On Twitter & Prediction Markets
About a year back I blogged on Twitter and how I thought the future was in data-mining. Since I have lost my posts the only remaining post is here (via istockanalyst). In that post wrote that there were 4 main ways in which Twitter could justify the inefficient use of database space (and essentially become profitable):
Today, I read a post on Twitter (via Abnormal Returns), the microblogging service that Fred Wilson, someone who really “gets” Web 2.0 is invested in and whose opinion I respect tremendously. The post, which I will comment on briefly, made me think about Twitter and its use and application.
In the post, Om Malik, writes that Twitter faces a “scalability” issue. I agree that it most definitely does, as there is tons of database space that is inefficiently used. I believe that it can be solved if 1) Twitter becomes profitable enough to justify such space, 2) Limits and/or Subscriptions are implemented, 3) A sophisticated and “smart” taxonomy is applied to “tweets”, and/or 4) Twitter evolves into “something else”.
As Twitter appears to be getting exponentially more popular it seems that it will become even more important to ensure that the newfound popularity isn’t putting the company further into the red.
Today they announced that they would implement “premium accounts”, which appear to be focused around charging for the ability to add more words to each tweet. While I said that a “subscriber” model would definitely help them, I see this as somewhat counterproductive and dangerous. The one thing I love about twitter is that the tweets are short. As someone who tends to be long winded it forces me to get to the point and in a society that is collectively ADD, this is a good thing. The goal of Twitter should be extracting more value from each tweet, not charging for longer tweets.
I had mentioned last year that datamining is the future of twitter. I like the idea of smart-taxonomy that allows users to WANT to organize their tweets into more mineable data. The concept behind Stocktwits is great because all it does is say, “Add a “$” before the symbol and $$ after the tweet and we will attached this tweet to this stock in our architecture.” That is genius because now it gets all of those interested in stocks and Twitter together to share information quickly and efficiently. It also allows traders to mine the data however they see fit. The drawback of course is that eventually it should follow the evolution of the YHOO message boards – as the market learns that its being mined, they will use that to “manipulate” the market itself.
This is why I believe the future of Twitter is linking it in with prediction markets. Prediction markets (real-money and play-money or “prize” based) can be set up to reward those who submit the most valuable data and limit people in how much data “equity” they have. Users can be rated and “bought” and “sold” and thus the most valuable tweets will come from the most valuable users, and vice versa. This would allow for data to be saved more efficiently as it would effectively allow the market to say what data is worth something and what data is worthless.
Eventually I see the convergence and integration of traditional Search algorithms with some of the cutting edge market making algorithms as a way to design a sophisticated macro and micro market structure for an information market where “market makers” will be able to adjust prices based on whether they have transacted with an “informed” trader or a “noise” trader. This will reduce the storage cost of information in general by allowing those who create noise to effectively pay for those who create value.
I am just investigating this idea, so if anyone has any information on those who are currently doing this, I would be interested in hearing about it.