2026 SPRING
Against the Oracle
DATE
Sep 2025 - May 2026
CATEGORY
Prototype
DESIGNER
Jane Jianhan Lu
MENTOR
Elise Co, Maxim Safioulline
Sep 2025 - May 2026
CATEGORY
Prototype
DESIGNER
Jane Jianhan Lu
MENTOR
Elise Co, Maxim Safioulline
Play a game of chess, and I’ll tell you your fortune.
This thesis examines structured play as a cultural technology for anticipating the future. Systems such as chess engines, AI prediction models, and strategic simulations are widely trusted as rational tools for forecasting, yet they operate through the same underlying mechanism as older divinatory practices: pattern recognition derived from historical archives. Both translate accumulated past events into probabilistic guidance about what may come next.
The project explores this shared logic through a chess-based prototype that makes predictive computation perceptible rather than hidden. Instead of presenting a single optimal answer, the system exposes evaluation, risk, and suggestion as signals the player must interpret. Prediction is reframed as a process of reading patterns and navigating probability rather than receiving certainty.
In this framework, structured play functions less as prophecy and more as rehearsal. Predictive systems reorganize historical patterns to shape present decisions, training participants to recognize recurrence, constraint, and uncertainty. The work ultimately positions forecasting—whether computational or symbolic—as a soft instrument for thinking and acting within the limits of what the past can tell us about the future.
This thesis examines structured play as a cultural technology for anticipating the future. Systems such as chess engines, AI prediction models, and strategic simulations are widely trusted as rational tools for forecasting, yet they operate through the same underlying mechanism as older divinatory practices: pattern recognition derived from historical archives. Both translate accumulated past events into probabilistic guidance about what may come next.
The project explores this shared logic through a chess-based prototype that makes predictive computation perceptible rather than hidden. Instead of presenting a single optimal answer, the system exposes evaluation, risk, and suggestion as signals the player must interpret. Prediction is reframed as a process of reading patterns and navigating probability rather than receiving certainty.
In this framework, structured play functions less as prophecy and more as rehearsal. Predictive systems reorganize historical patterns to shape present decisions, training participants to recognize recurrence, constraint, and uncertainty. The work ultimately positions forecasting—whether computational or symbolic—as a soft instrument for thinking and acting within the limits of what the past can tell us about the future.
Flow chart: how it works
Omen reading output sample
Version 1.0
THE COMPASS
![]()
The compass began as a question about visibility. If a chess engine trained on my own games already knows my probable next move, what happens when that prediction is shown to my opponent — while I remain blind to it?
The first prototype was simple: an 8×8 LED matrix in a chrome shell, sitting between me and whoever I was playing, lighting up my most likely next move for them to see. I typed moves into Terminal; the board responded. No ritual framing, no printed readings — just the prediction, made visible.
Playtesting with two cohorts revealed something unexpected. Novice players treated the prediction as instruction — a way to orient themselves in a system they couldn't yet read. Experienced players used it differently: they were more interested in the inaccuracies than the accurate predictions. One opponent told me that through the moments the compass was wrong — when I played offscript — she got to know me as a person better than through any correct prediction. The glitches weren't failures. They were the person behind the pattern becoming visible.
Both cohorts found reassurance in the machine regardless of its accuracy. That was the finding that mattered: the value wasn't in being right. It was in providing something to relate to in the face of uncertainty.
The compass began as a question about visibility. If a chess engine trained on my own games already knows my probable next move, what happens when that prediction is shown to my opponent — while I remain blind to it?
The first prototype was simple: an 8×8 LED matrix in a chrome shell, sitting between me and whoever I was playing, lighting up my most likely next move for them to see. I typed moves into Terminal; the board responded. No ritual framing, no printed readings — just the prediction, made visible.
Playtesting with two cohorts revealed something unexpected. Novice players treated the prediction as instruction — a way to orient themselves in a system they couldn't yet read. Experienced players used it differently: they were more interested in the inaccuracies than the accurate predictions. One opponent told me that through the moments the compass was wrong — when I played offscript — she got to know me as a person better than through any correct prediction. The glitches weren't failures. They were the person behind the pattern becoming visible.
Both cohorts found reassurance in the machine regardless of its accuracy. That was the finding that mattered: the value wasn't in being right. It was in providing something to relate to in the face of uncertainty.
Deployment of the Compass 1.0
Final Apparatus
HARDWARE & DEPLOYMENT
The complete apparatus introduced three additions to the compass logic: a reactive LED danger map, a win-probability OLED, and a thermal printer generating symbolic readings keyed to piece archetypes, I Ching-derived flux marks, and Stockfish evaluation.
The printer triggers only at moments of consequence — when a player deviates from the compass suggestion, or makes a blunder, sacrifice, or brilliant move. Each slip is earned rather than given.
Four participants playtested the complete system. No two used it the same way. One player ignored the win probability display entirely — she didn't want the numbers to discourage her — and used the compass only retrospectively to check moves she'd already made. Another treated the compass and probability display as two friends giving him advice, consulting both to see if they agreed before trusting himself. A third barely looked at the hardware at all, reading every printed slip carefully, connecting its language to her playing style and then to her personality more broadly: who am I in chess equals who am I in life, she said, unprompted.
The sharpest finding came from the most experienced player. After receiving a reading, she went through the entire slip trying to match its language to specific moves. The tonal framing alone — positive or negative — convinced her that certain moves had been correct and others wrong, even when the text had no direct relationship to the engine's evaluation. She described it as definitive.
The system wasn't describing her moves accurately. It was describing them symbolically — and that was enough to produce conviction.
Exhibition
