I much prefer this StackExchange topic over the Ubuntu topic:
http://unix.stackexchange.com/questions/88693/why-is-swappiness-set-to-60-by-default.
The Ubuntu topic misrepresents how swapping works. It doesn't work simply by amount of RAM consumed; the LRU algorithm makes it more nuanced than that.
Take note of this quote from the answer:
As with any performance parameter, adjusting vm.swappiness should be based on benchmark data comparable to real workloads, not just a gut feeling.
The correct way to tune swappiness is to leave it at the default value and setup server monitoring that tells you when pages are actually being moved in to and out of swap. Every time pages are moved from RAM to swap (swap out) is a time you effectively get extra memory for free. This is when unused code and data in you programs is being moved out of RAM, making more space for used code, data, and file caching. Only worry about tweaking swappiness when you start seeing a significant amount of pages moving from swap into RAM (swap in). At that point, tweaking may not help much unless you are just on the edge of not having enough RAM. You need to make sure any programs that manage their own memory aren't trying to use too much (like MySQL innodb cache) or you may just plain need more memory.
TL;DR: Leave swappiness alone. Monitor swap in and swap out. Don't worry about swap out. Worry about swap in.
Edit: improve description of what to do about swap in and add TL;DR.