← Instructional Design · Project Management

Case Study

One Chapter, Taught Two Ways

Rebuilding the movement rules for 20 Below, a tabletop RPG I’ve produced since 2021, so the same rule works whether a table plays on a map or entirely in their heads. The work went one decision at a time, with every number simulated, and it was written once for two very different learners.

Role

Producer: decisions and rulings.

Team

Claude (AI assistant), with team review and a playtest scheduled.

Timeline

24–26 September 2026.

Result

One movement chapter for map and theater-of-the-mind tables, every number simulated, changed in all eight places it appears at once.

The Problem

The movement rules fit in a few paragraphs, and they had drifted. The same all-out run appeared twice, as “Run” at four times your Movement Rate and “Sprint” at five. A line claiming a sprint “crosses Near in a single turn” wasn’t true for the slowest characters. Climbing, swimming, jumping, bad ground, chases and travel had no rules at all, even though equipment and creature cards already assumed them. Snowshoes cancelled a deep-snow penalty that didn’t exist. A motorcycle gave Advantage on Chase rolls with no chase rules to roll. Creature cards listed flying, swimming and burrowing speeds with nothing saying how to use them.

And the rules assumed readers knew what “theater of the mind” meant. Half of the tables who’d pick up this book have never played without a map, and the other half have never played with one.

Two Learners, One Rule

The instructional-design decision that shaped everything else: every rule is stated once, then applied two ways.

Because both tracks use the same numbers, a table can switch mid-session: a map for the warehouse fight, no map for the rooftop chase after it. Every section ends with the same worked example played both ways.

How the Work Ran

• Sequenced simple to complex. Seven steps: the basics, a single round, leaving Melee, terrain, climbing and swimming, chases, travel. Each step used only what came before. Two more, combat actions and special actions, were added to the scope as the work surfaced them.
• One ruling at a time. Partway through, I asked for decisions to come to me one question per message instead of in bundles. It made the pace slower per message and faster overall.
• Scope logged as it grew. Seven additions came up mid-step, from a materials table to stats for every creature. Each went straight onto the running to-do list instead of derailing the step in progress.
• Visualize everything. Every ladder and simulation got a chart, and every chart was checked against the text before it was trusted.

Benchmark Learners

Every simulation ran against the same five test characters: Fragile Frankie, Mediocre Mike, Average Annie, Tough Tony and Maxed Maggie. Each has the same score in every attribute (1, 3, 5, 7 and 10). They’re simulants: fixed benchmarks for design rather than characters anyone would build. Mike was added partway through, to fill the gap between Frankie and Annie, and he turned out to mark exactly where one finding flipped.

Design Decisions

1. Bad ground, tested at every skill level. Rough ground costs double, Wild Land triple, and an Athletics roll at Difficulty 6 pushes through. The simulation ran every character at every training tier, which also tested a concern the playtesters and the team had already raised about Novice. It confirmed it: at low attributes, a Novice rolls worse than someone with no training at all, because the Novice’s Disadvantage costs more than the attribute adds. Mike sits exactly on the crossover. For movement, my ruling was to leave it: in a modern campaign Athletics is an Everyman Skill, so everyone starts at Trained, and a standing rule already lets players argue for a different Element. Three days later, taken up as a question about Skill use rather than movement, Novice was retired.

2. Chases: from a footrace to a chase. The first simulation said speed decided everything: from 20 meters, a slower character never escaped a guard dog, and a faster one nearly always did. My ruling made escape mean getting out of sight, then winning a Stealth roll against the pursuer’s Perception. Now where you run matters, and so does being sneaky. Stealth training roughly doubles an average character’s chances.

3. Travel, grounded in the real world. I asked for the pace to come from real walking data, not a guess. Real-world benchmarks were mined from the web, and three candidate rules were charted against them. The best pace still let the frailest character walk a fit hiker’s day, and the cause was the fixed eight-hour day, not the speed. Making the travel day four hours plus Stamina put every test character inside the real-world range that matches them.

4. Actions: say what each one is for. A Slow round had quietly promised both a called shot and Advantage. That’s two bonuses for one choice. It became a list of empowered actions: Aim gives Advantage, a Called Shot doesn’t, and Study a Target this round gives next round’s Called Shot its Advantage. Two rounds of commitment buy what one can’t. Cautious Attack folded into Full Defense. A charge, Bulldoze, came from a concept mined from outside sources and was rebuilt entirely from rules the game already had. Cover got one rule for everything.

How Design Choices Get Made

More than sixty rulings went into this chapter. None of them started as a blank page, and none of them were made by the assistant. Every one followed the same loop.

1. Audit what the book already promises. Before any new rule, the assistant read everything the rulebook already said on the subject, including the places that pointed at rules that didn’t exist. Snowshoes, a motorcycle’s “Chase rolls,” gauntlets that resist a “disarm attempt,” and a compass that means you’re “never lost” are all promises made to players. A new rule has to keep them, so they narrow the options before anyone proposes anything.
2. Lay out real options, with a recommendation. Each decision came to me as two to four concrete options. Each option had a one-line trade-off and a number where one mattered, like “Frankie 28%, Annie 64%.” One was marked as the assistant’s recommendation, with its reason.
3. I rule. Often it’s one of the options. Often it isn’t: “2, with the conditions in 1,” or a version nobody offered. The ruling goes into the decision log with the date and the reason.
4. Throw dice before it’s final. Anything with numbers is simulated against the five benchmark characters and charted. If the chart shows a problem, the problem goes back to step 2.

The principles behind the rulings

Reuse before inventing

A new rule is a new thing to learn, so the first question for any gap was whether a rule the game already had could fill it.

• The endurance rule the book already had (free time, then a roll that gets harder each time) became swimming, running flat out and forced marching. Only the time unit and the starting Difficulty change.
• Rough ground’s “costs double” became climbing, crawling, squeezing through a gap, and what happens when you fumble a chase roll.
• The strength contest used for breaking free of a grip became Grab, Disarm and Bulldoze. Bulldoze came from a concept mined from outside sources and was rebuilt entirely from these parts.
• The stat-block format from an earlier revision (“Perception 7 vs. Stealth”) became how a chase ends: only the player rolls, with the creature’s number as the Difficulty.
• Advantage’s existing stacking rule already settled how cover and getting low combine. No new rule was needed, only a chart that showed it.

When the numbers push back

Three times a simulation said a draft was wrong. Each time, the fix changed a parameter and kept the idea.

• Swimming. Counted in minutes, nearly everyone had gone under within twenty minutes of calm water. The rule was fine; the clock was wrong. Calm water now runs in hours and rough water in minutes.
• Chases. With escape set at a fixed distance, speed decided everything. Changing escape to “out of sight, then Stealth” turned a footrace into a chase.
• Travel. The best pace still gave the frailest character a fit hiker’s day. The fix was the length of the day, four hours plus Stamina, not the pace.

Where the ruling differed from the recommendation

The assistant recommends and I decide, and the record shows it’s not a rubber stamp.

Question	Assistant recommended	I ruled
Should Full Defense keep its Defense bonus on top of Disadvantage?	Cut it; the Disadvantage alone is enough	Keep both. Against a guard dog, Annie is hit 22% of the time instead of 79%.
Is Dive for Cover Fast only?	Yes	No. In a Normal round you can move in, then hit the deck.
Do we need more band-specific actions?	None	Add Snap to Cover: a single step into cover, staying on your feet.
How should a chase be tracked?	A range-band track	Counted in meters, with the band track’s end conditions.
What does a critical Chase roll do?	Halve the other side’s movement	Move 1.5 times as far yourself.
Where does the forced-march clock start?	Endurance rolls right after 8 hours	Stamina hours free past 8. That later became the Stamina-length day, which the assistant proposed and I adopted.
What does a creature need when it grabs or gets grabbed?	Use its Soak as a stand-in for Potence	Give every creature real sub-stats. A bigger job, added to the to-do list.

Naming is a design choice

A name teaches the rule before the text does, so names got options too. “Attack on the Run” won over “Running Attack” because it fits a pistol and a machete equally, and doesn’t suggest a Dash’s distance. “Bulldoze” was picked from fourteen candidates. “Belly Down” separates choosing to get low from being knocked flat, which is still “Prone,” and that split became a rule of its own: Prone is only ever something that happens to you.

Checks and Balances

The work ran as a conversation, and both sides caught each other’s mistakes.

Outcome

One chapter now teaches movement to both kinds of table with the same numbers. It covers a single round, leaving a fight, getting low and taking cover, bad ground, climbing, jumping, swimming, creatures that move differently, chases and travel, plus every action a round can buy. The rules page, the book chapter, the print file, the glossary, the status-effects page, the quick references and two reference sheets changed together, and the project’s consistency checks passed on every file.

Next is the part no simulation replaces: the team’s review and a playtest this weekend. What the table finds goes into the record the same way everything else did.

Credits

Decisions and rulings	Me, as producer
Audits, drafts, mining sources, simulation and charts	Claude (AI assistant), checked by me
Team review and playtest	The 20 Below team and playtesters, scheduled