Every boundary you have ever seen on a screen is wrong. Not metaphorically wrong, not wrong in a way that requires a philosophy degree to appreciate. Wrong by a measurable distance that nobody displayed to you.
That is not a criticism of the maps. It is the nature of the thing. A boundary is a legal concept, a line on a screen is a rendering, and there are at least four independent sources of error stacked between them. Knowing what they are turns a boundary from something you trust into something you can reason about, which is considerably more useful.
Error one: the original survey
Most of the American West was laid out by contract surveyors working on foot in the nineteenth century, dragging a metal chain across ground that was frequently steep, frequently wooded and occasionally hostile. They set monuments at section corners: a post, a stone, a marked tree, a mound of earth.
They were mostly conscientious and the work was mostly good. It was also done to nineteenth-century standards with nineteenth-century equipment, and some of it was rushed, some was estimated, and a small but genuine amount of it was never actually walked at all.
Here is the part that surprises people. Where the original survey was wrong, the original survey generally still governs.
American boundary law works on an order of priority, and monuments outrank measurements. If the field notes say a corner is eighty chains west and the actual original monument sits seventy-eight chains west, the corner is where the monument is. The recorded distance was a description of where the surveyor thought he was. The monument is where he actually was, and the land was granted by reference to it.
This is a very old and very deliberate rule. Its purpose is stability: everyone who bought land relying on that corner keeps what they thought they bought, rather than the whole grid shifting every time measurement technology improves. It also means a modern GPS measurement, however precise, does not relocate a boundary. It measures one.
Original Monuments, Retracement Monuments, and First SurveysError two: the coordinate system
The second error is invisible and it is the one that catches technically minded people out.
A coordinate is meaningless without a datum, which is the model of the earth's shape and the reference frame the coordinate is expressed in. Change the datum and the same physical spot gets a different number.
The United States has used several. Older maps and older surveys reference a datum from the middle of the last century. Most modern American mapping references a newer one. Global satellite positioning uses a global reference frame of its own. The differences between these are not trivial. Between the older national datum and the current one, the shift is commonly on the order of tens of meters, and how large it is depends on where you are standing. Between the current national datum and the global one used by satellite systems, the difference is smaller, on the order of a meter or two in the lower forty-eight, and it changes slowly because the continent is moving.
The national reference frame is in the process of being modernized, and when a new frame is adopted every coordinate in the country gets a new number for the same piece of ground. Nothing will have moved. The numbers will have.
For anyone reading maps in the field, the practical takeaway is short: a coordinate copied from an old document into a modern device may land you a long way from where the document meant, and the device will display it with total confidence.
Datums: WGS84 vs NAD83Error three: your receiver
The third error is the one people are most aware of and still routinely underestimate.
A phone or handheld receiver under open sky, with a clear view of a lot of satellites, is typically good to a few meters. That is genuinely impressive and it is not the same as exact.
Conditions degrade it substantially. Under dense canopy, in a canyon, against a cliff, or in a river bottom with steep walls, the receiver sees fewer satellites and sees some of them via signals that bounced off rock or trees before arriving. Those reflected signals arrive late, the receiver treats them as true, and the calculated position moves. This is the reason your track log wanders when you sit still in a narrow drainage.
The accuracy circle a device shows you is an estimate of its own error, produced by the device, based on assumptions. It is a useful indicator and it is not a guarantee.
Put that next to the previous two errors and the conclusion is unavoidable. A device that says you are thirty feet inside a boundary, when the boundary polygon may itself be off by more than thirty feet, is not telling you which side of the line you are on. It is telling you that you are near it.
Error four: the world moved
Boundaries described relative to physical features have a problem, which is that physical features do not hold still.
Rivers are the obvious case and the most consequential one for anglers. A river that shifts gradually, eroding one bank and building the other over years, generally takes the boundary with it under the common-law doctrine of accretion. A river that jumps its channel suddenly, in a flood, generally does not: the boundary stays in the old channel even though the water is now somewhere else. That distinction, gradual versus sudden, has enormous practical consequences and it varies in its details from state to state.
Which means a river can be flowing across ground that legally belongs to whoever owned the old bed, and the map may show either state of affairs depending on when it was last updated. Combine that with the separate question of where the public boundary sits relative to the water, which we cover in the ordinary high water mark, and you have the most genuinely uncertain kind of boundary in the country.
Coastlines do the same thing on a longer timescale. Lakes rise and fall. Trees called out as monuments in a metes and bounds description die and rot and are replaced by nothing.
Error five, which is really a rendering problem
There is a fifth thing that is not error in the surveying sense but produces the same result.
A line on a screen has width. At a zoomed-out scale, a boundary drawn two pixels wide may cover a considerable distance on the ground, and where exactly within that band the line falls is not something the display is telling you. Zoom in and the line gets narrower in real terms, which is why a boundary that looked clear at trip-planning scale becomes ambiguous when you are standing on it.
Apps also generalize geometry for performance, dropping vertices from complex polygons at low zoom. The shape you see may have had detail removed.
The photograph underneath is also displaced
Worth separating out, because people treat aerial imagery as ground truth in a way they no longer treat maps.
Aerial and satellite imagery is fitted to a coordinate system through a process that corrects for the camera's angle and for the shape of the terrain. Where that terrain model is good and the ground is flat, the fit is excellent. Where the ground is steep, it is not, because anything with height gets photographed from the side as well as from above and ends up drawn leaning away from the camera position.
The practical version: in mountainous country, the top of a cliff, the crown of a tall tree and the ridge of a roof all appear displaced from the point directly beneath them. Imagery captured on a different day from a different angle displaces them differently. That is why the same feature appears to move slightly when an app switches imagery providers or refreshes its basemap.
So when a parcel line appears to run right along a visible feature, you are looking at agreement between two independently approximate things. It is weak evidence and it feels like strong evidence, which is the worst combination available.
What to do about all of it
The honest answer is not to buy a better device. It is to change how you hold the information.
Treat every boundary as a band rather than a line. Assume the band is wider in old rural country, in steep terrain, near water and under canopy, and narrower in a recently platted subdivision on flat open ground.
Build in a buffer and make it a habit rather than a calculation. If the line matters, stay well off it. The cost of being a hundred feet further from a boundary than you needed to be is nothing. The cost of being twenty feet on the wrong side of one is a conversation you do not want.
Believe physical evidence over the screen. A surveyor's monument, an agency boundary sign, a corner marker or a posted line is direct evidence about the real boundary, and the map is an inference about it. When they disagree, the ground wins. This is also true in the reverse direction, which is worth saying: a fence is not evidence of a boundary, because fences get built where it was convenient to build them.
Ask the agency, on public land. As we said in public land layers explained, the managing agency surveys and posts its own boundaries and its answer beats any app.
And if the boundary genuinely matters, in the sense that money or a legal outcome depends on it, that is the point at which a licensed surveyor is the correct answer and nothing on your phone is. Resolving a real boundary is professional work and, if it goes far enough, a matter for a court.
Where this bites hardest
Three situations are worth naming because they combine several errors at once.
Corner crossing, where the entire question turns on the precise location of a point at which four parcels meet, and where the underlying grid was laid out by exactly the nineteenth-century process described above. We covered the legal shape of it in corner crossing explained. The mapping shape of it is that the corner has real uncertainty attached and your device has more.
River access, where the boundary is defined by water that has moved, in a state whose rules about that movement you may not know, with a separate legal line for public use somewhere near it. The place to start is wading versus floating rights.
And any unmarked line in steep timbered country, where survey error, canopy multipath and terrain-driven imagery displacement all pull in unrelated directions at the same time.
There is a fourth that deserves mention because it is the friendliest one: the boundary that is wrong in your favor. Public land layers lag acquisitions, so ground that an agency bought recently may still show as private on a consumer app. If a place looks private and something about it seems off, a call to the local office occasionally turns up a recent purchase and a piece of newly public ground that nobody is using yet. The error runs both directions, and only one direction gets talked about.
The gear
The theme here is that the useful purchases are the ones that tell you about uncertainty rather than the ones that promise to remove it.
A mapping app that displays a real accuracy estimate rather than a confident dot. Knowing the size of your uncertainty is more valuable than a position claimed to a tenth of a foot.
An external GNSS receiver, if you genuinely need better than a phone. They exist, they are affordable, and they help most in exactly the conditions where phones are worst. They do not fix survey error or datum error, which is worth remembering before buying one to settle an argument.
Offline maps and a downloaded parcel view, because these questions arise where there is no signal. The approach is the same one described in navigation without signal.
A compass and the ability to use it, since it does not care about satellites and does not have an error mode that produces a confident wrong answer.
The agency or county contact for the ground you are on, before you go rather than while you are standing there.
A camera, for recording monuments, signs and posted lines as you find them. Physical evidence is worth more than recollection.
And a willingness to walk away from a marginal spot, which costs nothing and is the only tool on this list that works every time.
The short version
Maps are models. Models have error. The error in boundary data is larger than the crispness of the rendering suggests and it is not displayed anywhere.
None of that makes the tools useless. Parcel maps and public land layers are the best planning instruments anyone outdoors has ever had. It just means the correct relationship to them is the one you would have with a good weather forecast: extremely useful, worth acting on, and not something to bet your afternoon on being exactly right.
