I think you have to do it iteratively. Take all the serious ideas, and their votes as guidance, think about the user experience when you combine some of the ideas, implement something or form another proposal combining several things, and at some point come back to this community and say "ok, this is what we did - now what do you think?" Followed by another brainstorm... repeat.

That way the product will improve iteratively. It's impossible to make it excellent all in one go, and anyway without trying out changes, the best ideas cannot emerge yet.

For example the idea for combining email with SMS conversations. There's so many ways to do that. But what does "Clear conversation" translate to? Some people delete messages and emails after a while, but some people think deleting old emails is weird and archaic - and presumably same applies to SMS if they are merged, all should be searchable and archived. IMs are logged and archived forever by some apps and not others. What's best?

I think to a large extent, those sort of things and many others can't be answered without a certain amount of getting on with implementation or good prototypes, and repeatedly getting feedback and more refined ideas each time, and occasionally fresh new idea emerge that nobody could think of earlier. Or limitations become clearer, and ideas emerge for how to handle them.

In a more conventional devoper + developer's customer scenario (where customer is paying for development), there is a lot of iteration like this if it is done well.

Same applies here, imho.