Even just converting science into a Wikipedia-like format would be useful for the sake of open access. Imagine if all citations in a paper were a hyperlink away, and the abstract would display if you hovered your mouse over the link.

YES! YES! YES! And this could be done pretty much automatically. Also, links in the reverse direction: "who cited this paper?" with abstracts in tooltips.

But there is much more that could be done in the hypothetical Science Wiki. For example, imagine that the reverse citations that disagree with the original paper would appear in a different color or with a different icon, so you could immediately check "who disagree with this paper?". That would already require some human work (unfortunately, with all the problems that follow, such as edit wars and editor corruption). Or imagine having a "Talk page" for each of these papers. Imagine people trying to write better third-party abstracts: more accessible, less buzzwords, adding some context from later research. Imagine people trying to write a simpler version of the more popular papers...

The science could be made more accessible and popular.