It's likely that the biggest startup ideas won't be on this list. But, it's still a great list.
I also think the idea of measuring and comparing complexity should be included. Even though this is arguably a pure math problem... A software startup will likely be involved because 1) any useful implementation will likely be software based 2) It'll be useful to most businesses in the world 3) most hackers know math.
http://news.ycombinator.com/best