Degrading network capacity may improve performance: private versus public monitoring in the Braess Paradox

Abstract

The Braess Paradox (BP) is a counterintuitive finding that degrading a network that is susceptible to congestion may decrease the equilibrium travel cost for each of its users. We illustrate this paradox with two networks: a basic network with four alternative routes from a single origin to a single destination, and an augmented network with six alternative routes. We construct the equilibrium solutions to these two networks, which jointly give rise to the paradox, and subject them to experimental testing. Our purpose is to test the generality of the BP when the network is enriched as well as the effects of the information provided to the network users when they conclude their travel. To this end, we compare experimentally two information conditions when each of the two networks is iterated in time. Under public monitoring each user is accurately informed of the route choices and payoffs of all the users, whereas under private monitoring she is only informed of her own payoff. Under both information conditions, over iterations of the basic and augmented games, aggregate route choices converge to equilibrium. Keywords Braess Paradox · Congested traffic networks · Route choice · ExperimentElectronic supplementary material The online version of this article

Distributed Decisions in Networks: Laboratory Study of Routing Splittable Flow

“…The major contribution of this article is the analysis of route choice behavior in traffic networks with splittable flow in which coordination is centralized within but not between cohorts. Our analysis of game G(S4) further extends the findings of earlier studies (e.g., Gisches and Rapoport , Morgan et al. , Rapoport et al.…”

“…Implementing a within-subject design, our results show that play in the basic network converges to equilibrium quite rapidly within 15 or so rounds. In contrast, route choices in the augmented network gradually approach, but do not converge, to equilibrium at a considerably slower rate (see Rapoport 2012 andRapoport et al 2008 for similar results). If convergence is to be established in the laboratory, longer sessions with more rounds might be required.…”

Counterintuitive properties of the fixation time in network-structured populations

“…A similar phenomenon has been reported in transportation systems, where the flow through a system can be increased if connections are removed [34]. The Braess Paradox describes the situation where adding a seemingly helpful link can have a negative effect on the flow through the network [31].…”

Replicating human interaction in Braess Paradox

“…Others investigated the effect of private monitoring, that is, at each turn individuals were informed only of their performance and not of the whole network performance as in public monitoring (Morgan, Orzen, and Sefton 2009). Yet, no patterns of route choice that increase efficiency were observed (Gisches and Rapoport 2012).…”