Most real-world data contains repetitive components. Examples include sequences of constant values such as a video image that doesn’t change from one frame to the next, a document with sections that are all one level (e.g. white), or repetitive data values in a file or database.
Two possible problems are introduced by this non-random data:
1) long sequences of certain values (zeroes, ones) will result in a signal that may not have enough transitions to allow for clock recovery (for data recovery and synchronization). You can read basics of "clock recovery" even from wikipedia.
2) Morevoer, when these (long sequences of zeroes or ones) values are transmitted, the periodic components of the signal result in peaks in the spectrum that have larger than average power. These strong discrete frequency components can cause interference.
To solve these problems most communication systems use “scramblers” to remove undesirable patterns in the data.