The problem with short URLs

I’ve started to use Twitter again, mostly as a way of doing quick linkdump posts that aren’t anything more than “Hey, I found this cool thing, go have a look”, but with occasional random chatty things as well. I was chuffed to bits when I randomly typed “I have achieved one of my goals in life. I have worked out how to play Dancing Queen in a minor key and have it sound creepy and disturbing.” and found that it was exactly 140 characters long. (Which doesn’t always happen.)

Twitter’s 140 character limit was originally a technical requirement imposed by the fact that the original idea of using SMS as a transport mechanism meant that a message, sending username included, could only be 160 characters long. 140 was, I suspect, derived by saying “this is the future, usernames are longer than 8 characters these days, and we need a colon; how about we say we reserve 20 characters for the username, colon, and future expansion”.

And these days a Twitter Pro with no post length restriction sounds plausible enough at a first glance, but was never likely to be anything other than a clever April Fool’s prank. There are other reasons for why Twitter made it big - one plausible reason is that by telling you when you’re being followed, but not when someone stops following you, Twitter avoids the problem earlier social networks had with the social stigma of being unfriended. But its enforced lapidary nature is surely a major part of what it means to tweet, and be twittered. So the 140 character limit is probably here to stay.

Still, there are problems with this, mainly when it comes to passing around URLs. There are plenty of URLs that are longer than 140 characters - while the SEO-friendly Amazon URL for the 30th anniversary edition of Richard Dawkins’ The Selfish Gene (to choose a random example from Amazon’s front page just now) is the fairly trim http://www.amazon.co.uk/Selfish-Gene-30th-Anniversary/dp/0199291152, the URL I get from Amazon’s front page is e.g. http://www.amazon.co.uk/gp/product/0199291152/ref=s9topdgwtr01?pfrdm=xxxxxxxxxxx&pfrds=center-8&pfrdr=xxxxxxxxxxxxxxxxxxxx&pfrdt=101&pfrdp= xxxxxxxxx&pfrd_i=xxxxxx (potentially personally-identifiable information occulted). Plenty of URLs are Enterprisy rather than friendly, and in any case if you want to talk about this cool website you found, chances are you want the URL to merely be a springboard for what you have to say, which means it should be as short as possible.

Twitter’s default solution is to turn URLs into shorter versions, via tinyurl.com. Tinyurl.com is a mature site, and lets you see where an otherwise anonymous link is taking you, but a number of URL-shortening sites have sprung up recently that are supposedly better. The problem is, these URL shortener sites are the wrong solution, with problems of their own: just as we were getting used to how URLs work, and learning not to click on suspicious-looking links in phishing emails, suddenly we’re back to having to blindly click on links, and in cases where we’re following links from a months- or years-old post, hope that the link still works. And that’s assuming that the URL shortener service is well-intentioned and benign, which is not always the case (Digg appear to be in the process of cleaning up their act since then, though). Hence why there’s an emerging standard for specifying canonical short URLs, so you can paste a huge long URL into Twitter, or a Twitter client, and it will automatically fetch what the remote site says is the canonical short URL, rather than generating a new one via a service that may not necessarily be reliable or trustworthy. (In Amazon’s case it might tell you that http://www.amazon.co.uk/gp/product/0199291152 is good enough.)

All of which is a very roundabout way of kvetching about the way Twitter’s website chooses to shorten URLs. I just posted the folllowing tweet about a fun Youtube video I found via Boing Boing:

How to make a baby: http://www.youtube.com/watch?v=luf6ZepNY6o (via http://www.boingboing.net/2009/04/20/revealed-where-babie.html)

That message is 131 characters long. Yet Twitter decided to shorten it to

How to make a baby: http://tinyurl.com/dgxhup (via http://tinyurl.com/d35ltm

Twitter didn’t need to do that; the message was already within the 140 character limit. But in doing so, it made two adjustments:

1. It shortened the YouTube URL from 42 characters to 25 characters, but threw away the important information that this was something on YouTube.

2. It decided that the ending ) was part of the URL, so if you follow that link to Boing Boing, you get a page not found error.

And of course by turning the URLs into something else, your browser no longer knew that you’d seen that site already.

Of course, the savage, bitter irony is that I’m going to post to Twitter that I just posted this blog entry, and because I haven’t patched Movable Type to generate authoritative short URLs on illuminated.co.uk I’m going to have to use a third-party URL shortener.