against itself, repeating an observation and checking how tightly the results agree. Accuracy is measured by comparing a solution against something external and independently trusted, a known control point, a different correction source, a different survey method entirely. Confusing the two means mistaking a tight cluster of shots for a bullseye, without ever having looked at the target.
A rover occupies the same monument five times over an hour and gets nearly identical coordinates each time. Has that demonstrated the coordinate is accurate? No. It has demonstrated the coordinate is precise. Accuracy requires comparison against something outside the rover's own repeated measurements.
Precision measures agreement among repeated measurements. Accuracy measures agreement with the truth. RTK FIX guarantees excellent precision. It says nothing about accuracy on its own, because accuracy can only be established by comparison against an independent reference.
So far this companion has assumed FIX itself is genuine. The next chapter examines a more unsettling case: a receiver that reports FIX with complete confidence, while having resolved the wrong answer entirely.