Starting with the assumption that whatever startup idea I can have it's already been done in some form somewhere. How would you go about digging out those attempts ? (I mean beside googling which more often than not will yield nothing relevant)
The most useful refactoring of this is, "have non-sales conversations with actual intended customers, learn about how they see this problem including how they're currently dealing with it, which may or may not include existing tools."
At which point, we're talking about basic customer development interviews, which are way more useful than "google up a list of possible alternatives."
If there is a company already solving the same problem and you can't find it within 1h of internet research then they're doing bad marketing and you have a good chance. The potential customers won't find them either.
In my case, Google did not find a competitor that ranks within 30k on Alexa, has the keywords on their homepage, and has $21m in funding. I found them after weeks of research. Don't remember the exact blog/comment/thread, but it wasn't within an hour of beginning my research. :)
Entrepreneurship starts with creating a business plan. Part of that business plan is identifying your target market, your competition, and defining how you are unique. It's not so much that something has been done before but rather what sets you apart. If your concern is legal, then patents (google scholar) and trademarks are what you should be looking at. Copywrite is a bit more difficult but after performing your market research you should have an idea of what's out there. In other words, who is your target market and what are they using now that your startup is aiming to satisfy?
Unless you're copying someone else's code, text or graphics, you're not infringing their copyright. Copyright would be the "easy one" since it's near impossible to accidentally infringe, unlike trademarks or patents.
