This is actually due to an unfortunate default value for lines.markeredgewidth = 0 in seaborn.mplstyle.

For now, a possible workaround is using the flierprops parameter:

plt.boxplot([0,0,0,1], flierprops={'markeredgewidth': 1})

This is actually due to an unfortunate default value for lines.markeredgewidth = 0 in seaborn.mplstyle.

It's also due to the fact that matplotlib's style sheets are out of sync with actual parameters seaborn sets:

import seaborn as sns
import matplotlib.pyplot as plt

sns.set()
plt.boxplot([0, 1, 2] * 50 + [10])
print(sns.__version__)

0.9.0

(By the way, the original example surprises me for other reasons. In Tukey's definition, box plots should only represents values that exist in the data set. How is 0.2 getting in there?)