I took a two-pronged approach. Ultimately, I've observed that it could hurt or it could help or it could do neither. Here are some thoughts, and I will note that I applied to some STEM and non-STEM programs, 10 in total.
1. Some departments or faculty webpages are very explicit in saying "do not contact advisers in advance". I did not contact those faculty or departments where this was the case. There is a reason that these departments or faculty make the request to not contact them public and I wanted to honor that.
2. In the case where no written policy was stated (or if the faculty/department encouraged contacting people ahead of time), I did that. But my approach was not to simply get them to notice me. My gut says you can probably learn about 1/3 of what you need to know about a department/faculty member just by looking at their basic information online. You can learn a bit more if you draw inferences from things like their manuscripts, etc. (e.g., a faculty webpage may show publications but be out of date, but a simple look in google scholar or sciencedirect can give you a clearer picture of productivity).
So if I did contact faculty, it was for a specific purpose and only to get answers to questions that were not already answered. Also, if I contacted someone, I would only do it once and would only re-contact if they replied to me. My initial correspondence was kept short and to the point. If they requested more information from me (they would often ask for a CV and some other information if they wanted to get to know me more) I would provide it to them, but I would not provide it to them unless requested.
Here is an excellent website that gives you perspective of a professor at a top school. Many of his sentiments are true for professors at most schools: https://sites.google.com/site/chuckeesley/before-emailing-me
Since it is still early in the decision process, it's hard for me to say whether or not my approach was effective. I will say that the faculty I contacted and corresponded with early on was a very good experience and I was left with very positive feelings and believe that correspondence will help them remember me when evaluations begin. Furthermore, a couple of the correspondences ended up enlightening me to the fact that the school/faculty was not a good personality or research fit, even though my "desktop" analysis indicated otherwise.
To sum it up with stats:
Total programs applied to: 10 (MIT, Berkeley, Yale, Stanford, U Chicago, UIUC, etc.)
Total programs where I contacted at least one faculty before the application season: 5
Acceptances from schools I contacted before the application season: 1
Interviews from school I contacted before the application season: 1 (offer of admission is forthcoming)
Total programs where no decision has been made: 8