I stumbled upon an article of Katapult Magazine (a German outlet specializing in critical cartography and social science and more) about 2025 Press Freedom Index in their Wiki (Signup-wall: https://wiki.katapult-magazin.de - Katapult Magazin, 2025). The author published a map of the press freedom index using data from "Reporter ohne Grenzen", the German division of RSF ("Reporters without borders" or in French "Reporters sans frontières").
By reverse-engineering the Katapult map and comparing it to its source data from Reporters Without Borders (RSF), we can observe distinct approaches to visualize the same dataset.
1. Projection Analysis: Katapult Magazine
Let's check the projection first, which is definitely not Mercator, as you can immediately see from the size of Greenland. The outer edges of the projection are slightly bent, which means it might be a pseudocylindrical type of projection, like Robinson. Greenland keeps its overall shape, which means it could be a compromise projection (Robinson or Natural Earth), as an equal-area projection like Equal Earth would distort its shape slightly more. To spot the difference between Robinson and Natural Earth is quite hard for me, so I consulted another page for projection comparison:
While similar to the Robinson projection, the distortion of Nunavut (Northern Canada) suggests that the Natural Earth projection was likely chosen. Unlike the Mollweide projection, which would distort high-latitude regions like Alaska significantly due to curvature, Natural Earth offers a compromise that maintains recognizable shapes.
A notable design choice is the shifting of the central meridian approximately 10 degrees west. This creates a contiguous view of the Russian landmass, preventing the truncation of Chukotka (Russian Okrug next to Alaska) that occurs in standard 0-meridian centered maps.
2. Data Generalization and Geopolitical Boundaries
Let's check the data as well. On the first sight the map includes the big lakes (which GeoCanvas default world dataset does not). They excluded Antarctica (there is no data for press freedom in Antarctica anyway) and used a simplified version of a world map, likely to improve readability. However, this generalization results in the exclusion or distortion of smaller geographic features.
We are missing some smaller countries, and the northern island of New Zealand (nothing special about missing New Zealand on a map, but here it is half missing). The east Caribbean states also fell victim to the simplification of the geometry, as well as Malta and the smaller islands of the Baleares. Besides that, there are noticeable other decisions:
- Falkland Islands is greyed out, even though its not in the legend, probably means no data. As Falkland Islands are disputed between Argentina and the UK it was a decision to use no data. Other British Oversea territories are not rendered, because they might be too small for the resolution anyway.
- Jamaica was maybe missed in the design, as the base data offers values, but it is grey as well.
- Greenland holds the same value as Denmark, while the original data provider marked it as not available.
- Western Sahara is shown in its de jure borders, holding probably the same value as Morocco, while the primary data provider did not distinct it and even left out the de facto geometry of West Sahara. This differs from many de facto datasets which separate the territory.
- Geometry simplification also affected Lebanon and Palestine, which left a kind of unfamiliar shape of the Mediterranean Sea.
There are more border disputes, and not everything can be mapped in the highest detail, so there might be more, but those are – in my perspective – the most obvious differences to spot in the country dataset.
The colour palette also diverges from the source; Katapult uses a light green for "satisfactory" (German: "zufriedenstellende Lage") ratings, whereas RSF utilizes yellow. This adjustment arguably improves semantic clarity, as green intuitively aligns with a positive "satisfactory" status, but depends on the impression the authors wanted to transport. RSF used a more alarming yellow for "satisfactory".
For a reproduction, we could use one of the Natural Earth POV datasets (Natural Earth - Admin 0 – Countries point-of-views) which includes the border of Western Sahara as de jure, compared to the default Natural Earth de facto world dataset.
3. Comparison with the Source: Reporters Without Borders (RSF)
Those were the design decisions made by Katapult, while the cited data provider RSF (https://www.reporter-ohne-grenzen.de) made some different decisions on their map.
First, they use the Mercator projection in the default rotation on the homepage. The data for the world map they use is probably derived from a de facto border dataset (e.g., Natural Earth 110m) as it has the de facto border of Western Sahara. It holds more details than the map by Katapult, as we have Palestine on the map, but it is also differs from the Natural Earth dataset: The Bahamas are missing, as well as a bigger island in northern Russia (Земля Георга / Zemlya Georga / Prince George Land). The resolution is therefore somewhere in between Natural Earth 110m and the one Katapult used in their simplified version. Going with the de facto border state, they excluded (or included by the data) Western Sahara and Morocco and used the Russian de facto border on the Crimean Peninsula.
I am more curious about the fact that they render a world map but exclude countries they hold data for: Singapore, Malta, OECS (Organisation of Eastern Caribbean States), and some others. But they do include them in their report map though.
The main organization of RSF and its map viewer is different in the way it shows the data on the map (https://rsf.org - Reporters Without Borders, 2025b). They are using the Robinson projection. This could have been figured out by analyzing it again, but looking at the code they use is way easier:+proj=robin +lon_0=0 +x_0=0 +y_0=0 +datum=WGS84 +units=m +no_defs
They also did not use any rotation but some decisions were made differently compared to the German division. They are using a simplified version but loading data depending on scale. That means we can see Malta, Singapore, and OECS states when zooming in. They are listing 180 entries in their index, while the map on the German division's homepage only offers 177 features (and those already include Antarctica). RSF also includes Morocco and Western Sahara in one shape but includes the de jure border line. Other differences are:
- Greenland is Denmark
- New Caledonia is France
- Crimean Peninsula is de jure Ukraine
Both datasets also use different values in their data by a really small margin; for example, the Netherlands is 88.65 on the German division map compared to 88.64 on the Org level. The methodology they link is the same (https://rsf.org - Reporters Without Borders, 2025b) and also the actual rank list uses different values compared to the map (https://media.reporter-ohne-grenzen.de - Reporter ohne Grenzen, 2025c). There seems to be an issue in the data handling, maybe a rounding error on their map view.
4. Discrepancies in Third-Party Reproductions
The complexity of mapping this data is further highlighted by third-party reproductions, which often differs on other maps.
The map is used as header image for an article in an Austrian magazine Warda, which published a section called "Critical View", in German "Kritischer Blick" (https://warda.at - Warda Network, 2025). A prominent example of how distortion works on the Mercator projection. The magazine probably didn't create the map, a reverse search leads us to a former publication of RSF themselves, which is not available online anymore. The view even leaves a wrong impression of the position of the Equator because of the cut-off Antarctica, while Greenland would cover almost all the land of North America. Though from 2025, the data differs compared to other releases.
The map with the highest level of detail I found was created by the Wikipedia author NordNordWest. Probably a compromise projection, likely Robinson, it is also rotated by about 10 degrees. It uses a high-detail base map of countries and independent regions. It even includes the Holy See (Vatican City State). The data of this version is from RSF 2024 report, but it is not exactly clear what was used, for example, for New Caledonia; while available as geometry and with its self-name, the data used is probably derived from the data of France. The author used blue tones for the higher values instead of the green/yellow of RSF.
The German division also offers another view on the data (https://flo.uri.sh/visualisation/22641845/embed?auto=1). It uses Natural Earth II projection. I think it is interesting, as it just leaves out Greenland after all from the world map as there is no data. The linked version is the German version, but the map appears quite often on other pages and there are other language versions probably available. The time slider also shows the alarming decrease in press freedom worldwide over time.
The last example from an online magazine called "Adda247". While I would not consider it a reasonable source, it proclaims itself as a learning platform. I am not judging on the map, but the underlying data it shows. It seems as it either uses different or blatantly wrong data for the map which converge far from what RSF published as press freedom index 2025 while still citing them on that. The colours in the map do not exactly match the colours in the legend. Though a reverse search does not hold any broader circulation of this map.
Conclusion
The Katapult map stands out for its aesthetic accessibility and logical projection choices, making it ideal for general public communication (I would send that to someone asking about Freedom of Press). However, for academic citation, the RSF international viewer remains the standard, despite minor inconsistencies between the map visualization and the raw data values. When recreating such maps, cartographers must balance geometric simplification with the need to accurately represent smaller nations and complex geopolitical realities.
Keep in mind, that this is a very basic analysis and there are many more factors to consider when creating a map. While I try to be as objective as possible, this analysis inherits the bias I have. I am not a cartographer by profession (or expert by certificate) but working with data and maps on a daily basis.
Thanks to all the creators of the maps, to provide a different view on the topic.