Commentary

Why URL shortening as we know it is bad

There was a time when computers on the Internet were accessed via a numerical address. Early on, even before TCP/IP, the idea to map these sets of numbers to human-readable addresses seemed like a good one. The precursor of what is today the Domain Name System (DNS) was born. Fast forward to today and the Web is filled with millions of domain names. Familiar ones such as http://google.com are typed daily instead of http://74.125.45.100. As more and more domains were added and more and more server-side technologies emerged, URLs of webpage addresses became long and cumbersome. A need for a shortened URL was needed.

Early in the 2000s a new website, tinyurl.com, became the first notable site to provide such a service. Using TinyURL any website address could be given an alias that redirected to the actual address, but the alias often contained many fewer characters than the actual address. This was seemingly ideal for pasting to chat windows, sending in email (where email clients would sometimes not parse web addresses that contained long query strings with question marks and ampersands correctly), etc.

Then in 2006, Twitter appeared on the scene, and with its character limited message format of 140 characters, the need for a URL shortening method was all the more in demand. With Twitter’s meteoric rise to prominence, URL shorteners were given a large helping hand to mainstream visibility. In the time since, dozens have sprung up. But without a revenue stream to sustain them these services often folded just as quickly. And herein lies the problem. The Internet is built upon links of information that are relevant and accessible. For the most part these links are created by humans, and once created add to the organization and usefulness of the Internet as an information resource. Through Twitter and elsewhere URL shorteners are creating a vast catalog of links that are dependent on one website service to exist. If the service disappears, the alias no longer redirects, and the link is dead. Unlike most dead links it also doesn’t offer a hint as to what the original resource was, because the address has been obscured beyond recognition (although, at least one service, twi.bz, attempts to retain some hint of the original address). This contradicts the very idea of the Internet itself, whose beginnings began in de-centralization.

Little more than a year and a half ago mashable.com had an article that listed “90+ URL Shortening Services.” Browsing through this list it can be discovered that many are already defunct. With so many services to choose from who’s to know which may be solid, sustainable enterprises and which may be a mere pet project. Here today, gone tomorrow. Smalltime services may not have the resources, financial or otherwise to sustain themselves, yet their aliases persist on the Internet. Take cli.gs, who lost many aliases earlier this year due to being hacked. While this could happen to any website, it emphasizes the centralized vulnerability of these services. A similar problem can occur in cloud computing, as was illustrated by the demise of Ma.gnolia due to a hardware failure.

There are so many of these services because they do offer something valuable that is not possible with the current Internet protocols. Take tr.im, who announced last week that they would be shutting their doors at the end of the year. The public reaction that followed made them change their mind. This certainly illustrates the demand for these services, and is good that for the time being, tr.im aliases will remain alive, but it is a band aid on a problem that needs a grander solution than one popular website can provide.

Projects such as shortlink and short_url aim to tackle this problem by removing the third-party aspect of URL shortening and creating a specification for creating shortened URLs on the site that hosts a particular URL. These projects are not yet polished, but it is a start.

Taken as inspiration from this article about a service called Vanish, one interesting use for URL shortening that uses aliases may be in links that are intentionally made to “expire” after a certain period of time. Since a particular URL shortening service controls where an alias redirects to, they also can control for how long as well. However, if such a service was used more generally it would lead to the same issues of dead links (linkrot) as above.