A long time ago someone came to present a paper to us that showed that people trading without information (eg momentum traders you mentioned above) impact the stock price much less than those trading with information. So I suspect that the more people know the more efficiently they themselves move the market. This might solve the "too much information" paradox you mention...
On the topic of the efficient market hypothesis, there was a paper published a while back that purportedly disproved the hypothesis in the general case (basically, the potential number of data points can grow larger than the available computational/cognitive power of the market). I'll see if I can dig that up later. I think specific cases could potentially still hold, though, at least theoretically.