The Search service has been evolving for some time at Twitter. The Search service is focused on serving relevant tweets to ad-hoc user-driven queries. This means that a user has provided some time of criteria â such as a keyword, a #hashtag, a personâs name, a URL, or a location â and the search service responds with what it âthinksâ with confidence best matches what the user was asking for. While Twitter users can potentially be minerals, ideas, companies, or things â in the context of Search, the intended end-user is a human. This is the search serviceâs focus, and the API for search is a programmatic interface bridging human queries between your application and Twitter.
All that preamble to say that the kind of tweets search chooses to index and respond with for a particular query evolves. Retweets are very valuable information constructs on Twitter, but their primary value is realized in their appearances in home, user, and list timelines and the fact of having simply existed at all, both singularly and in aggregate. But they are less valuable to serve to a user as part of a response to a query when the originating tweet or a tweet authored by a user more relevant to the searcher can be served instead.
User timelines continue to be the canonical source of tweets authored or retweeted by a specific user.
Any attempts at accurately counting or capturing all the instances of something, like retweets matching a specific term for instance, should be done using a Streaming API.
Thanks!
