I can only speak to the blogger POV and extrapolate to the "broader social milieu" from there. The advantage of the POV is that you get a broad look at all the services and can compare them. The disadvantage is that it is a decidedly partial view.

For bloggers, such sites represent the main sources of traffic spikes of various sizes and personalities. The quid pro quo on the other end is that blog links basically pour fuel on existing fires within each community. What you say as a blogger matters less than how the community decides to fit it into its internal discussions as currency. The stronger the community, the more it will transform your link into something else entirely. Usually into a stick for one sub-group to beat up on another one with, or for the whole group to beat up on the link-producer and create inner social capital.

By contrast, the forums/BBSes/Usenet era communities tended to get stale and insular much faster, since link-sharing was not the main driver of the community stream (in a way, link-sharing is anti-evaporative-cooling protection because it makes closed communities more open and therefore more sustainable/long-lived).

Overall, it is a fair trade. Both sides benefit. Bloggers get their new user acquisition via the traffic spikes, the communities get fresh fodder to keep their inner culture from getting stale.

You can classify link sharing sites along two useful dimensions from the blogger POV, each of which can be measured. Spike pattern, and conversion percentage. As a general rule, newer, more "2.0ish" sites send a broader distribution of spikes and convert better. Older sites send spikier traffic and convert worse. Some specifics:

1. Slashdot is the oldest, technically, and it shows. You get huge, server-crashing spikes (I've had two) and significant bragging rights, but the retention rate is fairly poor. My two slashdottings were each worth about 30,000 visits each over the first 2 days. Each led to an RSS bump of 400-500. Or about 2%. The people who stay self-select and are generally positive. The discussion on Slashdot is guaranteed to degenerate into ugliness.
2. Digg is like Slashdot, but rather more arbitrary. Sharp, rare spikes. Unlike Slashdot, there is no discernible tone to the traffic you get. In theory it is community site. In practice, the power users are nearly as autocratic in their ability to drive stuff to the front page as Slashdot editors.
3. StumbleUpon sends lots of spikes of all sizes ranging from a few dozen to a few hundred. It is more truly crowd-based, but not really a "community" in the sense of creating social-proof amplification effects. It tends to be the easiest/first source of spikes for a new blogger. My first ever little spike was an SU spike.
4. HackerNews, in aggregate, has been my best source. Its spikes are not very high (about 4000 visits in a day seems to be the max for me), but you get a range, depending on the community's mood and the other news that day. Once a few community members get you on their radar, you keep getting traffic spikes with a sort of power law distribution. So cumulatively, it has been more valuable than Slashdot. Conversion is harder to track because HN converts over multiple exposures. But I think the conversion is at least 2x. As with Slashdot, inner discussion tends to be more negative than the stuff that shows up on your blog comments, but the tone difference is not as extreme.
5. Reddit, Metacafe etc. etc. much more self-absorbed than the other sources, and for me personally, the spike frequency is lower, but my sense is that they are basically like HN, but with different core discussions. You could say I print HN dollars more frequently than I print Reddit dollars.
6. I should mention the one-person link-sharing industry. I've shown up on Kottke a few times, AndrewSullivan once, BoingBoing a couple of times (via Cory Doctorow) etc. This tends to be a curious traffic flow, since the referrals tend to relate more strongly to the reblogger, and how they choose to frame your stuff. How they frame the referral strongly determines how well the spike will convert.
7. Finally, Quora. Quora may not seem like a linksharing community, but it is, since people post links in answers. Quora traffic is low for me in volume terms, but in terms of conversion quality, it is phenomenally high. People who find my stuff through Quora tend to become very high-value readers, and I am more likely to develop personal relationships with them. Quora is also the only community that is set up to make active blogger participation worthwhile, because it balances individualism/collectivism well. The egalitarian nature of other sites tends to make it counterproductive for bloggers to participate (the few times I've posted on HN in discussions of my own articles, I've mostly been ignored... I get a "we'll take it from here, thanks" vibe).
8. For completeness: Facebook, Google+, Twitter and "Old Media" should be considered. I get traffic from FB but I don't really understand the dynamics of that traffic. From talking to other bloggers, you really need to actively work on an inside-FB page to maximize value. I have a FB page, but basically neglect it. Twitter is like a faster, time-sensitive version of the one-person aggregator industry. Each link retweet cascade is like a little temporary ad-hoc community. The traditional media is basically not important unless you make the front page or something. I've had a few mentions on NY Times blogs and bragging rights aside, the traffic impact is not worth squat.

Stepping back, I'd say that within the larger ecosystem, link-sharing communities are a combination of market-making intermediaries and leaky discovery+distribution channels. "Leaky" because you ultimately don't need them for retention, only for "passing offense" acquisition. I hope the metaphor is clear: organic, WOM growth of a blog, mainly via personal email forwards (still the biggest link sharing mechanism) is a rush offense. The spike traffic gets you the passing offense yards. Trend-following bloggers can therefore grow radically faster than non-trend following ones like me, because they get far more spike-driven growth.

In terms of larger milieu, they are in a nicely complementary relationship with the blogosphere (think of the blogosphere as a crowdfunded link-currency federal reserve and the link-sharing sites as the markets: we bloggers print the money, others trade with it to their own ends). There is some tension and adversarial stuff, but basically, it is a symbiotic relationship.

By contrast, they are really bad for the traditional media I think. There, it's a one-way street. Sources like the NY Times get temporary link amplification via the sharing sites, but get no loyalty or acquisition of direct relationships in return.

They are even worse for the content farming operations (Demand Media, most of AOL...), because they serve as a social filter that basically cuts these guys out of all traffic funnels other than organic search. And with Google cracking down on content farms (they have finally decided to stop trading off search result quality for ad revenue), this 1-2 punch basically means content farms may get shut out of the game entirely.

I have a theory about why the relationships play out this way: basically a more personal relationship can always disintermediate a less personal one. A reader-blogger relationship is more personal than a member-community one. Branded old media is even less personal and content farms are the least personal of all. So bloggers are on the right side of link-sharing sites to benefit, while old media and content farms are on the wrong side.

There's probably a 2x2 diagram here somewhere, maybe with openness and strength of internal relationships as the key variables.