On 6 April 2010 17:27, Raffi Krikorian <ra...@twitter.com> wrote: > our search and relevancy algorithms are constantly changing. we take in a > slew of signals like "engagement" or "conversation" around tweets, and use > that to pull it higher in search results. whether we will provide the exact > details of how that algorithm works, i'm not sure. its analogous to google > page rankings -- the general notion is well known, but the exact details are > constantly changing behind the scenes. > > we're still trying to figure out things internally regarding these "top > tweets" / "popular tweets" / "relevant tweets", but, as always, one could > just connect to the streaming API and get "true real time" tweets for > "earthquake". > > Maybe I could, but my 70 year old other couldn't. I also was talking from a users point of view, not a developers, and even for a develper, you might just want the data a little faster than you could knock up the working code to check the streaming API. At the moment, with 3-4 tweets from "popular" at the top, it's not too much of a problem, but my worry is that twitter intend to roll out the popularity algorithm to larger and larger chunks of search, thus losing the real time search aspect of which it should be proud.
-- To unsubscribe, reply using "remove me" as the subject.