Anonymity in Predicting the Future

TL;DR

This paper explores the limits of anonymous guessing strategies in predicting a system's future state based on past information, examining how various uncertainties affect the success of such predictions.

Contribution

It introduces and analyzes new variations of anonymity in guessing strategies, extending previous work by considering additional uncertainties like unknown time rates and intervals.

Findings

Agents can sometimes guess successfully despite uncertainties.

Certain anonymity conditions lead to poor guessing performance.

The study delineates when anonymity hampers or allows accurate predictions.

Abstract

Consider an arbitrary set $S$ and an arbitrary function $f : \mathbb{R} \to S$. We think of the domain of $f$ as representing time, and for each $x \in \mathbb{R}$, we think of $f(x)$ as the state of some system at time $x$. Imagine that, at each time $x$, there is an agent who can see $f \upharpoonright (-\infty, x)$ and is trying to guess $f(x)$--in other words, the agent is trying to guess the present state of the system from its past history. In a 2008 paper, Christopher Hardin and Alan Taylor use the axiom of choice to construct a strategy that the agents can use to guarantee that, for every function $f$, all but countably many of them will guess correctly. In a 2013 monograph they introduce the idea of anonymous guessing strategies, in which the agents can see the past but don't know where they are located in time. In this paper we consider a number of variations on anonymity. For instance, what if, in addition to not knowing where they are located in time, agents also do not know the rate at which time is progressing? What if they have no sense of how much time elapses between any two events? We show that in some cases agents can still guess successfully, while in others they perform very poorly.