Every map is a deliberate simplification of the reality it represents. The cartographer must decide which features to show, which to omit, which to emphasise, and how to resolve the many conflicts that arise when the real world - complicated, three-dimensional, and constantly changing - is compressed onto a flat sheet of paper or, more recently, a screen of pixels. These choices are not usually presented as choices: a finished map looks authoritative, as if its content were a simple record of the world rather than a specific interpretation of it. The study of the choices that cartographers have made, and continue to make, has become a recognised field of historical research over the last fifty years.

The most fundamental of these choices is the map projection - the mathematical rule by which the curved surface of the Earth is transformed into a flat image. No projection can preserve all properties at once: a projection that preserves area distorts angles and directions, while a projection that preserves angles distorts area. The Mercator projection, widely used from the sixteenth century onwards for navigation, preserves angles - a straight line of constant compass bearing on the map corresponds to a straight line on the Earth's surface - but substantially exaggerates the size of landmasses at high latitudes. Greenland appears larger than Africa on a Mercator map, though the real Africa is about fourteen times the size of the real Greenland. Since the mid-twentieth century, several alternative projections that preserve area more faithfully have been proposed for general-purpose mapping, and some publications have shifted away from Mercator for any map that is not specifically intended for navigation.

Projection is not the only locus of cartographic choice. The scale of a map determines what can be shown: at a scale of 1:1,000,000, individual buildings cannot be represented, and entire villages may be reduced to a single dot. At a scale of 1:10,000, individual streets and buildings can be shown, but the cartographer must still choose which to label, what to do with small features too close to each other to be depicted separately, and how to handle features such as footpaths that might or might not be visible depending on the season. The Scottish geographer Fiona Urquhart has argued that the central challenge of cartographic design is the resolution of scale-dependent conflicts: the cartographer must give the reader the information needed for the map's intended purpose while omitting material whose inclusion would produce visual clutter that defeats the same purpose.

Symbol choice is another significant decision. The small graphic symbols used to represent features such as churches, hospitals, bus stops, and so on carry cultural assumptions that are not always noticed. A cross symbol for a church works in most of Europe but is inappropriate in places where the overwhelming majority of places of worship are not Christian. The old Ordnance Survey convention of representing a church with a tower and a church with a spire differently reflected a specific social context in which such distinctions mattered for navigation; modern maps have often simplified these conventions because most users do not need the distinction. Similar questions arise for the representation of political boundaries, which can be shown with degrees of precision ranging from a clear line to a diffuse zone, and whose treatment can be politically sensitive.

Colour schemes carry their own burden of assumption. The traditional convention of showing different countries in contrasting colours, often associated with educational atlases, invites the viewer to see political boundaries as more important than physical ones. A map showing the same region coloured by elevation, or by vegetation, produces a very different visual impression. Urquhart has pointed out that the ubiquitous colour convention of showing land in brown and sea in blue is not a neutral choice: in a climate where water is scarce and valued, an alternative convention might make more sense. The conventions that modern Western cartography treats as natural are, on her account, historical artefacts of the environments in which Western cartography developed.

Historical maps reveal how these choices have changed. Medieval European maps often placed the east at the top of the page, with Jerusalem at the centre, reflecting the religious priorities of the societies that produced them. The shift to placing north at the top, which became standard in European cartography during the age of printing, reflected in part the practical convenience of navigators using magnetic compasses that pointed north. Chinese maps of the same period used entirely different conventions, including a southern orientation on some medieval examples. The modern global convention of north at the top is not a natural given but a specific decision that, once made, became difficult to reverse because of the expectations it created.

The digital era has introduced new kinds of cartographic choice. Interactive online maps can show different levels of detail at different zoom levels, potentially offering the user more information than any paper map. The underlying data, however, must still be organised into a hierarchy of detail, and the decisions about what to show at what zoom level reproduce many of the traditional choices in a new form. The prevalence of a few dominant platforms for online mapping - Google Maps, OpenStreetMap, Apple Maps - has produced a degree of standardisation in current map conventions that would have been difficult to achieve through any deliberate effort, along with a concentration of cartographic authority in a small number of corporations. Urquhart has argued that the concentration of cartographic authority raises questions about public oversight that are only beginning to be addressed.

The study of cartographic choices has practical implications. Users of maps can make better decisions if they understand the choices embedded in the maps they read: the same physical reality can be legitimately represented in different ways for different purposes, and recognising the interpretive element of any given map makes it easier to use several maps together. Cartographers themselves can design better maps if they understand the choices they are making explicitly rather than accepting inherited conventions without reflection. Urquhart has described the ideal as "informed cartography" - map-making and map-reading that treat the map as an argument about the world rather than as a direct representation of it. The argument may be a good one or a poor one, but either way it is an argument made by a specific person for a specific purpose. Seeing it as such is, on her account, the first step toward using maps well.