There tends to be an assumption that more information is always better but it ultimately depends on how one interprets and uses that information (time and absorptive capacity is important). Most models that push for more information assume perfect decision making (utilizing all the information effectively and optimally). That is not always the case (I might even argue often).
Information overload can lead to analysis paralysis, adding complexity to properly weighing information, and ultimately land into poor decision making processes. In the case of salary information, perhaps I perceive that I'm a below average worker. When handed the average, I might be willing to low ball myself because my thought process may think pay is directly correlated to skill or value given to a company and if I perceive my ability to produce as lower than others, I may be willing, during a negotiation process, to accept low balled values.
In the era of the data deluge, we're learning it's not always the case that more information is better. More and even high quality information needs high quality decision making. If you think of human decision making as flawed (hint: we all make flawed decisions so it is), then our processes will not always make good use of new information, even with high quality concise information. It's very possible for more information to result in undesired or negative effects.
We like to pretend we have all sorts of oracle systems out there but we really don't. In the case of knowing salaries I tend to agree with you: knowing what your competition make is better, most of the time.