I've been following the progress in number theory proofs for the past few years, and email and forums have worked fine.
That some projects require a large amount of expertise that no single person is likely to have is true. But there are disadvantages to having more authors as well, e.g., "design by committee" immediately comes to mind. And I don't think a majority of current projects require expertise that no individual could have to an extent that would explain why collaboration is expected in science. Ultimately I think a mix of the two would be ideal for science in general, with a larger contribution from individuals than is seen at present.
Also: If my example is a shallow anecdote than your CERN example is as well. I don't have hard data on any of this and you didn't mention any either. I'd be happy to look at any hard data that exists, but I don't know if there is any.
https://www.independent.co.uk/news/science/that-figures-prof...