I built a Twitter app today! It’s called Quadroopl! It’s a version of your Twitter feed that’s been cut down to size by removing all but one of each of your friends’ tweets, leaving only the most retweeted one. That means that, if you want to see if what your friends are doing is interesting or important, you can use Quadroopl to cut through the spam and get a quick update on the good stuff. You should try it. The code is right here on my github in case you’re that kind of dude/lady.
I was in Ohio with my cousin recently, and, being the geeks that we are, were talking about the people we follow on Twitter. One of the first things he mentioned was that, as a result of following me, he’d picked up a lot of interesting people to follow (because I’m really cool and have cool friends, obviously). He said, though, that he was kind of annoyed by their ‘spam’ – that is, a lot of the people he knew were interesting would often tweet in decidedly uninteresting ways (not naming any names….). He didn’t want to unfollow them, because he did want to be updated on their important doings, but disliked the majority of their tweets, which could be called non sequiturs, to put it politely. I’m guilty of this too, by the way. He said, “wouldn’t it be cool if there was an app that just showed you the tweets that you theoretically care about, while filtering out all the crap? It would probably be pretty easy to make. You should build that.”
So I did.
I put together Quadroopl over the course of the last three days, flying by the seat of my pants as I learned both OAuth and the Twitter API (and, arguably, Javascript and Ajax) at the same time. The logic goes like this: your whole (500 tweet) feed is retrieved, and each tweet in the feed is processed, being put into a session hash of hashes. A tweet is added to the hash if the user who created it doesn’t have a tweet in the hash already; or, if they do, it’s added if it’s been retweeted more times than the one currently in the hash. The result is a new feed that’s made up of the most-retweeted tweet by each user. If a user hasn’t been retweeted recently, they don’t show up on Quadroopl. This method places a lot of importance on retweets; I think it’s an ok solution to the problem of the “top tweet”, but I’d love to look for a way to retrieve the number of replies to a tweet. Another vector for the comparison of tweets could only make the results more relevant, I think.
I’m very happy to have gotten this out so quickly, and I look forward to adding features/revising the logic to make a more enjoyable user experience. Try it out, and please direct any constructive criticism to @emmett9001. Direct all other criticism to your mother. Thanks.