1) Internet tech doesn't necessarily discourage local, face-to-face friendships. Right -now- the Web isn't used for local connection but I think that's just because of the way it's -framed-, just a momentary lack of vision by the people/firms building it. And I think that's a temporary anomaly that's disappearing as wi-fi and locative tech takes hold.
Remember the net evolved from a set of LANs, and even as recently as the 80s, the folks who inhabited the dial-up BBS world were very locally-focused (you dialed BBS's in/near your town most often because those phone calls were cheapest) and many of these people got to know the other local BBSers through face-to-face get togethers. These "GTs" were an important part of BBS culture. More recent examples -- Google the study "Neighboring in Netville" to learn fascinating things that happened when researchers wired 1 out of every 3 homes in a typical suburban housing development outside Toronto w/ very simple terminals attached to a basic message forum system tied to a proprietary LAN. The people who moved in weren't techies, but nonetheless after a year this neighborhood was measurably more cohesive and local connections were much stronger than in neighboring unwired subdivisions that otherwise were almost identical in physical structure and demographics. In short: in the wired subdivision a lot more people knew their neighbors and other folks nearby, and the community as a whole was much more politically active in tracking and responding to issues that affected the good of the neighborhood. All because networked communication tech was -framed- as something that connects you to people nearby -- not just as something that connects you to the placeless Web, not -just- something that's for finding people on the other side of the planet who share precisely the same interests that you do.
2) Back to this "Internet to blame" study, note an important point the researchers themselves make: that the wording of the survey questions might have strongly affected the results and their interpretation. (i.e., 2004 respondents might have thought "discussing" doesn't include e-mail/IM.)