So I’ve been experimenting with Twitter a bit more over the last month – not a obscene amount like a number of tech geeks out there (I’ve posted only 7 times since Mar. 30), but definitely a ramp in usage for me (during the 10 months prior I posted only 4 times).

In any event, my experimentation has lead me to come to two conclusions:

Observation #1:

People are no longer using Twitter just for broadcasting “what they are doing” despite what Twitter’s home page says. To prove my hypothesis I looked at the most recent 20 “tweets” from the people that I follow (this is what the home page displays by default) and classified them into the three primary use buckets that I’ve observed: (1) status message (i.e., the original use case for Twitter), (2) 1-to-1 messages, and (3) random thought/linkblog/microblog (i.e., everything else that is not a 1-1 message or a “what you are doing” status update). The end result was:
– 40% status messages
– 10% 1-to-1 messages
– 50% random thought/linkblog/microblog

Obviously this is a small sample set, and one that is heavily skewed to the heavy Twitter users that I follow. Nonetheless, it gives some sense for how people are using it. Interestingly, if I look at the last 20 tweets from everyone (not just the people I follow):
– 60% are unknown (i.e., I don’t know because they are not in English)
– 5% are status messages
– 5% are 1-to-1 messages
– 30% are random thought/linkblog/microblog

What does this all mean? It means that Twitter is evolving from a single type of communication into a mode of communication. If you are Twitter, this is all very good news. Their addressable market is expanding.

Observation #2:

Twitter is becoming a dumb pipe. Both on the authoring and receiving ends, it seems that many people are using end points other than Twitter.com (see here and here and here). Once again looking at the last 20 tweets from my friends, I see the following distribution of “authoring” end points:
– 35% twitter.com
– 30% twhirl
– 20% blog it (facebook app)
– 10% text/mobile
– 5% twinkle

Of course, using a limited sample data set from my friends only is even more problematic in this context as compared to the “message type” context described in observation #1. In this context, there are many more end point options than message types, and as such you need a much larger data set to make meaningful conclusions about which are most used. That being said, the last 20 tweets from everyone (not just my friends) indicated a similar pattern (i.e., only 45% of tweets were made via the web on Twitter.com). So while I’m very likely way off on the popularity of the various third party applications vs. each other, I’m probably not that far off on the popularity of Twitter.com vs. all of the third party applications combined.

And what does this mean? Not good for Twitter. Over time I will not be surprised to see some of the end points generate more value than Twitter the company. The end point is where users interact with the service. That is where companies can serve ads, bundle toolbars, include search boxes, etc. There is no doubt that opening up the network to third party developers was a huge driver of growth for Twitter the network. But if Twitter doesn’t invest in their own front end applications they might find it very difficult to ever generate significant value for themselves in the future.

Jon Postel may have been the driving force behind the SMTP standard, but he never achieved the same commercial success of Microsoft Exchange, Yahoo! Mail, Hotmail, or even Gmail. Developing a standard can generate lots of value for users and developers across the Internet, but it’s not necessarily a good recipe for commercial success.