Note that for all three of these we are talking about interpersonal relationships, a field where many technologists and engineers feel uncomfortable at best. The reality of life is "people count" ... and relationships are a critical element of your future success. So let me try to give some definition and distinction between these concepts.
Networking: This is simply becoming known to other persons, and critically getting to know them. It parallels the "direct connections" between nodes in a communications network, and thinking about it as "establishing links" and knowing which links to use to effectively communicate a given message (routing) you won't be too far off. Except that communications links between people have differing filters, priority structures, protocols, and encoding.
Social Networking: Social networking is the subset of the networking concept above that exists, in theory, to develop friendships, acquaintances, and such. Having a party, or attending one -- inviting your friends out for beer or coffee or a movie -- joining a bowling league, a quilting club, a hiking group, etc. Notice that doing these things from the your "set of neighbors" or "set of co-workers" or as part of a professional conference you are attending puts a slightly different spin on the initial topics of discussion, etc.
Social Capital: Trust relationships (networks) that provide a basis for obtaining assistance or collaboration. Notice that social capital involves a bit more than just 'recognizing' the other person -- it involves having a mutual awareness of common values and interests and an "appropriate level of trust" related to the requested investment of said social capital. If I have a neighbor I've met at a party, and I know she works at a company that hires folks in my field, it is typically "ok" to contact that person and ask if they know of openings at the company and who I might contact; and probably ok if I contact her to contribute to some charity activity. If you don't have an awareness of common interests, it is less clear that you can ask a person to join you in setting up a Robotics team at the local elementary school, etc.
Regular interactions with a community of folk (a few times a year typically), with opportunities to inter-act informally (hall-way discussions, at the lunch table, etc.) -- provides a chance to pickup on areas of common interest. If you have interactions over a few years (not uncommon if you are on common committees - standards, management, periodical review, conference, or even coordinating the church bazaar) -- you will find out about common interests in black holes, RFID privacy, Yucatan cenotes, bridge, muti-core processors, etc. The result is you now have the 'routing table' that will let you know whom you might consider inviting on that next dive trip to Mexico, or local lecture at the college, ask to speak at a chapter meeting, or to join you in a new start-up company. And, they have the same insight in return.
The more practical question, is how much, or which things can you do online? The answer may differ between the "Generation Net" and older folks who don't IM their friends at school to see about lunch. Personally (being one of the older folks), I find having met a person, "in person" before helps me gauge their style of interactions. Early in my career I was being mentored by an "old hand" whom I'd never met, via phone calls. His style was gruff and strongly opinionated -- I was intimidated. I met him at a meeting and quickly realized that there was a 'twinkle' in his eye and humor that was not obvious in the phone calls. From that point on I could picture his personality as a factor in the phone calls, and these became much more productive. In the reverse, I had a manager at one time with a very short fuse and limited tolerance -- watching him interact with others, and a few times myself, I quickly learned to read every email I was about to send him twice and make sure it was 'going to work' the right way. In short - I am skeptical that the same level of trust can be developed without some level of face-2-face interaction. At the same time, I have collaborated online with colleagues I've never met in person, co-authored papers, and established a level of mutual awareness.
The only reasons I can envision avoiding face to face awareness is either a strong personal sense of inadequacy ("I am so ugly..."); or perhaps a more rational awareness that the other individual has prejudices that might interfere with professional interactions. (I wish I could assert that this doesn't happen; but there are solid reasons why your resume should be obscure about race, sex and age -- these biases can be played out with little or no awareness on the part of hiring managers, etc.) Which leads to....
These represent one variation of social capital. You have that group of folks you went to school with, or even the ones you did not know that are alumni. This provides a 'hidden' advantage -- your resume says you went to that school, and they immediately identify a common connection that probably favors your situation. Why -- the known, or implied shared experience that gives you a basis for interaction.
Old boys clubs develop the same way, but often independent of "school". Folks interact, they form groups they are comfortable with -- people they trust -- sharing common activities: golf or a drink at the club. The "old" is often a reality for a few reasons -- it takes time to have these de facto structures form, and often the old folks hold the power -- and that becomes an implicit common ground. I suspect there are "young" boys clubs that have formed in Silicon Valley and other areas where wealth and power have emerged in a younger group. Do they have to be "boys?" -- I don't think there is any biological reason. Some of the "clubs" are explicitly restricted by sex, race, religion, etc. There is a more insidious reason which has to do with how social capital is developed. Who do you trust? How do you build those relationships? Often it is established among "persons like me" -- In many cultures, men have had the power, they associated with other men in this context, and those who gained 'entry' to the club were other men. This does not require explicit or intentional discrimination, those persons feeling comfortable moving in closer to the center and the folks in the center feeling comfortable with the addition of the new persons just works that way.
Here professional societies can have a real advantage. Few professions have implicit gender or age barriers, and it is hard to imagine ones that could justify racial barriers. (It is hard to get on an NFL football team at my age, or if you are a woman -- so there are some barriers along these lines.) Groups like IEEE CS are explicitly open to all related professionals -- and while we may not attract or engage as diverse of volunteer base as we might want -- folks are welcome and efforts are made to support all of the diverse communities. There are specific groups that focus on Women in Engineering, and Graduates of the Last Decade -- these are efforts to assure we minimize both the "Old" and the "Boys" aspect of our operations. No doubt more energy and investment is needed to broaden our engagement of professionals world wide -- and in diverse fields. Areas like India and China may face economic barriers in becoming fully involved; fields like Information Technology may feel 'excluded' by our references to "engineers" ... so we continue to face challenges along these lines.