Statistical Inference — comparing MSE of a constant guess to variance of the usual estimator

If $\hat\theta_1=\frac{X}{n}$ and $\hat\theta_2=\frac13$ are the estimators of the parameter $\theta$ of a binomial population and $\theta=\frac12$, then the values of n for which the mean square error of $\hat\theta_2$ is less than the variance of $\hat\theta_1$, are