BONUS! Maximizing Surprise

Zhihuan Huang, Yuqing Kong, Tracy Xiao Liu, Grant Schoenebeck, Shengwei Xu

Abstract

Maintaining audience engagement in multi-round competitions requires balancing the significance of early performance with the potential for late-game drama. This paper presents a formal mathematical framework for determining the optimal "bonus points" to award in a final round to maximize "expected surprise"—a metric for the shift in audience beliefs. Using Beta distributions to model priors, we demonstrate that the ideal bonus is often proportional to the "expected lead" of the dominant player. Our findings provide actionable rules of thumb for game designers to sustain excitement throughout a competition.

Introduction: The Challenge of Competitive Drama

In sports and gaming, a common pitfall is the "decided outcome": a scenario where one player establishes such a dominant lead that the final rounds become a formality. When the result feels certain, audience interest wanes. To counter this, many formats—ranging from IndyCar racing to the fictional Quidditch—employ "catch-up" mechanics, such as awarding double or triple points in the final round.

However, setting this bonus is a delicate balancing act. A bonus that is too small fails to revive the drama, while one that is too large trivializes the preceding rounds, making the early competition feel meaningless. This paper addresses the "Key Question": Given a multi-round competition between two players, what point value for the final round maximizes the overall expected surprise?

Modeling Audience Beliefs

The authors define "surprise" as a formal mathematical metric measuring the magnitude of change in an audience's belief about the eventual winner from one round to the next. To quantify this, they model the audience's prior beliefs about player abilities using the Beta distribution, $Be(\alpha, \beta)$.

Figure 3: Beta distribution priors used to model audience beliefs across different scenarios, including symmetric match-ups and cases with high certainty.

As shown in Figure 3, these priors allow for the representation of various competitive landscapes:

  • Symmetric ($\alpha = \beta$): Players are perceived as evenly matched.
  • Certain ($\alpha, \beta \to \infty$): The audience has high confidence in the players' relative skill levels.
  • General: A flexible model for asymmetric match-ups with varying levels of uncertainty.

The Mathematics of Surprise

The competition consists of $n$ rounds. In the first $n-1$ rounds, the winner receives 1 point; in the final round, the winner receives $x$ points. The paper’s primary technical contribution is a lemma proving that for Beta priors, the ratio of expected surprise between any two early rounds is a constant, independent of the final bonus $x$.

This simplification reduces a complex multi-round optimization problem into a manageable trade-off between the surprise generated in the final round versus the surprise in the penultimate round. By analyzing this trade-off, the authors developed an $O(n)$ algorithm to find the exact optimal bonus, a significant improvement over naive $O(n^3)$ approaches.

Key Insights: The "Expected Lead" Rule

The research yields several "rules of thumb" for game design. A primary finding is the Expected Lead Rule: when the audience is fairly certain about player abilities, the optimal bonus $x$ is approximately equal to the number of points the weaker player is expected to be trailing by before the final round.

Figure 5: Contour plot illustrating the optimal bonus ratio as a function of skewness (player asymmetry) and uncertainty.

As illustrated in Figure 5, the optimal bonus is highly sensitive to two factors:

  1. Skewness: As the match-up becomes more uneven (higher skewness), the required bonus increases. To keep an audience engaged in a "David vs. Goliath" scenario, the final round must offer a massive reward to maintain the underdog's mathematical viability.
  2. Uncertainty: In symmetric cases, higher uncertainty about true skill levels actually necessitates a larger bonus. Uncertainty often results in lopsided scores by pure chance; a significant bonus acts as a "dam" for information, ensuring the most dramatic shift in belief occurs at the climax.

Conclusion and Future Work

While this framework provides a robust starting point, the authors acknowledge several avenues for refinement. The current model treats surprise in early and late rounds as equally valuable, whereas psychological "recency effects" suggest audiences prioritize late-game drama. Additionally, many real-world games use dynamic or escalating scoring rather than a single final jump.

Future research could involve A/B testing these mathematically optimal bonuses in digital environments, such as e-sports or quiz apps, to validate whether maximizing theoretical surprise translates directly to increased user retention and viewer satisfaction.

Summary updated

Abstract

Multi-round competitions often double or triple the points awarded in the final round, calling it a bonus, to maximize spectators’ excitement. In a two-player competition with n rounds, we aim to derive the optimal bonus size to maximize the audience’s overall expected surprise (as defined in [7]). We model the audience’s prior belief over the two players’ ability levels as a beta distribution. Using a novel analysis that clarifies and simplifies the computation, we find that the optimal bonus depends greatly upon the prior belief and obtain solutions of various forms for both the case of a finite number of rounds and the asymptotic case. In an interesting special case, we show that the optimal bonus approximately and asymptotically equals to the “expected lead”, the number of points the weaker player will need to come back in expectation. Moreover, we observe that priors with a higher skewness lead to a higher optimal bonus size, and in the symmetric case, priors with a higher uncertainty also lead to a higher optimal bonus size. This matches our intuition since a highly asymmetric prior leads to a high “expected lead”, and a highly uncertain symmetric prior often leads to a lopsided game, which again benefits from a larger bonus.
BibTeX
@inproceedings{huang2021bonus,
  author = {Zhihuan Huang and Yuqing Kong and Tracy Xiao Liu and Grant Schoenebeck and Shengwei Xu},
  title = {{BONUS! Maximizing Surprise}},
  booktitle = {The Web Conference},
  year = {2021},
  doi = {10.1145/3485447.3512049},
}
Figure