Hacker News Reader

Comment by matt1 | Hacker News Reader

1) Get ahold of a zone file listing all the registered .com domain names

2) Pick five common words

3) Scan the zone file, noting which registered domain names start and end with one of those common words.

4) Now parse those results and pick the words that are joined with them to form domain names (ie, the prefixes and suffixes)

5) Scan the zone file, noting how many domain names start or end with those prefixes/suffixes

6) Filter that list for the ones that occur > n times

7) Randomize the resulting list of prefixes and suffixes so it's not sorted by popularity

8) Add 100 per week

View on HN

Very cool, and a clever algorithm!

I've been playing with the site, and the sheer number of results can be a bit overwhelming. It might be neat to help address that by grouping together the suggested domains into categories of roots?

For example, there could be Places roots, filled with roots like hub, spot, -ocity, etc. There could be Frugal roots, with roots like cheap, cheapo, bargain, steal, etc. There could Newspaper roots, which add words like Observer, Times, Post, Chronicle, etc. to the users input. Other users could also create their own Roots, and share them... it'd be a great way to encourage community and create stickiness. If I have my 100 favorite Roots categories saved to your site, I'm much more likely to continue to use it to hunt for new domains.

When I search for a keyword, the results could be displayed in conceptual groupings (i.e. the different Roots categories)... that would help me quickly scan the results.

Just a thought... feel free to use or disregard. I'll definitely be using your site in the meantime!

This is a really excellent idea.

If you (or anyone reading this) is interested in categorizing the 2,000 modifiers like this suggests, please drop me a note: matt@leandomainsearch.com -- I'd be more than happy to compensate you for your effort.

I'd be glad to take a pass at it! Will drop you an email...

This is pretty epic; thanks for making it!

One feature request, though: while almost no self-respecting site would use most of the TLDs out there, any chance you could add in support for domain hacks and/or otherwise give an option to aggregate data for specific non-.com TLDs? I've noticed that a lot of the hipper two-letter TLDs are actually becoming prime real estate (namely .io, .fm, and .am), and they're young enough that a lot of really good names are still available with them.

Great work with your site matt!

When do you update the file containing all the registered .com domain names. How do you handle the registered domains after you create the file?

Reply on news.ycombinator.com