A person going by “Bilaw.al Hameed” sounded the alarm over the possibility that the href of an anchor tag could be changed by clicking on it. He felt this opened up a security hole and the feature should be disabled.
This is not a new trick. This is how Google’s search results page works.
Inspect the results for a typical search; you’ll see that they look like this:
<a
href="http://en.wikipedia.org/wiki/Test_cricket"
class="l"
onmousedown="return rwt(this,'','','','2','AFQjCNGPXOKClui7vHgzV25lOsr4nAq50g','','0CDgQFjAB','','',event)">
<em>Test</em> cricket - Wikipedia, the free encyclopedia
</a>
The rwt
there probably means rewrite, as in, rewrite the URL. Hold the mouse down on the element and watch it change to:
<a
href="/url?sa=t&rct=j&q=&esrc=s&source=web&cd=2&ved=0CDgQFjAB&url=http%3A%2F%2Fen.wikipedia.org%2Fwiki%2FTest_cricket&ei=BBlGUZXbLdLSqAHKkoDQBQ&usg=AFQjCNGPXOKClui7vHgzV25lOsr4nAq50g"
class="l"
onmousedown="return rwt(this,'','','','2','AFQjCNGPXOKClui7vHgzV25lOsr4nAq50g','','0CDgQFjAB','','',event)">
<em>Test</em> cricket - Wikipedia, the free encyclopedia
</a>
And that’s why, when you hover, the link-preview appears to show you that you’re going to en.wikipedia.org
, but if you click you are actually going to a Google page which redirects you to what you wanted. They do this to track the effectiveness of their search results (with the side effect of learning what you clicked on).
If an attacker can inject JavaScript into your page you are basically doomed anyway. I don’t think there’s a reason to disable this feature, which also has some common uses like the above.
A version of this was posted to Reddit’s /r/netsec.