"The idea of Twitter as a citywide suggestions and complaints box came into the public consciousness over a year ago, when a group of researchers at Purdue University searched thousands of geo-tagged tweets from presumed Chicago 'L' train riders to gauge broad sentiment and identify common complaints. The word clouds created from their analysis show how the frequency of certain words and phrases rose and fell as trains were delayed or, for example, faced a signal fire or fallen tree.
A key discovery was that people rarely make positive comments - you'll look long and hard before finding a tweet along the lines of "the Purple train is running wonderfully this morning!" -- instead saving their energy and 140 characters for problems, delays and unexpected difficulties. This may not sound terribly friendly, but it's ideal for municipal officials, who have little need to hear about systems running smoothly."
Thanks to Colin Scarff