The Link Letdown: When URL Shorteners Fail
There's been some heated discussion lately about URL shorteners and whether they're useful tools or pure "evil." My initial reaction was that much of the debate was overreaction -- after all, you'd be hard-pressed to send links on Twitter without services like bit.ly to cut down their characters. This week, though, I've seen some new evidence that's made me rethink my position.
The Ecosystem Argument
The argument started with a blog posting by Joshua Schachter, creator of social bookmarking site Delicious. Schachter described URL shorteners as being generally bad for most of the online "ecosystem," claiming they weigh down the Web by adding "another layer of indirection" and allowing for spam-oriented links or worse malware links to be masked. The part of his post that particularly strikes a chord with me now, though, is his stance on the potential problems with reliability.
"A new and potentially unreliable middleman now sits between the link and its destination," Schachter says. "The long-term archivability of the hyperlink now depends on the health of a third party. The shortener may decide a link is a terms of service violation and delete it. If the shortener accidentally erases a database, forgets to renew its domain, or just disappears, the link will break."
Those scenarios may seem hypothetical, but I got a small taste of what could happen -- make that, two small tastes of what could happen -- in the past few days. I've been using tr.im for URL shortening on Twitter, and I've generally been pleased with their service. Tuesday, though, I opened my tr.im control panel only to find a "500 Internal Server Error," and nothing more. Sure enough, every single URL within the tr.im domain returned the same code and failed to redirect to its appropriate target.
The issue didn't seem to last terribly long, and I was willing to write it off as a fluke. Thursday, however, it happened again -- and, from my measurement, the second go-round lasted even longer than the first.
Investigating the Issues
I tried tweeting and e-mailing the folks at tr.im to get a better understanding of what was happening and what was being done to make sure it wouldn't happen again. After all, how can we rely on a URL redirection service when we don't know when it might stop redirecting? It's hard to estimate how many links from past hours, days, and months led people to error pages instead of appropriate destinations even during those limited times. And, to an average end-user, that sort of occurrence reflects poorly on the person who posted the link.
Unfortunately, tr.im didn't get back to me in time for this story. The company did, however, post the following two messages on its Twitter page:
"We have fully recovered from this morning's outage. We are getting a serious uptick in traffic since the release of Nambu"
"We assure you tr.im is reliable! An increase in traffic is no excuse. We are increasing our margin of spare capacity."
The same service suffered a smaller issue on Monday, when its statistics service stopped displaying click totals for some time. I was able to reach the team at tr.im then. They assured me that no data was lost and that the "technical issue had been resolved." They also indicated that tr.im had an uptime of 99.67 percent in the month of March, and that the majority of the downtime was a result of planned maintenance to better handle the growing traffic.
My point here is not to unnecessarily single out or pick on one company -- in the grand scheme of things, the outages thus far this week have admittedly been minimal. But, given the timeliness with the discussions happening in the online world, they've certainly given me pause. Links posted to Twitter, Facebook, or blogs remain available indefinitely. If a service were to become periodically or permanently unreliable, and you'd long relied on it for sending out your links, you'd effectively be screwed.
The implications, of course, reach even further. Even back before Twitter's recent boom, the gang over at Slashdot was growing worried about where URL shorteners could lead us. "What if a service starts sending a pop-up ad along with the redirect?" one poster asked. "Are [these] services ... taking the WWW towards a single point of failure?"
One thing's for sure: On the whole, we as a Web community are putting a lot into these services' hands. TinyURL, still the top shortening service by most counts, has nearly tripled in growth over the past year, according to traffic measurement estimations by Compete. Other popular services such as bit.ly, is.gd, and tr.im have seen less staggering but still quite significant increases of their own, gauging by recent measurements from Twitter tracker Tweetmeme. And the list of options only continues to grow.
No Easy Solution
Schachter suggests moving to a more transparent model in which shortening services provide logs of redirects so that users can "reclaim" their links if the service disappears. He also mentions the idea of encouraging more Web sites to provide their own built-in shortening services.
While a Twitter-integrated shortener would probably be the least reliable solution of all (let's be honest here, the site's not exactly known for its stability), the idea of content sites offering their own trimmed-down links isn't half-bad. USA Today has already taken that step, providing bite-sized links for stories via the usat.me domain. The concept isn't only user-friendly -- it's also good for the site, keeping links within its control and adding an extra serving of Google juice into its mix.
A poll of roughly 1600 readers of tech blog TechCrunch found 57 percent saying URL shorteners are in fact "evil," compared to 39 percent saying they weren't. (The rest had no opinion, yet for some reason felt the need to vote anyway.) I'm not sure I'd go as far as to stamp the Lex Luthor label on these companies -- the shortening services do, after all, offer a lot of value and functionality, and they've yet to completely fail us. I, for one, am certainly not about to jump ship. Still, as this week's events have reinforced to me, the system is anything but foolproof, and that's something well-worth keeping in mind every time we generate a handy new 10-character link.