Making real-time work is very complex at scale, and there's a reason there are only two real-time blog search indexes (Google Blog Search and Technorati). The volume of data presents multiple challenges: the data becomes nearly irrelevant shortly after it appears (often within days and certainly within weeks -- over 90% of all searches on Technorati are looking for something less than a month old); it’s much easier to spam* (Twitter is just beginning to experience this -- just wait...); it’s hard to balance recency and relevancy together; and lastly, it's expensive -- spinning large quantities of data so it's readily available to query is really expensive, and the entire live web is a really large place (Technorati only focuses on the blogosphere).