So I wrote a plugin to process the links. When Twitter Tools builds a digest, the plugin calls out to the remote site, follows redirects, retrieves the final URL and (if possible) extracts the page title. Then it replaces the cryptic-looking link with a human-readable link, transforming this:

Check out this site: http://bit.ly/9MhKVv

into this:

Check out this site: Flash: Those Who Ride the Lightning

If it can’t retrieve a title, it uses the final hostname. If it can’t connect at all, it leaves the link unchanged.

The download is here, and that’s where I’ll put future versions:
» Plugin: Twitter Tools – Nice Links.

Future

One thing I’d like to add at some point is cleaning up the title a bit. They can get really long, even without people trying to stuff keywords and descriptions in for SEO purposes. All it takes is a page title plus a site title, like this one. That’s a much more complicated problem, though, since there isn’t any sort of standard for which part of a title is the most important. I suppose I could just clip it to the first few words.

I’d also like to clean up duplicate text. Often the link title and tweet content are going to be the same, or at least overlap, especially if it’s generated by a sharing button or extension. That should be easier to check.

But why?

There are two main reasons to shorten* a link:

• There’s a technical limit, such as SMS message length or email line width.
• You expect people to manually enter the URL.

Right now, with Twitter messages limited to 140 characters and links forced to share that space with the rest of the post, URL shorteners are critical. But they’re working on a plan to accept longer URLs, and specifically shorten them for SMS messages. The full link will be available on the Twitter website, desktop clients, and other platforms that don’t have that hard and fast limit.

That will cut down on the demand for shorteners, but they’ll still be useful.

For one thing, there are other microblogging platforms out there like StatusNet.

For another, there’s email.

IIRC, the first URL shorteners launched because email programs often break up really long lines, including really long URLs. In plain-text messages, that leaves links not just unclickable, but inconvenient even to copy and paste, because you have to copy each line separately and paste them together. This will continue to be an issue as long as people continue to put visible URLs in email.

And then there’s the human factor. It might not be easy to remember http://is.gd/cGE8V, but it certainly takes a lot less time to write it on a scrap of paper than http://blogs.discovermagazine.com/badastronomy/2010/06/07/hard-to-port-eject-goose-eject/.

Which of those URLs would you rather type on your keyboard? Or worse, on your mobile phone?

*In this case, I mean making it really short and cryptic. There are plenty of reasons to keep links readable and sort of short.

1. It asks the wrong question
2. It was designed around limitations of cell phone text messaging

The Wrong Question

Twitter’s prompt is not something general like “What’s on your mind?” It’s “What are you doing?” That encourages people to post things like “I’m eating lunch” or “Just got into work,” or “Posting on Twitter.” Presumably what they mean is “What are you doing that you think people would find interesting?” but of course that’s too long a prompt from a usability standpoint.

The thing is, there’s no reason to broadcast the mundane to the world. Don’t tell me “I’m eating soup.” Tell me, “Just learned that gazpacho soup is best served cold. I wonder if they eat it in space?”

Unfortunately, that means the signal-to-noise ratio can get pretty bad at times.

Outgrowing its Limitations

Twitter posts are limited to 140 characters of plain text so that the your name and comments can fit in a standard SMS message. Now, this is great if you use Twitter via text messages on your mobile phone. It’s not so great if you use Twitter on the web, or through a smartphone app like Twitterific on iPhone or Twidroid on Android, or through any of the zillions of desktop apps.

I don’t have a problem with the 140-character limit itself (it can actually be liberating in a way), though it would be nice to have some formatting options beyond all-caps and *asterisk bolding*.

The real problem is that links have to share that limit. URL-shortening services have exploded lately as people try to squeeze links into the tiniest space possible to save room for their precious text. Even if you use something as short as is.gd, just including one link means you’re down to 122 characters.

Plus URL shorteners come with a host of problems, in particular the fact that they hide the destination. That’s no big deal if the target matches the description, or if it’s a harmless prank like a Rick Roll, but it’s all too easy to disguise something malicious.

Seriously, if you got an email that said something like this:

Look at this! http://example.com/asdjh

Would you click on that link? Even if it appeared to be from someone you know? That’s just asking to get your computer infected by a virus, trojan horse or other piece of malware. Or to see something you wish you could unsee.

Better Link Sharing: Facebook

I hesitate to bring up Facebook as a good example of anything, and I know the current layout is largely reviled by its users, but they really got posting links right.

When you want to post a link to your Facebook profile, you paste in the full URL. Facebook reads the page and extracts the title, a short summary, and possible thumbnail images. Then you have the normal amount of space to write your comment.

The one thing I don’t like about it is that it opens the link in a frame with a Facebook “toolbar” at the top, but aside from that, it’s much cleaner and more informative. I mean, compare this:

to this:

In the Facebook example, it pulled in enough information that I didn’t even need to add a comment. The link itself consists of meaningful words, there’s an indication of where it goes (googleblog.blogspot.com), an excerpt…and then of course there’s the image.

In the Twitter example, I made an effort to come up with a description (many people don’t bother), and then the link itself is just gibberish.

I’d like to see at least some of this capability in Twitter.

Maybe make links a separate field, so that the message itself doesn’t have to share those precious 140 characters, and have some sort of fallback for SMS users and older applications.

Or maybe just offer enhanced display on the Twitter website and on newer apps. Automatically follow the links (up to a certain number of redirects), extract a title (if there is one) and display something a little more legible than http://is.gd/rOoi

Come to think of it, that second option could be implemented in any Twitter client, or in a Greasemonkey script, without making any changes to Twitter itself. Hmm…

My first thought was that this was just like the CueCat, which was a bar code scanner that you could plug into your computer’s USB port, then scan bar codes in magazines, or on cans of soda, or whatever, and it would tell your computer to bring up relevant information. It was marketed in the late 1990s, during the tech boom… and it was a total flop. No one wanted them. The company went under and had millions of the little scanners sitting around unsold.

But now there are multiple schemes in use for object hyperlinking. In addition to graphical codes, there are RFID tags, GPS coordinates, and short text codes that you can easily type into an SMS message or a web portal.

So why is this sort of thing working now, 10 years later? Is it a societal change? Was the CueCat ahead of its time?

I think there are two reasons:

• CueCat was a single-purpose device. All the applications listed involve smartphones or other multi-purpose handheld devices. No one wanted a device that would only scan bar codes, but a phone/camera/browser/MP3 Player/bicycle that also scans bar codes? Sure, why not?
• CueCat was tied to the desktop. Sure, you could plug it into a laptop computer, but you’d still have to take the object over to your computer to scan the bar code. Unless you’re a lousy typist, swiping the CueCat across your can of Coke isn’t that much easier than typing in www.coke.com. As a home user, you’re not likely to be scanning a dozen objects in a row (unless you’re cataloging all of your books for LibraryThing).

All the applications listed on that page are mobile. A tagging scheme does give you an advantage when you’re out walking down the street and see something interesting. It’s much easier to punch in a short number than to try to type a URL on most phones, easier still to point your camera at a graphic, and dead simple to pick up an RFID tag or pull in GPS coordinates.

Another surprise: visits from commentary on last night’s Clinton/Obama debate. The comment thread includes a link to my post on JMS’s Londo/G’Kar campaign signs. Found while skimming the comments for links: T-shirts for the Capricorn ticket, Roslin/Airlock.

And then there’s the surge in searches for the Black Flash, no doubt inspired by people reading today’s Something Positive strip. In addition to landing directly on the profile, people are coming in from the Wikipedia article, and finding Flash Foreshadowing via an image search.

It’s a fairly long article about a web browser, and it mentions a few other web browsers including Firefox, Opera and IE. It also mentions websites Amazon.com and del.icio.us. But the only links in the article are to stock quotes and an earlier article.

I understand that it’s Business Week, and I’m not saying they should have linked to every website that was even tangentially mentioned—but you’d think they could have at least linked to the browser company they just profiled! I had to get that link from Slashdot! (Unfortunately, so did everyone else, so I won’t be able to look at the page until tomorrow.)

Edit: Compare the BusinessWeek article to Wired’s take from last month. Even taking into account that they’re written for different audiences, BusinessWeek still looks like a print article that’s been thrown up on the web.