mlinksva, right, power law of bloggiverse definitely implies decentralization is not of particular consequence to frequency distribution of connectedness. I wonder if it's similar for the effects of the UI. Needs more thought (and data).

Re: FB serving as a proxy for what a frictionless decentralized social web (FDSW) would look like, I think that's a reasonable way to draw inferences, but would focus on dissimilarities that allow conservative conclusions. E.g. my impression is (a) 'friending' someone is something that the UI pushes you into doing. Also, (b) friending is forcedly bidirectional, hence A being interested in B (ergo sending a friend request), results in higher in-degree for A as well (unless B has significant impetus to deny the request). So I think it's safe to assume histogram of user in-degree in the FDSW would have lower tail; perhaps higher peak, assuming power law still applies.

So my guess would be that (b) above could account for a higher peak in the twitter curve. No idea if it completely accounts for that; if not, public vs 'private' might come into play as you suggest. Should read up on this, surely some researchers have tried to analyze the data. Differential analysis after UI changes would be especially helpful (but, is that possible for representative sample of the FB graph w/o cooperation from FB? I don't know).

The speculation on email as a universal social network hadn't occured to me. The first thought that comes to mind is that, for this to have happened, ISPs (in the 90s) or free webmail providers (portals, etc) would need to have provided one-click creation of MLs, along with a default "self" ML[0]. It also would have provided for a degree of user lock-in and customer attraction benefits because of network effects (assuming traffic within-ISP would be essentially free) to the ISPs.

Why did no-one try that? My understanding of the ISP business landscape post-walled-garden-dreams is admittedly quite limited, but my uninformed guess is that 1) few ISPs could get the capital for that 2) fewer would want to bet on another walled garden and 3) because of first-mover advantages, there would be poor incentive for other ISPs to follow. Just trying to be wildly wrong here, in the hope that someone with a better view of this part of history will be tempted into providing their own speculation :-) Any pointers to resources describing the business side of things at thduring that era would also be highly appreciated.

WRT portals, their problem might have been a bit like pump.io's: offering mostly blank slate expressivity with little UI guidance or manipulation (trying to include usability in "UI guidance" there). Take geocities (or self-hosted home pages): no two of the latter alike, the former more of a playground for individual experimentation than a platform for reaching out (i.e. introverted, not extroverted). It seems to have taken a few iterations, with blog^Wjournal hosting platforms, then myspace, others, for FB to come along, further narrow the context for expressivity, use the UI to manipulate into extrovertedness, make best (back then) use of network effects.

Perhaps it was also a cultural thing then; programmers designing software at first; business people designing software with an eye towards effective user manipulation techniques (i.e. best business practices) later. Some of them stumbling and tripping on to reach critical mass (survivorship bias will always ascribe exceptional skill to success, of course).

And this blank-slate aspect might have a lot to say about the email alternative history speculation (although we're probably more into alt-history science fiction territory at this point).

[0] Surprisingly, such a network would have worked quite well as a self-hosted setup for users on dialup with intermittent connectivity. You know, from a technical PoV.