On December 17th — eleven days ago — I emailed my friend Ashish over at Setfive Consulting with a pretty simple idea: a topical meme tracker, using pre-selected RSS feeds to limit the scope of acceptable articles and bit.ly to weight and further filter the wheat from the chaff. Specifically, as a die-hard New York Mets fan who has a second identity on Twitter, @metstweets, I wanted to be able to tweet out “hot” Mets stories. I considered using TweetMeme, but I really didn’t want some generic article from an out-of-market newspaper, something non-English, spam, or even worse, something entirely unrelated to the “Mets” I’m talking about.
(Also, I wanted to test a model for a startup: a reverse-incubator. But that’s a story for another day — probably later this week — so if you’re interested, subscribe to my RSS feed.)
Limiting the universe of potential sources seemed like the right starting point. So I quickly wrote it up and sent it to Ashish to see if it was a reasonable build. The end result is BitMe.Me, and you’ll note that the front page has nothing to do with the Mets. The tool itself applies to any vertical, which is why I liked the idea so much, and can be adapted incredibly quickly. When you see how it works, you’ll see why.
We started with my target vertical — the Mets. I read enough Mets content to know that there are fewer than 100 key sources, and while I’m sure there are diamonds out there in other sites, I am willing to forgo them in my meme tracker. Why? Because as the volume of content expands out from beyond the core group, the signal to noise ratio decreases. Also, because the number of sources is relatively small and, in any event, finite, this allowed us to build an index without having to worry about spidering, discovery, etc.
Ashish instinctively expanded it to bigger verticals — technology, sports, gossip, politics, and news — and that’s what you see on the site right now. Yes, there’s a Mets section as well, but that’s not the focus of the site. (And we’re working on a problem which causes MetsBlog content to not register correctly.)
Pretty straight forward. Check it out and let me know what you think in the comments, or by shooting me an email.
It seems that the problem with tweetmeme is that, although it can determine popularity, it does not indicate which posts are on-topic for Mets. bit.ly suffers from the same problem though.
Couldn't bitme.me be built on tweetmeme instead of bit.ly but be restricted to links in the feeds currently used by bitme.me ?
Just curious.
Probably. I'm not sure what kind of API tweetmeme has. Bitly's was
(relatively) easy to use, given what I gleaned from conversations
w/Ashish about it.
The advantages of bitly: it's not limited to tweets (although in
effect, it may be) and you can score links via clicks — and therefore
engagements — as opposed to shares.
The advantages of tweetmeme: much wider net — I'd bet that it tracks
more tweets (100%?) than bitly does links generally; and it may be
better to score via shares than clicks.
Dan, congrats with the entrepreneurial endeavor. I love these types of projects. I was thinking about something similar – glad to see you nailed it. How's the reaction been?
Thanks!
It's definitely a side project — we're all busy with other things
(i.e. jobs) but we've been able to attract some help, which is good.
Not a lot of buzz around it, but also, not a real effort to get buzz.
We'll see how it goes, but it could be really cool.
Nice job, Dan! If you'd like another signal perhaps PostRank would help? It includes tweets as well as 20 other social sites and on site comment comment counts.
You can see the API documentation at our Dev Wiki: http://apidocs.postrank.com
Cheers.
Cool idea. We'll look at mixing it in this weekend.
-Ashish
Thanks Jim!
Thanks Jim!