This essay reviews Dani Rodrik's superb book Economics Rules and argues that it can serve as an ideal platform for discussing what economists can and should accomplish. The essay comments on some of the major issues in contemporary economics examined in the book: whether economics is a science, the meaning of economic models, the nature of “facts” in economics, and others. It also touches on issues that the book overlooks, such as the sociology of the profession, the teaching curriculum in economics, and the dismal situation of publishing in economics.

What on earth are economic theorists like me trying to accomplish? This paper discusses four dilemmas encountered by an economic theorist: The dilemma of absurd conclusions: Should we abandon a model if it produces absurd conclusions or should we regard a model as a very limited set of assumptions that will inevitably fail in some contexts? The dilemma of responding to evidence: Should our models be judged according to experimental results? The dilemma of modelless regularities: Should models provide the hypothesis for testing or are they simply exercises in logic that have no use in identifying regularities? The dilemma of relevance: Do we have the right to offer advice or to make statements that are intended to influence the real world?

A single seller of an indivisible good operates in a market with many consumers who differ in their ability to process information. The consumers' constraints are modeled in two submodels: the first in terms of the limits on the number of sets in the partition of the price space, and the second in terms of the limits on the complexity of the operation he can use to process a price offer. For the construction of the second submodel, the tool of a "perceptron" is borrowed from the parallel computation literature. Assuming a negative correlation between the seller's cost of supply of the good and the consumer's ability to process information, I demonstrate that the heterogeneity of consumer's abilities can be used by the seller to profitably discriminate among them.

A speaker wishes to persuade a listener to accept a certain request. The conditions under which the request is justified, from the listener's point of view, depend on the values of two aspects. The values of the aspects are known only to the speaker and the listener can check the value of at most one. A mechanism specifies a set of messages that the speaker can send and a rule that determines the listener's response, namely, which aspect he checks and whether he accepts or rejects the speaker's request. We study mechanisms that maximize the probability that the listener accepts the request when it is justified and rejects the request when it is unjustified, given that the speaker maximizes the probability that his request is accepted. We show that a simple optimal mechanism exists and can be found by solving a linear programming problem in which the set of constraints is derived from what we call the L-principle.

A new model of persuasion is presented. A listener first announces and commits to a codex (i.e., a set of conditions). The speaker then presents a (not necessarily true) profile that must satisfy the codex in order for the listener to be persuaded. The speaker is boundedly rational in the sense that his ability to come up with a persuasive profile is limited and depends on the true profile and the content and framing of the codex. The circumstances under which the listener can design a codex that will implement his goal are fully characterized.