Is it possible to somehow make a query that will just return a random sample of historical tweets, rather than tweets based on a keyword? I’m trying to put together a data set akin to the “firehose” that was released to some developers and academics several years ago.