On Stackoverflow.com, I often see this kind of question: "When I read my text on a different computer, it's all garbled."
The underlying issue is that people don't understand the concept of encoding and therefore, they use the default encoding which breaks their code when they least expect it. Worse, it often causes data corruption without throwing exceptions.
Therefore my suggestion: Deprecate and then remove all methods that use the default encoding. Users should always specify an encoding when doing text I/O, to make sure that data cannot be corrupted even when they don't know what they're doing.