I wrote yesterday on whether Twitter qualifies as public opinion. I opened with an attack on the notion, one based on the methodological challenges of using the micro-blog as a measure of opinion. I then ended by suggesting that the earlier me was wrong, that Twitter may better resemble our classic understanding of the concept of public opinion.
Let me take a stab at following up on my defense of Twitter. At the end, I'll try to extend this from a philosophical concern to a more practical concern for journalists.
Before the existence of sophisticated polling, early thinkers had a very different understanding of public opinion. Its roots can be found in the coffee houses of England and the salons of France. In an 1820 letter, Sir Robert Peel complained about "that great compound of folly, weakness, prejudice, wrong feeling, right feeling, obstinacy and newspaper paragraphs, which is called public opinion."
Sounds like Twitter to me.
By this, I mean that early concepts of public opinion were based on the idea of communication. While I'm a poll guy and use survey data extensively, public opinion is more than merely what public opinion polls measure. Public opinion is an 18th Century invention, and at its heart is public discussion. It is, to borrow a phrase, an "organic sociological process" that takes part in the "public sphere."
As John Durham Peters writes in an excellent chapter that sums up much of this early thought, "in reading the newspapers, the public reads about itself, and thus finds ways to come into existence."
Take out newspapers, put in Twitter. It's a nice fit. After all, in Twitter people can talk back and forth, at "influentials" and others who fill the twittersphere. From this, theoretically at least, public opinion is a changing, ephemeral, difficult to grasp. And hard to measure. Polls are much easier. And thus, public opinion becomes, not what it was intended to be, but what became easier to measure.
Polls matter. Surveys are useful. Uncanny in predictive accuracy for elections, excellent as snapshots of what people think about a question asked of them at a specific time, they are filled with usefulness in their generalizability. I love 'em. Can't live without 'em. But as measures of the classical sense of public opinion, they fall desperately short.
Can Twitter be any better?
Snapshot counting of positive versus negative tweets falls into some of the same traps that polls do. All the million-tweet analyses cranked out by computer scientists do little to address what we really mean by public opinion. It's a difficult concept. As V. O. Key wrote: "to speak with precision about public opinion is a task not unlike coming to grips with the Holy Ghost."
To work, this will require significantly more sophisticated analysis techniques to tease out not the snapshot, but the tendrils of communication and change that occur as a topic or topics bounce across the twittersphere.
It's a methodological nightmare. But it's probably a better sense of what we truly mean by public opinion.
So, what's this mean for those in journalism?
Very soon we (the royal journalistic we) will face folks trying to sell us Twitter counting as an alternative to polls. Surveys are expensive, counting tweets -- less so. I suspect by 2012 we'll see some of the major players, the CNNs and New York Times of the world -- messing around with tweet counting either done in-house or, more likely, through consultants. As I discussed in detail in the previous day's post, doing so is full of problems. Yes, cheaper. And yes, above I argue it more closely resembles our classical understanding of what public opinion means. But from a practical standpoint, we simply are not there yet. My warning is this: journalists, be very very careful in using such data, if you use it at all, except in tandem with traditional polling data. Do NOT rely merely on tweets. We aren't there. Yet.