web site architecture, size - folders - other factors that matter?

What architecture is best for SEO purposes?
- Star - every page points to sitemap which points back to each page?
- Organizational chart - home page points to 5 other pages which are each chapter headings which point to 5 closely related pages. These pages point to the others in their cluster. Should they also point to the other chapter heads?
- Sequential - Page A points to B points to C etc
- Combination of start & chart? Start & sequential?

Linking architecture
If I have a 100 links into my site, is it better to have them all pointed to one page or to spread them across all the pages?

If I have three websites and only one is the one (say A) that matters to me, what the best way to organize them?
- Should be B & C both point to A? Presumably yes. By the way, my A site (a learning site) has Educational Teaching Games and my B site (also used to learn about affiliate programs and how to learn to do websites myself) has first grade learning games. (www.time4learning.com and www.todays-learners.com respectively)
- Would it be better for outside links to point to B & C or directly to A?
- Would it best for advertising (for people) links to go to A but for spider-oritented links to be aiming at B or C?
- I gather that the "bad" or "black" seo types do all their black stuff with sites that only point to their main site, this keeps their main site intact. Also, they don't have a fixed "main" site, they just move it around.

Folders - Is it better to have sections? Do folders for the pages help the search engines find the sections or clusters? In the case of www.time4learning.com , would a folder called 2nd grade help if I put all the pages related to 2nd grade (2nd grade math, 2nd grade learning games, 2nd grade software, 2nd grade reading, 2nd grade curriculum) in it?

Are the search engines able to keep track of clusters that overlap? Using the example above, I have clusters that organize by grade level and also by subject area (learning games, math, learning software, curriculum, reading etc).

BTW - my assumption is basically while its good to gather info and follow the trends, to accept that the real algorithms will evolve quickly and that I should assume that google's ability to implement which more or less match my ability to dream up what they should do so I build a little bit independently of what I learn. Of course, if I was willing to read more and immerse myself in the evolving details of whats hots, I might do better. Maybe I will......