Who cares if it's easy to copy? Paul Buccheit says he built the first version of Gmail in an afternoon. Digg was famously built in fairly short order. Twitter, Flickr, Facebook-- all would be fairly trivial to clone. Less trivial to scale, obviously but... Feature-wise their V1 releases weren't that complex.
If you want something that you can do in your off hours AND something that isn't easy to copy, I'd say you might have an unreasonably low opinion of hackers around the world. Surely there's nothing you could build in your off hours that a few persistent hackers couldn't clone in a few months of full-time work, is there?
More on the "why should you care" front-- can you name any successful product/company that succeeded cloning someone else's idea? Maybe you can, but it's the exception, not the the rule. Now there ARE examples of companies jumping into the existing marketing and cleaning up by doing BETTER (iPhone, Flickr, etc) - but that's not cloning... That's innovating.
Additionally, you appear to be mostly considering high-profile web apps, which is not exactly the whole software universe. Head to the tucows.com, pick virtually any category and you will see it full of apps that are very similar. On the other end of spectrum, have a look at companies like NetScreen, which effectively grew their business by building cheaper Cisco devices. And this is not a singular example, it's more of a general MO rather.
All in all, I wrote about what worked for me and what worked exceptionally well. This does not imply that it's the only way.
The harder the idea is to copy, the more time you have for the app to mature. Any new idea needs real-world tweaking, e.g. the actual use of the app may end up being different from what you envisioned, some non-obvious features may be in a high-demand, unforeseen scalability or usability issues may surface, etc. The longer you have to address these issues without worrying about users leaving to the competitor the better.
Writing an actual copycat app is a different game altogether. I don't have an experience with it, but the common sense suggests that it's not worth doing unless there is a large, under-served market segment or all existing solutions really suck. Hoping to bite a piece out of established market with a couple of (flashy) features is not a very smart thing to do.
So I'd say 2+5+6 (+8) is the combination I had. Step (a) took about 2 years, (b) - under a year, (c) - 6 months. Your mileage may obviously vary :)