Note that with the Hacker News dataset now on BigQuery, downloading the dataset is unnecessary and the analysis in the post can be done in seconds: https://news.ycombinator.com/item?id=10440502 (although, parsing the domains using that particular dataset is harder since it requires a regex to extract the domain from a URL)

As a side note, I'm sorta annoyed that a data post with so few data and visualizations, but with a linkbait title and selfpromotional startup copy, was far more successful than any of my data posts. Did I miss the part where the OP explained how he used math / how he ranked the Top 50 startups?