Style TODOs:

Add hyphenation entries for all unknown words.
Clean and consistent tables.
Syntax highlighting of code (one option: pygments can output to tex).
Consistent example formatting (>>> style, str printer, note at beginning that from sympy import * is assumed).
Avoid too much itemize.
"Other projects that use SymPy" section should use a table.
Add Unicode pretty printing example (see discussion on #89).
Code examples use curly quotes (’) instead of regular quotes (').
Should we use listings for the code examples?

Content TODOs:

De-duplicate content in assumptions architecture section and assumptions features section.
Mention all major algorithms that are implemented, even if they aren't discussed in depth.
Write the Gruntz algorithm section as actual text.
Clean up "basic operations".
Move "sets" details in to citations/supplement.
Move solvers details into citations/supplement.
Clean up matrices section.
Move series stuff to supplement.
Move logic stuff to supplement.
Move vector and physics stuff to supplement.
Move SymPy Gamma/Live stuff to supplement.
Either flesh out CAS comparison section with other CAS's or move to supplement.
Address all inline TODOs in the .tex files.
Make sure features table isn't missing anything.
Rearrange basic usage before the architecture.

Make sure everything that needs to be cited is cited.
Make sure the citations we have are good ones. We should avoid software citations and "URL citations".
"Other projects that use SymPy" needs real citations (not URLs)
Fix citation of Mathematica.

Abstract (note: write this last)
Introduction
Polynomials module architecture (if we can't get anyone to write it, add a blurb and cite Mateusz's thesis)
Something on printers
Conclusion

[ ] Add keywords

[ ] Switch to author-year citations

[ ] Merge or decide how to refer to supplement