You're sort of moving the target here. I care very little about scoring Internet points, but I care very much about HNers not making decisions which will damag

You're sort of moving the target here. I care very little about scoring Internet points, but I care very much about HNers not making decisions which will damage their business, so let's reiterate: it is categorically not the case that Google does not have a notion of domains. Page A and Page B, given that they are on the same domain, can easily influence each other's rankings even without a link passed between A and B, by separately garnering trust for the domain. This informs several SEO strategies which will work such as e.g. publishing OSS on your domain to get link juice and other indicia of trust which you then use to rank things that real people (and your bottom line) actually care about. As to whether blog.example.com and www.example.com share trust metrics, sentiment among SEOs are mixed. Consensus estimate among SEOs who I'd trust with my business would be close to "blog.example.com and www.example.com are treated as at half-arms lengths for the purposes of accumulating link juice and virtually identically for the purpose of awarding link juice." This suggests that you want as few subdomains as reasonably possible for publicly visible content which will either attract links or which you want to rank. For similar reasons, I'd suggest most startups would require a darn good reason to have more than one toplevel domain. Google has said not to worry about this issue, but that a very common answer from Google when probed about any specifics of their algorithms. Google will also say things like "Quality content is 99% of SEO" which if it is true is true in the sense that Santa Claus is true -- perhaps a good message to tell children but I would not base my ability to feed my employees' family on the likelihood of finding a jolly fat guy if I went to the North Pole to check.