The semantics are crucial here, because solving a problem almost implies that it is something a customer 'understands' to be a problem and not just satisfed with the status quo - that is true for almost everything - give me a better house, car, laptop, email, search etc. etc. anyday and if its really good enough I will switch.
The biggest startups were not found by scratching an 'itch' but through founders idiosyncratic views of the status quo and how they belived (without an MBA, without a customer survey and without figuring out if it was really an 'itch') on what was a better way, and for a long time without any external validation.
Selling books online - how does that even make sense, the internet was supposed to get rid of books but Amazon found a way. The iPod - if no one even knew something like this can be built, how can it possibly be a part of their 'itch'. Twitter - I bet there was no itch here too.