Garrick Van Buren posted a recent post about URL shorteners where he suggests that “URL Shorteners Are So Last Year” where he runs through a variety of examples to make his case. To me, this generated enough yeah but responses that I decided to respond here.

For those not familiar with URL shorteners, they are services that allow you to take very long URLs and convert them into small versions

For example, here is a URL from a blog post I wrote in May 2009:

http://www.thedeets.com/2009/05/23/how-the-government-saved-
backpagecom/

That URL is 72 characters long, which isn’t too bad, but it isn’t exactly short. If I pasted it into Twitter, where there is a 140 character limit, I’d have less than half my allotted characters to comment on what I’m linking to.

I took that URL and shortened it at TinyURL.com into this:

http://tinyurl.com/y7284jm

That turned 72 characters into 26, so cut the size of the link in half. Clicking that link takes people to the exact same location. There is a drawback to this, as Garrick points out. It’s not as clear what people are going to find. The first link includes quite a bit of information in the URL, such as the destination site (thedeets.com); the year, month, and day the post was written (May 23, 2009), and the post’s title. That is all there due to settings I chose when setting up this blog. I could have created shorter, yet less readable, URLs on TheDeets.com which wouldn’t be all that much more user-friendly than TinyURL.com’s other than mentioning the destination domain, but didn’t do that for a variety of reasons not worth getting into right now.

If I took the same URL to another shortening site, Bit.ly, I can get it down to 20 characters with this:

http://bit.ly/dDfAKY

This is about as short as a URL can get.

Now, to Garrick’s points.

1. Shortened URLs can be used for spam. Yep, just like any other type of link. Sure, the chances of clicking to somewhere you weren’t expecting go up a bit with shortened URLs since you may be able to catch yourself before you click through to a spoofed PayPal site or some other bad neighborhood on the web if it’s in the URL you’re about to click on. However, how you get to a site isn’t as important as what you do once you get there, so I don’t think that’s a particularly big deal. Also, don’t click on links from people you don’t know (or suspicious looking links from people you do know) regardless of whether the links are shortened or not.

Top of the link URL Shortening services also test links that are shortened with their services to try to protect their users from ending up at spam sites. Clearly, the future of their services rely upon people trusting their service’s links, so they do what they can to proactively solve problems (not easy).

2. Shortened Links Expire. This is not true. URL shortening services would have a hard time existing if people clicked on a link a year from now only to end up at a different site than that shortened URL pointed to a year ago. Bit.ly explains this in their FAQ. TinyURL has provisions on what they’d do if they ever shut down (they have a plan on how to keep every URL that’s been shortened to date living forever).

Shortened links will expire if a URL shortening service doesn’t include provisions for what to do if they decide to get out of that business. For example, Garrick’s started and later stopped a URL shortening service using the domain culld.us, which he used to shorten a link here:

Garrick Using Culld.us

Since the service shut town, that culld.us link has pointed to this:

Screen shot 2010-04-19 at 9.04.20 PM

This makes me glad that I never shortened any links using that service. It would suck to hear from a client who decided to click on a shortened link in an old email telling me that the link no longer worked due to something outside of my control. It looks like Google can see 193 such links that were shortened and then published to the web somewhere that Google could see.

We also can no longer tell what David Brauer was talking about when he used a culld.us link in this Tweet:

Screen shot 2010-04-19 at 9.12.55 PM

Critical parts of Twitter conversations by Brauer, Tom Elko, Mike Kelliher, Jason DeRusha, Paul Jahn, Michael Janssen and Mookie (Chuckumentary’s cat) were destroyed when culld.us went offline. Here are some examples of pages within Twitter that Google can see that include now-dead culld.us shortened links:

Culld.us Link Rot

This is known as “link rot”. It sucks when links go bad, so be sure to shorten your links somewhere that has provisioned for this.

Case insensitivity is one of the keys to keeping URLs short without having to recycle addresses. With 6 characters after the / where Bit.ly uses a combination of numbers together with both lower and UPPER case letters (62 options per character), they have generate over 44 billion URLs before they’ll need to add another character.

3. Usability. Here is Garrick’s take:

At best – they make less usable URLs – because both the URLs shorteners domain name and the random string are meaningless (not to mention hard to remember) to people.

First off, if they were less usable, why would people use them? People don’t go out of their way to make life more difficult for themselves or for others. While it would be nice if we could link to everywhere on the web from a nice short URL, there are some good reasons why that’s not possible. The trade-off of linking to content using URLs that are more email and Twitter friendly is clearly favoring the shortened URL side of this equation. We can see this in the combined growth of just two of these services (bit.ly and tinyurl.com) who’ve had a combined 3X increase in traffic in the past year:

Bit.ly vs TinyURL.com Traffic

With bit.ly capturing a ton of that due to their superior service.

While URLs provide guidance, I don’t think many people are remembering long URLs. Sure, they may be able to remember that a company’s blog is at domainname.com/blog/ but that wouldn’t need to be shortened. Few people will remember 70+ character URLs.

Guidance can help, but true guidance comes from what’s around a URL as much as what the URL itself has to say about itself. If a friend sends me a link that says, “here are some pictures from last weekend” I know what I need to know about the link whether it’s shortened or not.

4. It’s webmaster’s faults for making long URLs in the first place. As Garrick puts it:

“URL shorteners are cheap hacks apologizing for poor content-management-systems”

By that, he’s saying that URL shortening services only exist because web developers are slackers who are too lazy to create human friendly (describe what they’re linking to) short URLs. He picks on Google Maps, but doesn’t provide an example of how Google (who has a few smart coders) is supposed to create human friendly (and short [possibly even memorable]) links to their maps. Google Maps’ links, as they exist today, can include up to dozens of variables (along with values). So once we get past maps.google.com, you’ll often see variable=value&variable=value combinations building out a long string with popular variables being:

Latitude
Longitude
Zoom Level
Map Type (map, satellite, hybrid)
Layers (bike, transit)
Driving directions (or other type of directions)
Starting point
Name of venue
Language
Character encoding

Is this really a case of bad URL architecture, or is there simply no way to get a URL that has to contain so much information down to a shorter format? I’m guessing the latter.

There are over 40 variables that can be used to customize a map view. Most people don’t interact with the variables directly, but every time you move or zoom the map, the URL needed to get someone else to that exact same view changes.

Once you get to the map location, zoom level, venue, etc. that you’d like to share, would you rather send your friend an email with a URL that goes on for multiple links and probably breaks, causing them to click on the link then email you back to say that the link didn’t work for them? Or would you rather shorten that down to something that easily fits on one line (and will even tell you if your friend clicked on it)?

5. Twitter built the need for URL shorteners and they’ll also kill it. Garrick explains that the demand for URL shortening services exploded when people became addicted to sending 140 character or less messages with links. There is no question that Twitter took URL shortening to a new level, but the original link shortening problem that I believe tinyurl.com was initially set up to address was making URLs easier to email. That problem still exists. When I send a long link to people today, I shorten it, because I don’t know if they’re going to be opening it using a desktop client, webmail program, mobile device, etc., and all may treat long URLs differently. With or without Twitter, there is a market for shortened URL services.

That being said, Garrick has a point that bit.ly and other services used by people to shorten URLs may take a dive if Twitter builds in their own service. He goes so far as suggesting that Twitter’s move into URL shortening is “a move which will promptly shorten the already tiny lifespan of other URL shortening services.”

While possible, that depends on a quite a few factors, such as:

a. Whether it’s available in Twitter user’s preferred Twitter clients.

b. Whether it has reporting that’s as powerful as what people find elsewhere.

c. Whether it’s as reliable as what people use today. Would you trust Twitter with your important URL shortening needs based on how reliable their basic service has been?

Sure, 3rd party URL shortening services will see some impact from Twitter getting into the game, but I see no reason to believe that this will mark the end of quality link shortening services like Bit.ly or Tinyurl.com.

Thanks, Garrick, for throwing some ideas out there for me to riff on. Good stuff.