You should be able to select peers (though, maybe you should be given some as well?) and "experts" to follow. You would only see their activity when you want to, but you could also participate in the larger network. As a practical matter of implementation, I think allowing groups to run segregated networks (ie. their own "node") which can later be joined to each other makes sense for getting actual adoption and for reducing development, infrastructure, and support costs.
I think there are likely many problems with such a crude model, but that is where we find ourselves it seems. Waiting to think up something better might be prudent, but that might be difficult without trying something out.
In addition to collaborative document authoring (which I don't believe LinkedIn provides), data, simulations, and analysis code could be shared and built-on.
I don't think that "anoiting" a forum/web service would be a good strategy. We need to keep open the possibility of a new and better system to take over.
My feeling is that two fundamental requirements must be satisfied. 1. Anybody should be able to submit a paper. 2. Anybody must have an opportunity to become a recognized expert in a field. By anybody I mean that academic or career credentials must not come into play.
Obviously this begs for a rating system on the articles and on the "experts" too. Obviously recognized "experts" should have a bigger weight in the rating than random members. This makes it different from reddit and hacker news. But the idea is similar. Members should also be able to filter papers by rating and be notified only when a paper reaches a desired ranking. This is another difference from hacker news and reddit.
I would be very interested in participating in a project of this kind.