The perception of scrolling vs. paginating has definitely changed. Users today are not opposed to scrolling if they believe that there is value to be had in doing so. At the same time, users sense of being overwhelmed with data has changed. The most successful interfaces that challenge users to wade through large data sets are provide some or both of the following:
1) As Todd says: An expression of quantity: a notion of how many results.
2) Mechanisms to understand and navigate dimensions within the total quantity. A good example of this is faceted navigation/filter sets. (See Getty Images result set left rail as an example).
The perception of volume has two sides:
If the data response is shallow, it can cheapen the value perception. “These guys don’t have anything”. Conversely, creating too many clicks will weaken the user’s motivation to move forward. Narrowing the set and localizing to relevant results usually changes the users perception on volume. If they believe you will give them the result they’re looking for, the tolerance to paginate and scroll is vastly increased.
The determining factors for pagination vs. full query set scrolling are context, end user goals and technical feasibility.
A note on pagination vs. single page scrolling: This interaction is typically a decision made due to technical limitations/feasibility. Depending on the data set, queries and data - retrieval can be seen as a performance risk in complex systems with massive data sets. Breaking up the query into well defined and finite RPC's basically breaking the data into chunks or in this case, pages, is a good way to mitigate that risk. The length of time users wait for data can make or the user acceptance and the brand and is more of a perception risk than the quantity of response data.
What we're seeing with AJAX and "lazy load" used by Twitter and Facebook among other sites, is the tendency to load a page or two of scrolling data initially, then "fetch" another set as the user hits the bottom of the screen.
In the case of Twitter, the audience can assume that the data set is too massive for pagination as a volume indicator to make any sense. And as the context-culture of Twitter is more topical and present tense, there's little value in archival functionality.
Facebook Timeline on the other-hand is sensitive to Life Experience context to their data, and have implemented a nice Annual/Monthly rollup archive on the right rail of what can be a seemingly endless set of life streaming posts.
The shorter answer is… it depends.