[Handle-info] [blog] "Link Rot on Steroids: The Problems with URL Shorteners"

["John S. Erickson" <john.erickson@hp.com>]

Twitter's 140-char limit has caused URL shortening to hit the 
mainstream. As you might expect, the death of one of the more popular 
upstarts, tr.im, has caused much hand-wringing...

"Twitter Hammers Nail in the Coffin for Tr.im"
http://tinyurl.com/nqavw3 (readwriteweb.com)

"Link Rot on Steroids: The Problems with URL Shorteners"
http://tinyurl.com/m8my35 (edrants.com)

"On URL Shorteners"
http://tinyurl.com/dahygo (schacter.org)

Here is the summary (if you haven't figured it out already):

1. Twitter's 140-char limit caused predictable problems for users when 
they tried to refer bogusly-long system-generated (or just long...) URLs 
to their friends and colleagues

2. URL shortening services are easy to implement (in the first 
approximation...), so developers created their own even when long-lived 
services like tiny.url pre-dated the rise of Twitter

3. Most of these services have implausible financial models and are now 
going under, causing "link rot." To quote Tr.im, "There is no way for us 
to monetize URL shortening --- users won't pay for it..."

4. Most of these services have considered only the immediate set of 
requirements --- making URL shortening easy to use from Twitter and 
other social networking applications. Their objective is obviously not 
persistent naming; I'm sure most of their developers wouldn't have a 
clue what you were talking about if you brought up the topic.

Using these informal URL shortening services is the equivalent of using 
nicknames for people; for example, without context, do you know who 
"Turd Blossom" is? But what if instead of simply referring to "Turd 
Blossom," I also supplied some context <http://tinyurl.com/b29bb>?

One Solution: A possible work-around to this problem is for referring 
applications to (automatically upon embedding, after configuration) 
retrieve actual URLs and use these as the referred-to locations. S for 
example when you see...

http://tinyurl.com/b29bb

...what actually gets linked is...

http://en.wikipedia.org/wiki/List_of_nicknames_used_by_George_W._Bush

A problem: This creates the "indistinguishable from phishing" issue, 
especially in (html or other rich-formatted) emails where the link us 
apparently to one location but actually is to another.

A far better solution is for organizations to supply their own 
shortening services; better yet, for them to use a persistent identifier 
service in shortening mode.

NOTE: The solution here is not simply to "use DOI." The reason is, as we 
know the insistence by publishers on using "smart" DOIs means that DOIs 
can be just as bogusly long as URLs; they can still be long enough to 
break tweets and (my personal peeve) get broken in email systems.

Thank you for your indulgence in hearing this rant...

John


_______________________________________________
Handle-Info mailing list
Handle-Info@cnri.reston.va.us
http://www.handle.net/mailman/listinfo/handle-info