I often find myself wishing for variations of popular sites and services since often I can find one that does what I want, but fits me better.
For example, when I was recently looking for some good bug-tracking software I used FogBugz as a trial and really liked it, but they're pricing was really ridiculous for my use case. They would charge $20 per user per month for whoever uses the site. That will work great for other folks who have a static team, but I work with a revolving pool of outsourcers. So for me, it would cost an extra $20 for a outsourcer of mine to perform a $10 task that month and use the bug tracking system. Great software, but their pricing model was a bad fit for me.
So, I started looking at other bug tracking software sites and found some that I'd never heard of, but that looked promising. I eventually settled on Unfuddle since it allowed me to have a revolving pool of users and not be penalized for it in the pricing. I tried Lighthouse and a few others, but there were slight variations in their functionality that didn't make them an ideal fit. Of course, those aren't exactly 'clones', but they duplicate significant portions of functionality and vary in small ways that can be significant for a user like me (i.e. pricing structure).
So, I'm all for clones - bring 'em on. It keeps the established guys on their toes and innovating and gives users slight variations to choose from.
It's basically slight genetic variation and natural selection in business.
P.S. The only time it really seems wrong to clone is when the site's text is copied too closely - or if the visual design is too similar.
P.S.S Also, if the site being cloned is super unique and pushing the realm of some new functionality, it seems bad form if clones don't at least acknowledge and give credit to the original site for their inspiration.