Show HN: Probabilistically Generating HN Post Titles

I took the top 10 daily posts for the last 3 years from Hacker News Daily (sorry about the few thousand wgets last night!), and tallied the occurence of words following other words (or groups of words). This generator probabilistically creates Hacker News titles by looking at the previous N words (N = lookback) and sampling from the distribution of words following them.

Code can be found here.
List of titles can be found here.

1. Python, the Web with Emscripten
2. Mark Cuban: What Business is Wall Street Journal
3. BankSimple: We have an effective vaccine for Malaria
4. Reasons to Open Source
5. It’s the founding CEO’s job to China'
6. Advanced Data Structures MIT
7. A list of lists
8. Our office is too slow for programming competitions
9. The Product is the Internet too much
10. More Americans see man who screwed an entire IPv4 /8 that it isn't using