Twitter has jumped the gun on its move to ubiquitous link-wrapping with its t.co service. They said they’d start on the 15th, with links longer than 20 characters, and they started on or before the 12th, with links longer shorter than 20 characters (that’s 19 above). I’m not happy about this change. In making it they’ve broken any sort of link parsing in any third party client that doesn’t auto-expand links, which means that any client that would, say, display images inline or filter out pesky 4sq.com links automatically can’t do that anymore, because everything is now t.co.
And there doesn’t really seem to be any point to it except as a grab to obsolete any other URL shortening service in favor of its own. Twitter doesn’t have an end game here.
They like to talk about how they’re fighting spam, and they’re announcing the change to link wrapping as “a means to protect users from malicious sites and scams.” Except they admit that they’re not actually resolving shortened URLs, they’re just rewrapping them. Once they figure out — IF they figure out — that a link is bad then they might send you to a warning page instead of just redirecting you, but it appears they’re not actually trying to do anything with those links.
And they’re not doing anything about spam either. That user is clearly violating the policy, but since there’s no effective automation of spam blocking, the only way the user gets flagged and disabled is if the recipients of that spam, like me, push the button. And there’s increasingly less incentive to do that.
Fix it, Twitter. If you’re going to force this on us, then do three things:
- Live up to your word. If you say you’re going to implement a change on a certain date, don’t do it early. And don’t wrap URLs that are already short enough. All you’re doing is filling up your namespace, with no other apparent gain to anybody, including yourselves.
- Resolve all URLs, either through the longurl.org API (I’m sure you could probably buy the service outright instead of licensing it) or by rolling your own equivalent to it. Display the fully-resolved URL on twitter.com and make it available in the API so any client can see it without doing extra work. Your current half-assed solution is useless.
- Show some evidence you’re actually attempting to deal with the spam problem by filtering URLs before they even get posted, as opposed to after the fact. A user like “holahanhzsp6” sending out the same URL in mention-spam should be easy to spot algorithmically. Hell, offer me a job and I’ll write the regexes for you.
