A worn slab of Roman stone, forgotten for decades in a Dutch museum, has just forced historians to rethink ancient strategy.
In the Roman town of Coriovallum, beneath what is now the modern city of Heerlen in the Netherlands, a modest limestone fragment has gone from dusty display item to headline-maker. Thanks to cutting-edge artificial intelligence and meticulous archaeological work, researchers now argue that this object is the earliest known example of a “blocking game” board in Europe, pushing back the history of this type of strategy game by several centuries.
A mysterious slab in a small Dutch museum
The story begins not in a high-tech lab, but in the collection of the Het Romeins Museum in Heerlen. There, curators had long kept a 21 by 14.5 centimetre stone, unearthed around the turn of the 20th century from the remains of Roman Coriovallum.
At first glance, the piece looks unremarkable: a rectangle carved into the surface, criss-crossed by four diagonal lines and one straight line. The edges are chipped, and the grooves are irregular. For years, it sat catalogued as a possible gaming board, but with no confirmed identification.
That changed in 2020, when archaeologist Walter Crist, a specialist in ancient games, examined the fragment more closely. He noticed that the scratch-like pattern did not match any known Roman game such as latrunculi or tabula.
This was not a doodle or a mason’s test cut, but a carefully planned layout that did not fit any documented Roman board.
3D scanning later confirmed his suspicions. High-resolution digital models showed microscopic variations in the depth and polishing of the lines. Certain intersections were far more worn than others, suggesting that pieces had been moved, repeatedly, along specific routes on the board.
That pattern of wear convinced researchers that this was a finished product in active use, not an abandoned prototype. Based on the archaeological context and stylistic analysis, the stone was dated to roughly 1,500–1,700 years ago, placing it in the later Roman period.
AI joins the game: reconstructing lost rules
The real breakthrough came when an international team from the Netherlands, Belgium and Australia brought artificial intelligence into the investigation. They turned to Ludii, a game-simulation platform developed at Maastricht University, designed to reconstruct and analyse traditional games using computational models.
Ludii combines historical data, formal descriptions of known ancient games and search algorithms. The system can generate thousands of possible rule sets compatible with a given board layout, then simulate virtual players competing under those conditions.
➡️ This humble American hospital ship sits at the center of another US–Greenland spat
➡️ Prince William reflects on a “brutal year” for the royal family as both his father and his wife face cancer diagnoses
➡️ Hairstyles after 50: These 3 lifting haircuts make your face look 10 years younger, according to experts.
➡️ 6 minutes of darkness get ready astronomers say this once-in-a-lifetime event may be the last visible for decades in some regions
➡️ Full list of one off payments Big Reveal due to land in bank accounts before New Year in March
➡️ Astronomers Just Mapped 1,285 Star Factories in One Galaxy. ALMA Just Revealed Something Unexpected…
➡️ A rare giant bluefin tuna is measured and confirmed by marine biologists using peer-reviewed protocols
➡️ Hairstyles after 60 the blunt truth from stylists who say keeping old lady looks is a choice to age faster and this one cut exposes it
By letting AI “play” thousands of hypothetical games, researchers could see which rules matched the physical wear on the stone.
For the Coriovallum slab, the team fed in the board design, the distribution of wear marks, and what we know about Roman gaming culture. The AI then tested variants of race games, capture games and positional strategy games.
The results kept pointing in the same direction: the most plausible solutions were “blocking games”, where the aim is not to capture pieces but to trap the opponent until they can no longer move.
What is a blocking game?
Blocking games are a family of strategy games where each move changes mobility on the board. Instead of removing pieces, players try to restrict each other’s options, forcing a deadlock.
- The goal is to leave the rival with no legal moves.
- Pieces usually stay on the board throughout the game.
- Success relies on anticipating movement patterns several turns ahead.
- Many medieval and modern abstract games share this core mechanic.
Until now, historians thought such games emerged in Europe during the Middle Ages. The Roman fragment from Heerlen suggests that people in the late Roman Empire were already playing something structurally similar centuries earlier.
Rewriting the timeline of European strategy games
The study, published in the journal Antiquity, carries weight far beyond one small stone. If the interpretation holds, it shifts the origin of blocking games on the continent back to late antiquity.
Roman soldiers, merchants and local elites might have used this board in taverns, bathhouses or private homes. It points to a social environment where abstract strategic thinking was part of everyday leisure, not just military planning or philosophy.
The Coriovallum board suggests that Romans in the provinces were already playing highly positional, rule-based strategy games before medieval courts popularised them.
For game historians, this adds a missing chapter between classical games, such as Roman dice and race games, and later medieval innovations including certain forms of draughts and early positional puzzles.
It also reinforces a broader pattern: the Roman Empire served as a crossroads for game cultures, absorbing influences from the eastern Mediterranean and exporting its own pastimes across Europe.
How archaeologists taught AI to “read” wear and tear
The Coriovallum project grew out of the ERC-funded Digital Ludeme Project, which set out to reconstruct ancient and traditional games using computational tools. A key concept is the “ludeme”: a basic building block of game rules, such as “move diagonally” or “capture by blocking”.
By breaking known games into ludemes and storing them in a database, the system can propose new combinations that still make sense structurally. When researchers confront an unidentified board, they can ask the AI to assemble rule sets that fit both the geometry of the board and the cultural context.
In Heerlen, 3D scans provided an extra dataset: some lines were deeper, some intersections smoother. The team treated these as clues. They asked: where would hands and pieces most often meet if the game had a blocking mechanic? The AI simulations that best matched those hotspots rose to the top of the shortlist.
| Evidence | What it suggests |
|---|---|
| Diagonal and straight lines combined | Pieces likely moved in more than one direction type |
| Heavy wear at intersections | Frequent stopping points or contested positions |
| Consistent layout, careful carving | Standardised board for a known game, not random marks |
| Concentration of wear on one side | Possible starting area or strong strategic zone |
This methodology offers a new way to treat boards, dice and other gaming artefacts. Instead of relying only on visual resemblance to known objects, archaeologists can now test how different rule sets would actually play out on the artefact’s surface.
Museums, forgotten drawers and hidden pastimes
The Coriovallum stone also highlights the quiet power of museum storerooms. The fragment was not a fresh excavation; it had been preserved for decades with limited interpretation. Only when an expert in ancient games looked again, and when AI tools were ready, did its full significance emerge.
This raises an uncomfortable question for curators: how many other “ordinary” objects, from scratched tiles to carved roof slates, might encode game boards or other systems we have yet to recognise?
Archaeologists now see long-ignored markings as possible evidence of complex leisure, not just idle doodling.
Re-examining old collections with digital tools can be cheaper and faster than large-scale digs. Scanning, simulation and pattern-matching can pull new stories from objects that have already been cleaned, catalogued and safely stored.
Why ancient games matter today
At first glance, the rules of a 1,600-year-old game might seem like a niche concern. Yet games provide rare, direct evidence of how people thought, competed and socialised. They show how abstract ideas were translated into shared, repeatable experiences around a table or a stone slab.
A blocking game, for instance, requires players to anticipate opponent moves, manage limited space and accept that victory often comes from forcing a stalemate. These are mental skills valued today in business, diplomacy and computer science, but they were already being trained informally in Roman taverns.
For modern readers curious about testing similar logic, simple blocking mechanics survive in several accessible games. Variants of “three-in-a-row” puzzles, certain abstract apps and some pen-and-paper games used in classrooms all rely on restricting an opponent’s moves rather than capturing pieces outright.
Key terms and future scenarios
Several concepts sit at the heart of this research and are worth clarifying:
- Game reconstruction: The process of proposing rules for an ancient game when only boards or pieces survive.
- Simulation: Running many virtual plays of a game to see which rule sets feel coherent and match physical evidence.
- Wear analysis: Studying tiny scratches, polish and damage on artefacts to infer how hands and objects interacted with them.
Looking ahead, the same AI and scanning approach could be applied to enigmatic boards from the Near East, North Africa or northern Europe. A future research programme might virtually “teach” AI entire families of traditional games, then systematically compare them against museum databases.
There are risks: algorithms could overfit to incomplete data, or researchers might accept elegant digital solutions that do not reflect messy historical reality. This is why archaeologists stress that AI-generated rules remain hypotheses, to be weighed against texts, iconography and ethnographic parallels.
Yet the benefits are clear. By combining patient human expertise with machine-driven pattern searching, even a small stone from a provincial Roman town can reshape timelines, connect cultures and remind us that strategic thinking over a board is far older than any modern chess clock or smartphone app.
Originally posted 2026-02-19 06:19:31.