Play with Growth: test your intuition
Make a choice, see what unfolds, and try again.
01 · One grain of rice
THE LEGEND OF SISSA
A reward that sounds almost too small.
In the Indian legend retold by Cambridge’s NRICH, Sissa invents a board game and the king offers him a reward. Sissa asks for rice: one grain on the first square, two on the second, four on the third, then twice as many on each new square. The king thinks the request is silly. What could be hiding in such a small beginning?
Sources, methods, and data ↓YOUR TURN
Follow the grains.
Tap a square, drag the slider, or let the board play. Start at the top left and read across each row.
1 → 2 → 4 → 8 → …
Grains on this square
1Total grains so far
1One grain. An easy promise to make.
The counters are exact. The small grain symbols on the board are illustrative once a square holds more than eight grains.
Two rules. Very different journeys.
Grains on each square · linear scale. The chart follows your selected rule; choose doubling to see both curves together. The vertical scale adjusts as you move.
Explore sources & data ↓THE REVEAL
The last square changes everything.
The request that sounded so modest contains an astonishing total. This is what the king’s promise would require.
18,446,744,073,709,551,615
grains of rice across the entire board.
The final square alone holds one grain more than all 63 earlier squares combined.
Adding one grain each time would need only 2,080 grains in total. Doubling reaches about 8.9 quadrillion times that amount.
For scale: assuming 20 milligrams per grain, the doubling total would weigh about 369 billion metric tonnes. This is an illustrative conversion, not a measured grain weight or a prediction.
FROM A CHESSBOARD TO OUR WORLD
A constant rule can create accelerating change.
Real technologies do not double forever. Resources, adoption and institutions shape their paths. But the board shows why a steady percentage increase can bring an ever larger change—and why our capacity to adapt matters.
Rapid growth can bring opportunity. The question is how knowledge, governance and tools such as AI can help us respond.
Explore acceleration & absorption →Sources, methods, and data
Behind the story and the numbers
This is a retelling of a legend, not a documented historical event or a quotation from the 2016 book.
- University of Cambridge · NRICH · Sissa’s Reward ↗The telling used here: Sissa, the king and the doubling request.
- Glen Pugh · Vancouver Island University · Sequences and Series, p. 11 ↗The geometric sum and an illustrative mass of 20 milligrams per grain.
Doubling
On square n: 2ⁿ⁻¹ · Total through n: 2ⁿ − 1
Adding one
On square n: n · Total through n: n(n + 1) / 2
The graph compares grains on each square, not cumulative totals. The add-one sequence is linear; its cumulative total is not. Exact whole-number arithmetic is used for all grain counters; chart positions and compact axis labels are rounded for display.
All Atlas sources & data →
02 · 2016 → 2026 quiz
YOUR EXPECTATIONS. THE EVIDENCE.
Return to the questions raised by the 2016 book. Choose what you think the following years revealed, then compare your intuition with dated evidence.
Four questions across population, connectivity and AI. This scores your answers, not the book’s forecasting accuracy. Evidence ends in different years; none of these questions asserts a complete 2026 outcome.
Sources, methods, and data
Inspired by
Gapminder’s Worldview Upgrader invites readers to test their assumptions against evidence. Our questions, answer choices, explanations and book connections were written specifically for this website.
- Gapminder · Worldview Upgrader ↗
- UN Population Division · Day of Eight Billion (2022) ↗
- UN · World Population Prospects 2022 announcement ↗
- ITU · Global Internet use estimate (2025) ↗
- Stanford HAI · AI Index 2025 · Inference prices ↗
Created for this website
The connection to 2016 follows the author’s retrospective in Part I of the 2026 book. Context sentences are paraphrases of its themes, not quotations or newly attributed numerical forecasts. Only the population milestone is presented as a dated expectation from that comparison.
Data & limits
Population milestones and Internet use are statistical estimates. AI prices compare models reaching a specified MMLU benchmark threshold, not every AI task or total deployment costs. Dates and scope accompany every answer. The alternative answers are game choices, not claims made by the original book. Sources checked in October 2026.
03 · Can you stop in time?
GROWTH × RESPONSE TIME
Grow the glowing colony as close to 100% as you can without crossing the boundary. The stop command may take several turns to work. How long will you wait?
The colony starts at 3% and grows by 60% each turn. A two-turn delay means two more growth steps after your stop request. Reaching 100% is allowed; crossing it ends the attempt.
Sources, methods, and data
Inspired by
John Sterman’s account of the MIT Beer Game explains how delayed effects complicate decisions. This game uses that general insight in a different, single-player stopping challenge; it does not reproduce the Beer Game.
Created for this website
The colony setting, turn-based rules, delay choices, scoring, graphics and code are original to The Accelerating World.
Model & limits
Illustrative arithmetic: filled percentage = 3 × 1.6^turn. The selected delay is fixed for an attempt. Growth stops completely when the command arrives. A successful score equals the final filled percentage; crossing 100% gives zero. This is not a biological model or a forecast. Real growth rates, detection and interventions are less predictable.
04 · The shrimp lagoon
A LIVING RESOURCE. TEN SEASONS.
Bring in a good harvest while leaving a thriving lagoon. Choose a catch each season—or skip a harvest to restore habitat. Can you finish with at least 600 shrimp units and harvest more than 1,300?
A fictional lagoon, measured in model units—not individual animals or tonnes. Reproduction depends on the stock left behind and the habitat’s capacity. You can restore habitat twice.
Season-by-season record
Sources, methods, and data
Inspired by
Fishbanks, developed by Dennis Meadows, John Sterman and Andrew King, explores renewable resources and collective decisions. Our lagoon has no competing fleets, boat market or copied scenarios. It uses original single-player harvest and restoration choices.
Created for this website
The shrimp setting, ten-season objective, two habitat restorations, comparison strategies, illustrations and code were designed for The Accelerating World. No code, artwork or rule text from Fishbanks is used.
Model & limits
Start with 600 stock units and capacity K = 1,000. After harvesting, remaining stock R gains 0.6 × R × (1 − R/K) units; stock is bounded between zero and K. Restoration replaces that season’s harvest and permanently adds 200 capacity units, up to twice. All parameters and the success targets are illustrative choices.
Results retain full precision; displayed quantities are rounded. Comparison strategies harvest 20% or 50% each season with no restoration under the same starting conditions. They are examples, not optimal strategies. The model omits shrimp species, life stages, seasons, water quality, disease, prices and external pressures. It is not aquaculture advice or a biological forecast.
05 · Make it spread
CONNECTEDNESS / A SMALL CHANGE, A DIFFERENT OUTCOME
Help a useful idea reach all twelve workshops. Choose three starting champions, pick an adoption rule, and watch the connections do the rest. Will your idea cross every community?
Tap a numbered workshop to select or deselect a champion before starting. Lines are two-way connections. Gold outlines mark your starting choices.
Every workshop checks its neighbours at the start of a round. New adopters can influence others in the next round. Choose exactly three champions to begin.
Sources, methods, and data
Inspired by
Nicky Case’s The Wisdom and/or Madness of Crowds explores networks and the difference between simple spread and adoption needing reinforcement. We use that general idea in a new workshop puzzle with a fixed network and selectable champions.
Created for this website
The three workshop communities, twelve-node layout, champion choices, optional bridges, paired replay and all code, wording and graphics were created for The Accelerating World. No existing game assets or code are reused.
Model & limits
Each four-workshop community is fully connected. The initial bridges are A4–B1 and B4–C1; optional bridges are A1–B2 and B1–C2. A workshop adopts when at least one or two distinct neighbours are already active, depending on your rule. Updates happen simultaneously, and adoption is permanent. The threshold is an absolute count, not a percentage of neighbours.
This deterministic illustration omits trust, persuasion, costs, unequal influence and changing networks. Extra links cannot reduce adoption under these particular rules; other models can behave differently. Wide adoption does not establish that an idea is true or beneficial. No real-world diffusion rate is predicted.
06 · The efficiency surprise
BETTER PER TASK. WHAT ABOUT THE TOTAL?
A workshop upgrades to machines that use half as much energy per task. Predict its total energy use in three demand scenarios. Then build your own combination of efficiency and growth.
The starting workshop completes 100 tasks using 100 energy units. All quantities are illustrative. A prediction within five units counts as accurate.
The energy axis stays fixed from 0 to 400 units. The result stays hidden until you submit your prediction.
Now change the assumptions.
Can you increase useful output while keeping energy at or below the original 100 units?
Sources, methods, and data
Inspired by
Research by Kenneth Gillingham, David Rapson and Gernot Wagner explains how rebound can offset some expected energy savings. Its review does not support treating increased total energy use as an inevitable result of efficiency improvements.
Created for this website
The workshop story, three prediction rounds, fixed-scale chart, scoring and adjustable final challenge are original to The Accelerating World. This is a new arithmetic game, not a reproduction of a published game.
Model & limits
Baseline: 100 tasks × 1 energy unit per task = 100 units. An efficiency multiplier of m reduces energy per task to 1/m. Total energy = tasks/m. The three rounds use m = 2 and task counts of 100, 180 and 300. The sandbox allows m from 1 to 4 and tasks from 100 to 400.
Demand is supplied as a scenario, not calculated from cheaper energy or increased efficiency. The game shows how much demand growth could offset savings; it does not estimate a causal rebound effect. Prices, embodied energy, other resource uses and changes in service quality are excluded. None of these numbers is a measured industry outcome or a forecast.
Each game is an invitation to explore, not a prediction. Open “Sources, methods, and data” to see the inspiration, what we created, and where each model stops being realistic.
