The issue of what is the best confidence interval for the probability of success for iid Bernoulli trials is still being debated, and researchers are still studying the problem. I will discuss both older and newish interval estimation procedures for the Bernoulli parameter, and summarize the ongoing argument about which method is best. When the opinions of several leading researchers in the area are compared, only one thing seems to be agreed upon: it is time to finally do away with the standard (Wald) interval which is commonly taught to introductory statistics students, and which is still routinely used by some statisticians who ought to know better.

If time permits I will say a little about the associated point estimation problem. Even for this seemingly simple problem, there are interesting alternatives to the commonly used sample proportion, which is both the maximum likelihood estimator and the uniform minimum variance unbiased estimator.