[Permalink] code inactive other web
gnowt A URL shortening service that uses memorable words
With Twitter's move over to shortening every url in every tweet, I quickly got sick of seeing URLs that look like
shrt.lnk/sdl83JU, i.e. a domain name followed by an unfriendly muddle of uppercase letters, lowercase letters and numbers.
Gnowt is a proof-of-concept URL shortening service that stores URLs and provides a short link like:
I felt that a URL shortening service that used words to identify the link would be more useful in a number of ways:
- I can remember a short sequence of words much better than a sequence of characters
- It's probably just as quick to type a few familiar words as it is to type an unfamiliar code
- You can tell another human being in person or over the phone how to get to your URL much easier if it's made of words
- "Go to shrt dot lnk slash ess dee ell eight three upper-jay upper-yoo"
- "Go to gnowt dot offend dot me dot uk slash out now here" This will be better if I buy a nice short domain name of course
- If the service knows it's looking for words, it can be flexible over input
What I mean by the last point is that, if you've entered a URL into gnowt and it says your new URL is
http://gnowt.offend.me.uk/out/now/here, you can actually also access that as:
http://gnowt.offend.me.uk/out now here
Whatever you find more aesthetically pleasing I guess.
One concern I had was that my shortened URLs might get considerably longer than other services once gnowt is storing a good number of URLs and thus losing the benefit of being memorable.
To offset the possibility, I made sure that all the words in the list were well-known and unambiguous words. I did this by:
- Taking a list of the top 6000 most frequent words
- Removing all but nouns, adjectives, and adverbs
- Removing plurals
- Removing profanities (I'd be upset (ok, maybe amused) if my URL ended up being gnowt.com/cock/balls/poo)
- Removing any words that are spelled differently in different forms of english (I'm thinking of colour vs. color here)
- Sorting the remaining words in order of size so that we use up shorter words first
- Removing any words longer than 6 letters
The word list contains 1693 words. This means that there can be:
- 1,693 one-word URLs
- 2,866,249 two-word URLs
- 4,852,559,557 three-word URLs
- 8,215,383,330,001 four-word URLs
By the time it gets up to four-word URLs, that's a total of 8,220,238,757,500 URLs.
Let me say that again: 8 trillion, 220 billion, 238 million, 757 thousand and 500 hundred URLs.
(When you get to 5, it's nearly 14 quadrillion)
I reckon that's enough to be getting on with for now. If I'm wrong, I'm wrong :)