In the Great Wargaming Survey, 2025 edition, a related question asked respondents, “What do you like least about miniatures wargaming?” As in the "most liked" question, responses were recorded as unstructured text with a maximum length of about 2,100 characters. A few respondents actually pushed this limit in their lengthy and generous responses. Some of these longer responses are fascinating to read and offer much to consider. In total, there were 4,029 non-blank responses recorded.
Technique Choice
Cost as an Independent Factor: Cost operates largely independently, showing low alignment with any single cluster across the first two PCA dimensions. This suggests that financial expense is a universal secondary concern across all groups rather than the primary defining trait of a specific subset of players.
Summary
The broad inferences, here, suggest that wargaming’s main barriers are threefold and arise at three distinct stages of the hobby. These are:
Technique Choice
To (hopefully) make sense of this large body of free-form text, machine learning techniques are introduced. Unlike the analysis on "What I Like" linked above, this analysis focuses on Principal Component Analysis (PCA) rather than on hierarchical cluster analysis. While both tackle the challenge of qualifying these text responses, PCA and cluster analysis approach the problem differently in order bring some meaning out of the seeming chaos. Both of these statistical techniques attempt to bring out underlying data associations lurking within these data. PCA helps with revealing these patterns, graphically, using a biplot technique. The first step in analysis is to preprocess these data and prepare them for further analysis.
After these routines are completed, the resulting dataset produced 3,488 unique terms with associated frequencies. A further data step involved keeping only tokens with 30 or more counts and then removing near-zero variance terms. The result produced a dramatic dataset reduction to only nine key word tokens. These nine tokens are "people", "painting", "gaming", "time", "competition", "miniatures", "cost", "rules_lawyer", and "find".
Preprocessing
To produce a PCA biplot, data reduction methods are utilized to reduce thousands of unique words (tokens) to a smaller and more manageable dataset. The goal is to transform these unstructured texts into a count of representative tokens while keeping the essential meaning intact.
As an example, a snippet of survey responses shows the following text in the Miniatures_Gaming_Least column. Parsing and preprocessing requires a number of text transformations (including correcting some spelling errors) to shape tokens into a standardized form. In addition, synonyms are used to group similar tokens. For example, parsing and tokenizing the highlighted entry produces the meaningful tokens, "general", "concerns", "persistent", "gatekeeping", "wargaming", "community".
| Raw text before parsing. |
Principal Component Analysis
Once tokenization is complete, the next step focuses on Principal Component Analysis (PCA). PCA simplifies complex datasets by reducing a number of related variables into a smaller set of uncorrelated dimensions called principal components. These components are ordered by how much variance they explain. The first principal component captures the most variation. The second captures the next most, and so on. One useful visualization tool is a PCA biplot which illustrates the relative importance (loadings) of each of the nine variables in 2D space. The PCA biplot produced by this analysis is shown in Figure 1. The plot shows broad association patterns only and does not identify any one individual. Color and length of each vector denotes its contribution to the PCA analysis. Variables pointing in similar directions are positively linked. Those pointing in opposite directions represent contrasting views. Vector length implies importance in the given dimension. The plot suggests that respondents’ least-liked aspects of wargaming fall into a few related, but distinct types of dissatisfaction based upon the nine tokens retained in the preprocessing phase. Based upon the results illustrated in Figure 1, four main groups emerge from the PCA plot as shown in Figure 2.
At a higher level of aggregation and inspection, these four clusters can be neatly bifurcated by each of the two dimensions (DIM1 and DIM2) as illustrated in Figure 3 and Figure 4. For each Dimension, hemispheres are given intuitive classifications to describe the groupings.
| Figure 3 - DIM1 |
| Figure 4 - DIM2 |
Interpretation and Inference
Graphical analysis of PCA results tends to lend itself (in most cases) to inferences that are easier to interpret than raw data and rudimentary descriptive statistics. This PCA analysis shows that wargaming complaints aren't just one big nebulous cloud of negativity. Instead, player dissatisfaction falls into three distinct "pain points" and one universal constant as shown in Figure 2. These pain points are:Burden (Miniatures, Painting, Time)
The strongest pattern connects the time and labor required to bring a force to the table. While respondents do not necessarily dislike collecting or historical gaming, itself, they tend to feel overwhelmed by the total preparation commitment. Those commitments can include assembling, painting, basing, storing, and managing a project backlog of The Lead Pile. The primary obstacle is the time lag between purchasing the figures and terrain for a new project and actually getting the figures onto the table for a game.
The strongest pattern connects the time and labor required to bring a force to the table. While respondents do not necessarily dislike collecting or historical gaming, itself, they tend to feel overwhelmed by the total preparation commitment. Those commitments can include assembling, painting, basing, storing, and managing a project backlog of The Lead Pile. The primary obstacle is the time lag between purchasing the figures and terrain for a new project and actually getting the figures onto the table for a game.
Access (Gaming, People, Find)
A second distinct cluster highlights the possible frustration around getting a game onto the table once painted armies have mustered out. This roadblock includes finding compatible opponents, joining local clubs, scheduling, or matching interests in a particular period, scale, or ruleset. For these players, the challenge isn't building or painting an army, but the ability to transform those completed units into regular, enjoyable games.
Culture & Friction (Competition, Rules-Lawyering)
A third, sharper pattern links competitive environments with rules disputes. This reflects a clear distaste for extreme adversarial gaming where min-maxing, rules-lawyering, or winning overshadows sociability, historical narrative, and mutual enjoyment. This dimension strongly differentiates players who favor relaxed, scenario-driven or pick-up games from those uncomfortable with tournament-style play.
A second distinct cluster highlights the possible frustration around getting a game onto the table once painted armies have mustered out. This roadblock includes finding compatible opponents, joining local clubs, scheduling, or matching interests in a particular period, scale, or ruleset. For these players, the challenge isn't building or painting an army, but the ability to transform those completed units into regular, enjoyable games.
Culture & Friction (Competition, Rules-Lawyering)
A third, sharper pattern links competitive environments with rules disputes. This reflects a clear distaste for extreme adversarial gaming where min-maxing, rules-lawyering, or winning overshadows sociability, historical narrative, and mutual enjoyment. This dimension strongly differentiates players who favor relaxed, scenario-driven or pick-up games from those uncomfortable with tournament-style play.
Cost as an Independent Factor: Cost operates largely independently, showing low alignment with any single cluster across the first two PCA dimensions. This suggests that financial expense is a universal secondary concern across all groups rather than the primary defining trait of a specific subset of players.
Dimensional Study
Dimension 1 (DIM1) in Figure 3 breaks the horizontal axis clearly between a production cluster on the left from the gaming/social cluster on the right. On the left are people who dislike painting, miniatures, and the time those activities demand. On the right, complaints focus on other players, gaming opportunities, competition, and even finding games or opponents. Perhaps this underscores the differences between the solo preparatory burden of miniatures gaming and the social or organizational realities of actually playing it.
Dimension 2 (DIM2) in Figure 4 divides the vertical axis between the access or burden dislikes of the hobby against the negative culture or style of play components. On the top, dislikes focus on participation, access, or preparation. The bottom, on the other hand, highlights the style or tone of actual play especially in competitive or argumentative settings.
The broad inferences, here, suggest that wargaming’s main barriers are threefold and arise at three distinct stages of the hobby. These are:
- Preparing to play. Acquiring and painting miniatures takes time, effort, and money.
- Finding a game. Connecting with like-minded opponents, clubs, and schedules.
- Enjoying the game. Matching culture, style, and competitive expectations once at the table.
Ultimately, these data indicate that player disengagement is rarely caused by a lack of enthusiasm for history or rules. Instead, these dislikes are driven by the sheer effort required to prepare, the friction of finding compatible opponents, and the occasional dissatisfaction with how games are played.
What are your main dislikes about the wargaming hobby? Do any of these tendencies and associations hold for you?