Concept
The query «safest country» promises a simple answer: a name, a first-place ranking, and certainty in the choice. But already in the source table, this answer fragments into several different dimensions. Right next to each other are the Global Peace Index, the terrorism impact indicator, the TravelSafe rating, and the U.S. News ranking. Their juxtaposition suggests comparability, even though they utilize different scales, refer to different years, and cover different sets of countries. The «Borders of Calm» project explores the gap between rating convenience and the conditions under which a rating is meaningful.
At the core of this work is a CSV file containing 195 countries and territories. The initial question, «who is safer?», gradually evolves into a more precise one: «what exactly are we calling safety, and whom are we including in the comparison?». This shift in questioning establishes the project’s dramatic structure. First, the reader sees the scope of the sources, then encounters the familiar top ten, after which they compare two evaluations of a single country, observe rank discrepancies, and examine the consistency across all four metrics. Each subsequent chart refines the scope of the previous one.
The working hypothesis is that the safety indicators correlate partially but are not interchangeable. This can be tested without creating a new composite index: it is sufficient to compare the completeness, initial values, and country ordering. Furthermore, the discrepancy in results is not attributed to an error in any single source. It may involve the indicator definition, the observation period, the sample composition, and data characteristics simultaneously. The attached table illustrates the divergence but does not isolate the contribution of each of these factors.
Data: what ended up in one table
The original file contains 195 rows and 15 columns. GPI represents issues from 2021–2026, while GTI represents 2022–2025. For the main comparison, GPI 2026, GTI 2025, TravelSafe (without a specified year), and U.S. News 2024 were selected. The table creator compiles this data under the overarching theme of safety; source markings and years remain visible in the project.
GPI describes pacifism through conflict, public safety, and militarization. Its lower value is considered more favorable. GTI relates to the impact of terrorism; a lower value is also more favorable. TravelSafe has the opposite metric: a higher rating indicates a more favorable security assessment. The U.S. News column contains rankings, so a lower number signifies a higher position. The full methodology for this specific column is not independently verified; it is used with attribution to the CSV compiler.
The difference between the metrics begins before the numbers. They do not share a single observation time, and at least one of the four ratings is missing for 130 rows. Therefore, the overall filename should not be interpreted as a promise of a synchronous snapshot of the world for 2026.
Method: Preserve Differences
Analysis was performed in Python. No duplicate rows or country names were detected. Data types, missing values, ranges, and unique values were verified. When reading the data, the NA code for Namibia was preserved: standard pandas settings might incorrectly interpret it as missing data.Missing values were not replaced with zeros. For each paired comparison, only countries with two required values were selected. For location comparisons, both series are re-ranked on a common set of 88 countries; the average rank is used in cases of ties. Spearman correlations were calculated for available pairs, and the number of observations is indicated next to each coefficient.Time dynamics are not plotted. The versions of the annual columns do not align with the revised primary source series. Specifically, the direction of the simple difference for Iceland in 2025–2026 differs from the change described in the IEP publication. This limitation has been retained in the report rather than attempting to explain it with an unverified reason.
1. Who Entered the Dimension
Number of non-null values across four columns; the total dataset consists of 195 rows from the original CSV.
TravelSafe is populated for 191 countries, GPI for 163, GTI for 98, and U.S. News for 89. These bars represent observation availability, not safety ratings. All four metrics share values for only 65 countries—exactly one-third of the table.
The emptiness itself becomes the first substantive finding. If without
2. Ten of the Most Peaceful Countries
The ten lowest GPI 2026 scores among 163 filled entries. Lower indicates a more peaceful country by this index.
Iceland leads with a score of 1.161, followed by New Zealand at 1.343 and Switzerland at 1.363. The scatter plot maintains the distances between scores—it illustrates not only the ranking but also that the intervals between adjacent positions are unequal.
3. One Country—Two Perspectives
159 countries with filled GPI 2026 and TravelSafe data. GPI is on the horizontal axis, and TravelSafe is on the vertical axis; the scales point in opposite directions.
The scatter plot shows a pronounced overall correlation: lower GPI values often correspond to higher TravelSafe ratings. The Spearman correlation coefficient on the original scales is −0.801. The negative sign here is related to the opposing directions of the scales, not the content.
4. The scale changes—the space changes
Eight maximum absolute discrepancies between the recalculated GPI 2026 ranks and the U.S. News 2024 ranks. Both series are constructed using the same 88 countries.
It is difficult to compare rankings from lists of different lengths directly. Therefore, this analysis first establishes a common set of countries, and then the ranking for each indicator is recalculated within that set. The connecting line shows the distance between the two internal ranks;
5. Consent Has Boundaries
Spearman’s paired correlations after aligning directions toward «more = more favorable.» The sample size used is indicated in each cell.
All six coefficients are positive. GPI and U.S. News yield 0.825 across 88 countries, and GPI and TravelSafe yield 0.801 across 159. GTI and U.S. News yield 0.338 across 65 observations. The metrics do indeed overlap in the order of countries, but the degree of agreement in the observed pairs is
Visual Design and Infographics
The project is built on a light field and generous spacing. The plum-colored text provides contrast, while the matte purple unifies the diagrams, lavender links elements, and pear-colored highlights specific cases. Color accents are always accompanied by labels. The composition does not use a geographical map—the central question relates to the method of measurement rather than the spatial proximity of countries. Instead, scattering takes center stage. The two upper sections provide the initial framework, and the two lower sections allow for the interpretation of the consequences of changing the metric.
Four offset apertures connect the poster’s cover and logo, remaining a decorative metaphor without a numerical scale.
Carrier: general view
Generative Mockup: a sheet on a semi-transparent display panel.
The overall view presents the composition as a physical object. The transparent panel continues the motif of layering and leaves ample negative space around the graphics. The viewer first registers the title and the central field before moving on to the details.
Medium: print surface
Generative Mockup: Snippet of the central graphic and paper edge.
The close-up shifts focus from the overall form to texture, color, and the spacing between elements. It showcases the contrast between analytical drawing and the soft materiality of the paper. This is a visualization of the intended print; precise coordinates and fine text should be viewed in the original PNG/SVG files, as generation may alter them.
Presenter: Reading and Discussion
Generative Mockup: Comparison of Ranks on the Desktop.
The third scenario illustrates an action: a person points out a positional discrepancy and correlates it with the legend. The infographic serves as the subject of discussion. There are no external captions overlaid on the image; the explanation remains a standalone text block within the long-form content.
Process and AI Usage
AI was used to prepare the code, formulate the concept, generate text, and create subject imagery. The cover was developed as a standalone, wordless metaphor; the three mockups are separate generations using a finished poster as a reference. The generator was not used as a source of numbers. The analytical version of the poster was assembled programmatically to preserve coordinates, scales, and the Russian text. The mockups illustrate the designed medium, rather than documenting an actual exhibition.