Twitter Launches Political Index

colinneagle writes "Twitter today launched a new tool that leverages its estimated 400 million daily Tweets to gauge public opinion on the candidates for the 2012 presidential election. Progress in political polling is long overdue, and with Twitter providing a constant, international conversation for web users to join or leave at their own will, there may not be a better time than now to make that change. However, there are some concerns. One of the interesting points made in Twitter's description of its new tool is where it claims to be 'illustrating instances when unprompted, natural conversation deviates from responses to specific survey questions.' That assumes conversation on Twitter is natural. If parody accounts, Twitter trolls, and spam bots have taught us anything (and they usually don't), it's that Twitter conversation can be manipulated just as easily as it can be used naturally. How will Twitter distinguish between positive Tweets coming from voters or news outlets and those from spam bots designed to drive the conversation surrounding a candidate one way or the other? How easy could it be for an organization with a vested interest in positive poll numbers for one candidate to craft an army of Twitter bots designed to drive Barack Obama's positive numbers down, or vice versa? How many people reading the data, which is sure to make its way to TV news as election coverage increases in the coming months, will be aware that Tweets can be manipulated?"

Twitter is only used by a certain sub-group!

[Anonymous Coward]

What use is a index, that is only taken from 40+ men in their midlife crisis, PR/marketing companies and other groups that want to look "hip" and "with the young people"?
Nobody who's actually young, uses it. Ever.

Oh wait... It's not supposed to show the political climate, but create it. My bad.

trending hashtag analysis is a research area

[Anonymous Coward]

A friend of mine just got a job doing some sophisticated, geo-ip indexed analysis of trending hashtags, which can be used for remarkably sophisticated real time marketing analysis. I don't know all the details because they're secret and he can't tell me too much but I know they can figure out in real time what is on the minds of people in various geographic areas. This can easily also be used for political analysis.

Re:Twitter is only used by a certain sub-group!

[Rewind]

You seem to have a very odd notion of Twitter and how it is used. A large variety of people use it for an incredible diverse list of reasons. I use twitter, but I never 'tweet'. It just provides me with a simple single page where I can check for updates on various communities/podcasts/webshows that I follow. I have no idea why you think that only 40+ year old men use it.

Centralized Comm is Perilous

[Bob9113]

How many people reading the data, which is sure to make its way to TV news as election coverage increases in the coming months, will be aware that Tweets can be manipulated?

My estimate is rough, but I would put it well under 20%, based on conversations about the topic with average non-technical people.

This is an example of one of the indirect perils of centralized communications. Even without the central authority controlling the content, the power implicit in comm centralization becomes a weapon against the free mind. If we don't replace Twitter, Facebook, G+, Hotmail, and the rest with decentralized alternatives, our society will increasingly be influenced by entities with the means and desire to alter public opinion.

We need to be running the chat servers, photo buckets, and mail servers used by our friends and family who are less technically skilled. We need to get Diaspora (or a competitor) nodes running on a much larger scale. I am doing some, and I am scaling up as quickly as I can.

Decentralized comm does not magically and completely solve the problem, but at least it would not serve up the means to manipulate public opinion on a silver platter.

Use betting odds

[Alkonaut]

A much better predictor is betting odds. Beats polls and any other index I have seen. As for the quality of twitter as a gauge of opinion, they just face a normal page rank problem. The weight of a quote about a candidate is determined by the rank of the account and the number of tweets from that account. The rank of the account depends on the followers and their rank, with validated accounts in the top. I didn't read tfa but it would surprise me if an algorithm like this isn't used. An army of bots would only have bots as followers. Just like splogs and other "ring" spam web sites. Deciding whether a tweet is good or bad is difficult. If you can't sense e.g. sarcasm in political tweets you will miss by a large margin. Doing that from 140 chars (plus history) would be a more impressive feat than using I for political indexes.