1. A worthy Google search competitor. Preferably open, open-source, federated, respecting privacy.
2. A platform where scientists can discuss publications. It should be a "home" for every paper. Of course, also open, run by a nonprofit organization, or perhaps even federated.
3. A good open (xkcd-927-defying) standard for chat that everybody will use. (Why can we have this for email, but not for chat?)
I just posted what I'd like to see built, so I don't necessarily have a vision :)
But a few remarks: I think that getting something like this off the ground is more a social challenge than a technical one. The biggest question is how to get everybody to use such system, and make it into a hub. I imagine it could help if such a system cooperated with existing places which are often visited by scientists, like sci-hub, arxiv. I'm not sure if Google scholar would be a good match (because of their non-openness and privacy issues), but perhaps they'd like to help. Of course there are technical challenges. Like how to run this in a federated space. And UX challenges, such as how to handle moderation. And also basic UX requirements like a good formula editor and diagram editor should be considered. And there will be questions like how to incentivize scientists to comment on their own results, and to help other people (while not necessarily being paid for it).
Lots of requirements and questions :) It is a pity that I don't have time to work on this myself. So good luck if you want to try this :)
I think the reason its not so good is down to the personalization. Google knows I want to search for Django as a web framework, while DuckDuckGo assumes I want the film.
2. A platform where scientists can discuss publications. It should be a "home" for every paper. Of course, also open, run by a nonprofit organization, or perhaps even federated.
3. A good open (xkcd-927-defying) standard for chat that everybody will use. (Why can we have this for email, but not for chat?)