del.icio.us word finder...

So, before I even get started, let me just preface this whole thing by saying, "Yes. I KNOW I am a geek." And frankly, I don't really care what you think about that. So if at any point while reading this, you find yourself thinking "Geez you are such a geek!" then I encourage you to just go instead read about how the Mariners are doing or the latest on what's happening on So You Think You Can Dance. Or at the very least, just go have a beer. Seriously, there was nothnig on TV...

So. This is something I have thought about ever since I found the site del.icio.us, which is, by the way a really cool website. I thought (and I know I wasnt alone) that this was such a cool and creative use of the DNS system and that the guys who thought to register that domainname were pretty damn smart. So, the quandary that perplexed me is this: How can one find a domain name that, when paired with the ".us" top-level-domain-name spell the most individual words? And as a corollary, ideally in the shortest number of characters. Obviously the domain "icio.us" is great one - it's owners, if they wished, could also spell "vicious", "auspicious", "malicious", and "suspicious" among 50-some others. That is a total of 60-some-odd single words from one 4-character domainname. Pretty cool thinking.

There must be others.

So of course, rather than just guess, I decided to use my favorite swiss army chainsaw tool, Perl and find out the exact numbers. So, I wrote a small program called dfw.pl ("Delicious Word Finder") that takes a few parameters:

dfw.pl suffix length [top_N]

By running the program, you will get back a "report" of the number of words that end in "suffix", containing "length" characters prior to the suffix that are used in the most dictionary words. As an optional parameter, it will show you the top "top_N" substrings in the list. So, if I were to run:

dfw.pl us 4 30

What I will get back will be the top 30 4-character final substrings that occur in words that end in "us". Doing so, I find that "icio" actually comes in at number 27 with a total of 60 possible words that end in "icious". But, interestingly enough, there are also 26 other 4-character strings that, when paired with "us" make up more words that "icious". In fact, the best choice of 4-character domains in the ".us" TLDN is actually "aceo". There are as many as 710 different words in common English usage that end in "aceous". Nice!

So, armed with my new little utility, I wrote a tiny driver program (called, appropriately enough, "dwf-driver.pl") that would step through each possible string length and look for the most number of words that can be spelled. So, for example, I was able to find that there are 473 differnt common words that end in "iferous" - that is, the 5-character domain "ifero.us" could be used to make something like "http://con.ifero.us". Perhaph it could be a site about trees and how important they are to the environment? At the same time, without registering another domain, you could also run the website "http://cobalt.ifero.us" - perhaps an advertising site for the cobalt industry. Nifty.

But so then I got to thinking, why stop with the .us TLDN? There are almost 250 other country-code-based TLDN's defined by IANA. Maybe some of those might be useful in spelling words. So with a little re-tooling, I hit up the IANA country list and grabbed their canonical list of countries and their 2-character country-codes. I did a little reformatting of that and created a CSV file out of it and then made a few changes to dwf-driver.pl to read that CSV file in and use it to drive through all the top few 4- through 10-character substrings ending in that TLDN. A little awk-action on the final report file and let me tell you - I need to find the registrar that controls the Libyan Arab Jamahiriya domain (".ly")... Over 2900 potential words there (with the best being the 4-character string "ical"). But Spain (".es") isn't all that far behind with over 900 potential words with "ness.es" as a domain name. Sweet!

So, this is all nifty, but it could use a little bit of improvement. For one thing, I am obviously (as usual) not the first to have thought of all this, because most of the 4-character and some of the best 5-character ".us" domains are already taken. So it might be useful to have either dwf.pl or the driver program actually perform the whois to determine if the domain is actually even still available. In addition, it would be kindof cool to hit up dictionary.com or something to also spit out definitions of each of the words found - then you could make decisions based, not just on the number of words available, but the *coolest* words available. For some definition of "cool".

Anyway, if you are interested, email me (or post a comment) and I will be happy to send you the perl code and let you search around yourself for a cool domain.