Twitter’s t.co Continues UX Failure of Link Shorteners
It’s been a few weeks since Twitter moved to its own link shortening service for tweets. Originally the shortener only kicked in for tweets over 18 characters, but Twitter recently moved to have it affect all URLs in tweets. Twitter’s argument was that this allows Twitter to reduce the number of spam and phishing URLs embedded in tweets. In Twitter’s own words (from the t.co site):
Twitter uses the t.co domain as part of a service to protect users from harmful activity, to provide value for the developer ecosystem, and as a quality signal for surfacing relevant, interesting Tweets.
Twitter’s Reasons for the t.co Shortener
Twitter’s explanation sounds reasonable but doesn’t bear itself our now that I’ve had some time to try it out. Let me explore…
As Protection from Spam/Phishing
I still get the same number of spammers on Twitter, I did not expect a link shortener to change that. Those spammers also use link shorteners, so whether or not the t.co service came into play it wouldn’t matter much — the link is still obfuscated. If the t.co service was doing its job, however, then those tweets would be caught or flagged. Even if it wasn’t a pro-active service (because we know people can change the destination of a link shortened by many services, rendering it malicious from an initial innocuous configuration), I would expect it to do its job when I follow a link. It doesn’t. Just this morning I received a spam tweet, and for the scope of this post opted to click the link. The t.co address showed up in browser, it spent time processing, and then sent me to the phishing site. Twitter’s first claim is false.
A couple weeks ago I followed a tweeted link from a local business that fed through the t.co service. The t.co service told me that the link was to a spam or malware site. It was not. The link was to an online petition for a local issue. While I was motivated to just grab the original link from the tweet, there’s no way to tell how may others may have had the same experience and were unable to weigh in on the petition. The risk here is that the t.co can also produce false positives, damaging anyone’s reliance on Twitter as a link dissemination tool.
As Value for the Developer Ecosystem
Let’s be clear here — as a user I don’t care how much easier it is for developers. I don’t let my web team just throw a bunch of fields on a web form without regard to the end user, no matter how much more quickly they can do it. But Twitter has a model that is less about the end user and more about driving organizations to rely on Twitter through its API and reporting features. The t.co shortener provides a boon to Twitter because it makes it easier for web masters to see how much traffic came to their site from a Twitter-shortened link.
Sometimes that t.co link isn’t from Twitter (it’s been shared elsewhere such as a blog, through originally from a tweet). Sometimes the same web page address has more than one t.co address. Sometimes that t.co link is to a bit.ly (or other shortener) link, which are often created for use on Twitter anyway.
As a Quality Signal for Relevant or Interesting Tweets
Given that I have no confidence in Twitter’s ability to filter malware, spam and phishing sites, I certainly cannot believe that quality is an appropriate word. Just seeing the t.co address in a tweet doesn’t tell me that it is relevant or interesting. Twitter may decide a link or tweet is relevant or interesting simply by measuring how many clicks it gets. In the absence of a clear explanation, those two metrics are also suspect.
Other Factors
Twitter Clients
I use TweetDeck on my computers and Seesmic on my phone. I have used Hootsuite and sometimes I use Twitterfall for a Twitter wall at events. They all display the t.co address instead of the full address underneath. These apps don’t update at the same pace as the Twitter web site and not all users will allow frequent updates (whether by corporate IT policies or lack of interest) to their Twitter clients for when they do support expanding the t.co addresses.
This means I regularly see a t.co address. This wouldn’t be an issue except I rely on the URL to know what will happen when I click a link: youtu.be means my Twitter client will play a YouTube video, twitpic.com means my client will show a picture, and so on. That link scent is now gone and I click fewer links as an end user because I don’t know what I will get.
Twitter-Provided Tweet Streams
I use the Twitter-provided JavaScript code to embed a Twitter feed on my personal site. This does not expand the t.co URLs to show the full address. I push the Twitter-provided RSS feed of my tweets to my blog. This does not expand the t.co URLs to show the full address. Tweets pushed to Facebook or other services come with the t.co, again hiding the full address and link scent.
Extra Bandwidth Burden
At peak times I have found a link that goes through t.co takes longer to redirect me to my destination address. Often that destination address is being hit by only a few users, maybe a few thousand. The destination site can typically handle the traffic. The t.co service is taking the brunt of all the traffic from all the users on Twitter. This increases the bandwidth used across the web (and on my phone) and results in a longer click-to-destination time.
Copy/Paste Hassle
This may sound like a minor issue, but I regularly copy an entire tweet or just the URL from a tweet, often wanting to share via email, in my blog, or elsewhere. When I do that I typically get the t.co link, and I think it’s obvious by now that I do not want that. If this affects me, then others who may do the same but not know how to tease the expanded URL out of a tweet could end up pushing traffic to my site from a blog that reports itself as a t.co referrer. In my reporting I will now be unable to distinguish traffic from a link and from a source that I may want to otherwise engage.
Blocking
I have read from a few sources that t.co is blocked in China. Given Twitter’s prominence in recent events such as the Arab Spring, London riots, and now even Occupy Wall Street, creating a single point of failure with t.co means not only are the links blocked, but the expanded link may be blocked easily by any organization or government that wants to quell activity on Twitter.
My Former Reliance
I use a photo sharing service that pushes a link to the photo to Twitter. I used to take the RSS feed, along with the geolocation of the tweets, and pull the URL from the photo service to quickly hack up into a path to the thumbnail. I would then embed this modified RSS feed into a Google Map to show my activities and travels — most recently for a trip to Italy.
Because I was not using the Twitter API I could not be considered a developer, so I don’t fall into Twitter’s stated support for developers. As such, when the RSS feed from Twitter converted the URL from my photo sharing service my maps didn’t display images and my followers stopped clicking links to my photos.
I also regularly craft URLs in tweets to remove the query string nonsense and unnecessary “www” prefixes (among other bits). I also regularly craft a tweet with the intent to make the URL visible to the end user because the address often feeds into my point or joke, Instead I find I exclude the “http://” from the address to get my point across, but it also means the link is not clickable for many users depending on their Twitter client.
A common annoyance is that Twitter now encodes URLs that are already far shorter than the Twitter-encoded t.co address. My own tweets have seen URLs double in character count after Twitter applies its link shortener.
Why Did Twitter Do This?
Three key reasons that I can contrive:
- Twitter owes much of its success to developers building apps and integrating it into other services. Shortening URLs reduces the efforts an end user has to make in a third-party tool to stay under the 140 character limit.
- Twitter’s importance as a driver of web site traffic is reinforced when webmasters see the t.co links in their logs.
- All the t.co links track information, which puts Twitter in a position to monetize the data it captures from each shortening and each click.
For all the reasons I state above, Twitter isn’t really helping the end user (either content consumer or non-developer). Twitter’s goals here are more for its own gains. In the end, since it’s a free service they have the right to do that. I would certainly appreciate a more direct and honest explanation and a consistent implementation across its API and own services (RSS, JavaScript widget).
Sadly Twitter has continued to set the mark for other developers and, like its infinite scroll and other user annoyances, it will continue to enable developers to make poor decisions that are counter to a good user experience.
Related
From Twitter
- About Twitter’s Link Service (http://t.co)
- Twitter’s t.co URL wrapper is now on for all URLs 19 characters and greater
My Posts about Link Shorteners
- URL Shortener Spam Overrunning Blogger Stats, Feb. 4, 2011
- You Get What You Pay For, Dec. 17, 2010
- Libya’s Terror Plot: Link Rot (Linkpocalypse?), Oct. 11, 2010
- More News in the URL Shortener Market, Dec. 15, 2009
- List of URL Shorteners Grows Shortener, Oct. 5, 2009
Leave a Comment or Response