Slashdot Mirror
URL Shorteners Get Some Backup
URL shorteners are problematical, as everybody knows, but with the rise of Twitter and its ilk they seem to be a necessary part of the landscape. Some of the biggest questions around services such as bit.ly, TinyURL, and is.gd is what happens when they go out of business (as tr.im did last August). Now a group of such companies, organized under the auspices of the Internet Archive, has formed a non-profit entity to hold URL-shortening databases in escrow, with the intent of continuing to resolve a member company's links should it get out of the business. At announcement, the 301Works organization has 21 URL-shortener members, including the largest, bit.ly. Many others are not (yet) on board. The members have agreed to cede control of their domain names to 301Works.org should they exit the field, and to back up their URL mappings regularly to the organization.
You'll never get rid of "link rot", only mitigate it. Even archiving services have a non-zero chance of going under.
URL shortners only server for twitter posts and other place where you need to count characters, these links become pointless within days of a post (some think they become useless even earlier than that), so why bother preserving them after that? let alone when a provider goes bankrupt.
p.s I'm only posting this so i can get some karma to go troll apple ;)
IranAir Flight 655 never forget!
A governing body should not have to step up to preserve these databases.
Wtf? There is an unbelievably simple way to deal with this. Make the URL shorter, not the link it points to.
"http://tech.slashdot.org/story/09/11/14/184256/URL-Shorteners-Get-Some-Backup?art_pos=1" could become
"tech.slashdot.org/story/..."
in the text, but with keeping the long link it points to, so you can still see it when you hover the mouse over it. This is what StackOverflow does in the comments, and it works perfectly.
If your own URL shortener becomes popular (it won't), it will have to serve at least a million clicks per day. bit.ly is currently at around 4-5 per day, I think.
You can just perform simple redirects, without logging anything... But then you don't have anything even remotely interesting. The natural urge is to log every visit and let people view logs of their links (if you don't, users won't like your shortener). DB storage quickly piles up. A little bit of AdSense won't help you pay the servers, storage and bandwidth.
For someone trying to resolve one of the shortened urls, having a working mechanism present on the domain is a lot less important than having the database (for instance, say bit.ly shut down, Twitter could put in place a mechanism where users could press a button and some program would go through their spew and replace all the bit.ly references with something else, having the service running on bit.ly isn't real important for things like that).
Nerd rage is the funniest rage.
You realize that cleaning out the links that haven't been clicked in X amount of days is exactly what is *wrong* with those services, right?
Seriously? You've NEVER looked at someone's post from 2--5 years ago, saw a link about something happening at the time, and wanted to follow it to see more? History should still be preserved for others, even if you think it's only of passing interest to you. In fact, most things are more interesting much later as culture changes and new facts are revealed than they are at the time, when the details of most things could be guessed at from the immediate context that everyone is immersed in just then.
Great idea.
Now please take this url:
http://example.com/insert_hexadecimal_dump_blueray_disc_image_here
and run it through your shortening function.
Who needs bittorrent!
That stupid characters limit is self-imposed, they're the ones who created the problem. The easiest solution is to not count HTML characters, meaning you could have a 200 characters URL but be limited in length for the anchor text itself.
It also does unnecessary traffic and adds a delay. Doesn't seem like much, but let's say in five years everybody does that:
1. You click a link
2. Your browser does a DNS request for the shortener domain
3. The shortener website takes the short URL, fetches it in its database
4. Your browser receives a redirect to a new domain
5. Your browser does a DNS request for the actual domain
6. The actual website is sent to your browser
Steps 1-4 are added only for the benefit of an artificial, self-imposed limit from ONE website?
Fuck Twitter.
Some websites have user-friendly URLs, such as "www.apple.com/macmini/". You don't even need to click that link to know what it's about.
Other websites have dumb, half-friendly URLs, where they add the backend technology inside the URL, such as "http://www.logitech.com/index.cfm/mice_pointers/" (what's with the "index.cfm" in the URL?). If they fix that problem, all the links pointing to the current URL will break. If they ever change technology, it's also going to break the links from other websites.
And then we have the URLs designed by web monkeys, such as the link for Dell's new Adamo laptop: "www.dell.com/content/topics/topic.aspx/global/products/adamo/topics/en/us/adamo-onyx?c=us&cs=19&l=en&s=dhs". What the HELL is that thing? Even if we forget the parameters at the end, look at the path of that thing! I don't care how your crap is organized on the server, the URL should be much simpler than that!
And last, we have completely brain-dead URLs that seem to be created for computers only, without any chance of figuring out what kind of content is waiting for us on the other side of that link. Crap like "http://www.sonystyle.com/webapp/wcs/stores/servlet/CategoryDisplay?catalogId=10551&storeId=10151&langId=-1&categoryId=16154&SR=nav:electronics:computers:notebook_computers:shop_compare:ss". We're lucky to see "notebook_computers" in the parameters, sometimes it's just a database reference number.
But even with crap URLs like that, unless you have to spell it, write it down or read it on a (paper) page, such links can be hidden behind simple anchor text such as Sony Laptops.
Twitter is its own problem, they should be the ones to fix their own mess. Someone could start a service similar to Twitter but without counting HTML code as being part of the X characters limit, which seems to be what the fuss is all about.
Those of us who still participate in discussion forums via Email or Usenet have a ~70 character character limit on URLs. If it exceeds that amount it will break across multiple lines and no longer function. Although readers could just copy-n-paste all the piece together, the TinyURL provides a way to convenient way to avoid that.
"I disapprove of what you say, but I will defend to the death your right to say it." - historian Evelyn Beatrice Hall
They also give you a way to send people to places they would normally have the good sense not to.
Cool!