h1. information overload and twitter p(meta). 17 April 2011 h2. How I use twitter I have been an on and off user of twitter since it first came out. There are hidden gems in the tweets I get, but most of it is just plain junk. It can be quite daunting to add another stream of information to go through with all the emails, blogs, and other information we are already processing on a daily basis. *What I consider junk tweets could be considered gold to others* so its hard to devise a system that works for everyone. I used to follow everyone who I thought is interesting, but as the rate of tweets have increased, I've had to remove lots of people. Now I am a lot more selective in who I follow because of all the noise that gets added to my twitter stream. I currently follow about 200 people right now and even now I think that number is too high. There are many people who have thousands of friends and I honestly can't imagine how they deal with it. I think we could all follow a lot more people if a better system were put in place. There are lots of interesting people to follow that talk about things I am interested in, but they also talk about lots of issues that I am not interested in. I'm going to outline who I use twitter and how I think it can be improved in a personal setting, I'll write about how I use twitter for my "company":http://socmetrics.com in a later blog post. !http://i.imgur.com/rRZgV.png! h2. tweeting I am still trying to figure out how to best tweet. I currently tweet any interesting articles I find. I also retweet any tweets and quotes that I find interesting. I use twitter as a memory bank to store interesting links and quotes that I like. Often times I will comeback to my twitter feed to search for that link, basically acting like a crude "del.iciou.us":http://delicious.com. I don't tweet out my own thoughts or content that often because I have this fear that I will aggravate some of my followers. So when I do tweet I try to only tweet content that is general for as wide an audience as possible. I think my approach is flawed and it limits me from tweeting more often. If there was a better way for people to filter out certain kinds of tweets I would be more inclined to tweet out content. I know hashtags are often used to categorize tweets, but I don't think there is enough support in the client to deal with hashtags. Most clients just have a simple search interface to find tweets from a certain hashtag. Twitter "talked":http://dev.twitter.com/pages/annotations_overview about allowing annotations and metadata to appear in tweets, but has not released this yet. I would prefer to have the hashtags appear in the metadata so that we had more space for our tweets. This would also allow us to add more hashtags to our tweet. h2. intelligent filtering If most tweets had hashtags and people were following a common naming convention for hashtags, then we could filter out on the client side tweets that matched certain hashtags. I would actually go through each of my twitter friends and figure out what hashtags I am interested in on a per user basis! For example I just browsed "@scobleizer's feed":http://twitter.com/#!/scobleizer and I see that he mentions his schedule quite often. I would like to filter out his #schedule tweets and read his #startup tweets. The types of content/hashtags I am interested in is typically about #startups, #entrepreneurship, #social media, and #software development. This hashtag system is not currently in place nor do I know if it will ever be there. I think we can get fairly close though using some other methods. _source http://www.fogomedia.com/blog/makeinfouseful/using-twitter-as-information-filter/_ I read tweets to find interesting content and learn from people who I respect. I don't need any foursquare, other check-ins, or pictures appearing in my feed, so we can filter all of those out. Most of the time the interesting content is usually a link to longer article so I could potentially filter out most tweets without a link. I also like seeing questions that people ask. It helps me understand how the people I follow think and it also allows me to find opportunities to engage and talk with them, so we can keep all questions. Another measure of tweet relevance is how many retweets it got. I think its fine to filter out the tweets that did not get any retweets, but I would want to see tweets that have a high retweet count. For certain users I would always want to see their tweets. For other users I would want to be able to completely mute them for periods of time, maybe they are at "sxsw":http://sxsw.com for the week and I can't make it out so I don't want to see all the tweets about how awesome all the parties are. I would also want to be able to see certain tweets from users based off of how often they tweet. So if someone who hasn't tweeted in months hasn't posted and all of a sudden they tweet something, I would still see it even though it was filtered out from the previous filters from above. For people who are constantly tweeting, the filters would apply more heavily to them so as to know overload you with tweets. !http://i.imgur.com/cqm00.jpg! h2. clients When sending a tweet, the client could have a list of your common hashtags and allow you to click on them to apply those hashtags to your tweets. This would save you time from typing out the hashtags and would make you more inclined to add the proper hashtags to your tweets. Followers reading your tweets could filter out your tweets based on their hashtag filters. When viewing the filtered stream I mentioned above, the default view would have strict filters applied. The interface could have some controls at the bottom of the screen to allow you to change the filter modes to see more tweets. This would allow the user to quickly understand what is happening in their "twittersphere", but also be able to see all the tweets if needed. With twitter's "new policies":http://groups.google.com/group/twitter-development-talk/browse_thread/thread/c82cd59c7a87216a, they have made it clear that they don't want new clients to be made for twitter, so its hard to see a filter system like this being built. I currently read most of my tweets on my iphone and so I would prefer a native and fast iphone client, but I could use a web based mobile client if it was fast enough. h2. conclusion There are currently 1,200 tweets generated a second, that is 100 million tweets per day. With the amount of tweets increasing at such a fast rate, the amount of information we need to deal with becomes painful and unmanageable. If a better solution is not presented, twitter will become impossible to use. Does such a client already exist to get us the relevant tweets? I've heard of some clients doing parts of what I've suggested, but I've never had the opportunity to use them. I'd think about building a client myself if there was enough interest in this. Are there better ideas than the ones I've suggested to improve twitter? Or are my ideas totally crazy? I'd love to hear about how you are dealing with the twitter feed.