The End of Public Opinion Polling
By Lisa Conn
I spoke as a “Luminary” at The Future of People conference yesterday, a gathering that imagined the year 2050 and designed a new framework for futurism. Fast-forward: What is your vision for 2050? Rewind: What has to happen to bring your vision to life? Play: What are you doing now? Read about my 2050 fast forward vision: the application of artificial intelligence to bring about a new flowering of democracy.
Fast Forward.
By 2050, public opinion polling as we know it today will be extinct.
To understand polling, we need to go back in time. Polling was born 80 years ago, designed to help manufacturers make and market products of mass appeal. From there, it was applied to politics and successfully predicted the outcome of many Presidential elections. It has taken 80 years to break, and it may take another 30 years or so to be completely replaced.
But by 2050, the traditional forms of understanding public opinion that are already being disrupted will be completely eliminated — and replaced with something new: the use of artificial intelligence to collect, analyze, and translate the public sphere in its own voices, as expressed on the channels that most of us use today: Twitter. Facebook. Instagram. Google. Snapchat.
Imagine a social scanner, as my colleague Andrew Heyward calls it, that makes sense of our digital voices, and in return, offer us meaningful, dynamic, constant measurements of real public opinion — and the ability to double-click into stories to give life to the statistics.
Technology has given everyone a voice, and now, through the rise of machine learning, the ability to listen.
Rewind.
In order to get there, two forces will occur in parallel: the decline of polling and the rise of social scanners.
Today. Already much of public life has moved online, new electoral coalitions are emerging, and trust in institutions, like polling, is eroding. Meanwhile, early uses of artificial intelligence are already underway to measure the public response (more on this later).
In the next 15 years. Polling — calling a limited number of people, asking them a set of predetermined and inevitably leading questions, and expecting it to represent everyone — will seem silly, because we will live in a time in which we all have a measurable voice. Meanwhile, AI will have created a web of detailed information about every aspect of our lives including the political issues we care about. And we will see the benefits of translating that data, through social scanners, emerge: dynamic measures of public opinion will allow for a more responsive governance, increased collective understanding of the public sphere will burst filter bubbles, increasing empathy and understanding, and the creation of a scientific truth will enable journalists better reflect the public sentiment.
By 2050. Traditional polling will be completely extinct and sophisticated social listening tools will be the gold standard.
Play.
I’ve spent the last six years working at the intersection of technology and politics. The Obama campaign. Mark Zuckerberg’s immigration group, FWD.us. Facebook. Currently, I work at a machine learning lab at the MIT Media Lab. We are the Laboratory for Social Machines, and we built something called the Electome with support from Twitter and the Knight Foundation. It is a first step in making sense of all those new voices that are speaking up online.
We have the data, the science, and deployment strategy. With access to the full Twitter firehose of approximately half a billion tweets per day, our cutting-edge algorithms pull out election-related tweets and classify them with 90+% accuracy. We deployed our findings to major media outlets, including the Washington Post, CNN, and Bloomberg — and were an official partner of the Commission on Presidential Debates, providing our data and suggested questions to all the debate moderators.
And I’m convinced this approach is more important than ever. No one looked to Twitter to predict the outcome of the 2016 presidential election because it’s considered too left-leaning, and because trolls and fake news eroded the credibility of the whole platform. But that said, our data showed something surprising and powerful: on this traditionally left-leaning outlet, Trump and his supporters drove the conversation. In the days leading up to Election Day, 84% of the conversation was about Donald Trump and his supporters tweeted 2 times more than Hillary Clinton supporters. That was a signal to which we all should have paid more attention.
Social media and the digital revolution can be a resurrection of the town meeting, where everyone has a voice that is listened to and considered.
To achieve this, we must support institutions, candidates, and causes that are pushing for transparency, authenticity, and commitment to accuracy in the public sphere; subscribe to them, volunteer with them, or create them at the local level.
And we must reach across the aisle. Force yourself outside of your comfort zone, populate your own social network with other points of view, try to avoid that comfortable complacency that we have all settled into. Strive to understand and engage to achieve positive social change.
Technology has given everyone a voice, and now, the ability to listen — and respond.
By 2050, we can create a new flowering of democracy. Let’s build it together.
