Apr 20 2010
Spend any time on Twitter, and you quickly learn that looking at the unfiltered feed of tweets is like drinking from a firehose. There are a number of human ways of getting around this problem: setting a “Twitter check” appointment, or only reviewing certain Twitter lists. But, inevitably, you miss so much good stuff.
Enter Cadmus, which dramatically improves the signal/noise ratio, and gets you to the good stuff on Twitter fast.
Despite Twitter’s massive growth, people that don’t use it still have a tough time understanding its usefulness. Usually, the first question you get is, “Why would I care what my friends ate for breakfast?” Then, after you explain that no one actually uses Twitter that way, the next question is, “I can barely keep up with email and Facebook, how am I supposed to keep up with that?”
And, despite all of the sturm und drang of the digerati, those prospective users are absolutely right. Twitter is a platform that is not very useful by itself, because it’s either a time suck (if you want to gain real value out of it), or you always wonder what you’re missing.
Many third-party apps have moved in to fill the void, but most suffer from similar problems: too much user intervention, too steep of a learning curve, and too much ongoing maintenance to significantly reduce the amount of time you spend. I won’t name names here.
Cadmus approaches the information overload from sound ideological footing. In a recent conversation, Jay, one of the two-man team that builds Cadmus, told me that, “there are several other [Twitter] tools that do a really good job in other areas but we would like to focus on answering one question; ‘What are my friend talking about?’ ”
This approach addresses the problem of filter failure head on – by making a better filter. This is really critical as more firehose services come online – users need better filtering tools, especially when they are Tweeting for work, and need to find valuable content quickly (as I discussed in March)
Cadmus does this using very smart algorithms to provide advanced filtering and prioritization of Tweets, all wrapped up in an easy-on-the-eyes interface. They use four primary ways to determine what are the most important Tweets for your limited attention span:
- Your last login: Cadmus indexes all the activity from all the accounts you follow from the last time you logged in to Cadmus, until the present. So, you are never in danger of missing an important tweet – it will appear first the next time you log in.
- Overall interaction levels on a specific tweet: If a tweet has been retweeted frequently by people within your network, it’s considered more important, and will appear higher in your list to review
- Overall interaction levels with a source: If you retweet 50% of a certain Twitter account, that account will continue to rank highly for you, and future tweets will appear higher. The percentage is what’s important here – it doesn’t matter how often they tweet.
- Your own interactions: If you retweet, @reply, or look at related tweets from a user, Cadmus determines that the user is more important to you. That will cause their future Tweets to rank higher.
The results are impressive. Suddenly, when I log in, I am updated with exactly what I want from Twitter: lots of tech news and some updates from friends. Here’s an idea of what I see, based on my login tonight:
By my count, Cadmus is all blue skies on vetting what’s important: two posts about the new iPhone story that Gizmodo broke (by a wide margin, the biggest tech story of the past 24 hours), a social media stat (always useful when prepping presentations), and news on e-reader platforms (another ongoing area of investigation for me).
Cadmus also fills in other crucial information: it calculates your personal trending topics (much easier than implementing a script to get rid of Justin Bieber), and does the same priority filtering for Twitter Lists.
And, Jay shared some of the great new features and functionality that Cadmus will be rolling out soon. It’s really great stuff, and will make Cadmus even more useful. Stay tuned!