In this paper we present TwitterSigns, an approach to display microblogs on public displays. Two different kinds of microblog entries (tweets) are selected for display: Tweets that were posted in the immediate environment of the display, and tweets that were posted by people associated with the location where the displays are installed (locals). The prototype was tested in a university setting on 4 displays for 4 weeks and compared to the information system that is usually running on the displays (iDisplays). Using face detection we show that people look significantly longer at TwitterSigns than at iDisplays. Interviews show that the relationship of viewer and poster as well as the tweet content are much more important than time and location of the tweet. Viewers recall and recognize mostly tweets from people they know, and of apparent importance for themselves (like a apparent bomb found in the city center). Furthermore, TwitterSigns change the way people use twitter (e.g. they feel ...