How the Green New Deal Has Played Out Online
Earlier this month, Rep. Alexandria Ocasio-Cortez and Sen. Edward Markey released their ambitious proposal for a “Green New Deal” to urgently address concerns over climate change. What did the reaction to their proposal look like across Twitter, traditional mainstream news coverage and web searches?
Ocasio-Cortez has become the de facto face of the proposed policy, but she is not the first to generate Twitter buzz for a “#GreenNewDeal.” That honor goes to 2012 and 2016 Green Party presidential candidate Jill Stein.
The timeline below shows the total number of tweets per day mentioning “#GreenNewDeal” or “GreenNewDeal” or “Green New Deal” from July 1, 2010 (the beginning of the data) through Feb. 17, 2019, using data from Crimson Hexagon.
The first slight bump from August 2012 through November 2012 largely centered on posts such as an Aug. 18, 2012 tweet from Stein in which she offered “Our #GreenNewDeal will redirect research funds from fossil fuels to research in wind, solar & geothermal. Like? PLS RT.” Mentions abruptly ended the day after the election.
Four years later, Stein managed to sustain Twitter conversation around the #GreenNewDeal for much of 2016 with a series of tweets criticizing fossil fuels, promoting the idea of job creation and noting the impact of climate change on minorities.
Once again, the Twitter conversation abruptly ended the day after the election.
Ocasio-Cortez is the latest to adopt the idea of a Green New Deal, resurrecting it as part of her campaign for Congress last year and adopting many of the same concepts and focal points as Stein’s proposal.
The graph below zooms in on the timeline above (from Nov. 1, 2018 through Feb. 17, 2019) and shows that the Twitter conversation did not begin immediately upon Ocasio-Cortez’s election. Rather, it was her Nov. 13, 2018 occupation of Nancy Pelosi’s office that began sustained conversation about the Green New Deal, reminding us of the power of physical activism for propelling topics back into the national discourse.
Attention remained relatively steady until a GND press conference on Feb. 7, 2019, peaked the following day as the press, public and policymakers digested the proposal, and then has steadily decreased since.
The timeline below compares Twitter against worldwide online news coverage mentioning “Green New Deal” in the 65 languages monitored by the GDELT Project. Since there is a substantial difference between the number of daily tweets and the number of daily news articles, the timeline normalizes their volumes into a “Z-Score” that reports the number of standard deviations from the mean rather than raw counts, allowing the two to be plotted on the same graph by comparing their relative changes rather than absolute counts.
News media coverage appears to have overall followed a very similar trajectory to Twitter. It is particularly noteworthy, however, that Twitter barely registered the Jan. 3, 2019 announcement that the new House Select Committee on the Climate Crisis would not focus on the Green New Deal language favored by Ocasio-Cortez and others, though the news media covered it heavily. Given Ocasio-Cortez’s fervent Twitter following, it is surprising that there was not more reaction to this snub.
Both the news media and Twitter covered Ocasio-Cortez’s Feb. 7 press conference heavily, but both also show a rapid fading of coverage over the following week and a half.
The timeline below examines this reaction to the press conference in more detail, zooming in on the day before and two days after at hourly resolution (all times are Eastern Standard), using the same Z-Scores as before. To examine how all of this media and social attention influenced public interest, it also plots Google search volume on the same axis.
While Twitter is often described as being so fast that it “beats the news,” in this case the heavy promotion of the event and lack of specific details as to what would be announced meant media outlets spent much of Wednesday afternoon speculating. Twitter, on the other hand, largely remained silent until the day of the press conference.
It is also worth noting that 82 percent of the tweets above were actually retweets and 57 percent included a link, rising from 50 percent during the press conference to around 70 percent by the end of the week. This suggests that much of the Twitter commentary involved users simply forwarding information rather than offering their own perspectives.
From 7 to 8 a.m. the morning of the press conference, media outlets released a flurry of coverage commenting on their expectations and rehashing Democrats’ concerns about climate change.
The press conference itself led to a vertical surge of Twitter commentary that lasted from the presser itself through 9 p.m.
News coverage, on the other hand, peaked during the press conference as outlets published their live takes and reports, then decreased for several hours as they digested the details, before peaking again in the 6-7 p.m. hour with the final reporting of the day. News coverage actually increased over the following two days as the press digested the ramifications of the proposal, while Twitter and search interest largely trailed off.
Google searches began trending upward in the 5-6 a.m. hour as people woke up and looked for the latest news on that day’s press conference. Unlike news and social media, Google searches increased only slightly during the press conference itself. Instead, they steadily increased through a peak in the 9-10 p.m. hour. They also remained elevated through the night and did not reach a peak until the 8-9 a.m. hour the following day.
The slow buildup of search interest reflects the fact that the press conference, while important to climate activists, was not otherwise a major event. Interest largely grew organically through the day as people had time to check the news and when they got home from work that evening.
The difference between search interest and news and social attention also reminds us that there is a tremendous and underappreciated difference between the production and consumption of information. When we analyze public reaction to legislation, we typically measure production in the form of news coverage and social media posts. Consumption, in the form of web searches, tells us which aspects of those narratives actually resonated with the public.
The timeline below plots the Google Trends results for four purported themes of the Green New Deal proposal that received particular attention: restrictions on air travel, Universal Basic Income for those “unwilling” to work, “farting cows” and forcing citizens to drink their own urine. (Universal Basic Income and “farting cows” were found in the original FAQ published on Ocasio-Cortez’s website but later deleted after proving controversial; the basis for the urine search is explained below.)
(Google Trends does not report the actual number of searches; instead it reports a score of 100 during the hour with the greatest number of searches and reports the remaining hours relative to that peak. To focus on their hourly timing, the four searches are displayed independently of each other in the graph below.)
Restrictions on air travel were the first of the four to generate substantial search interest, beginning around 5 a.m. and increasing sharply more than two hours before the press conference. Providing for those “unwilling” to work and cutting back on the number of cows in the country captured attention during the press conference. After the conference, the three received fairly similar attention.
The one outlier in the graph above is searches for urine drinking. In the hours after Ocasio-Cortez’ press conference, a hoax version of the proposal was circulated online that had been altered to claim that to conserve water, Americans would be required to drink their own urine. The fact that this topic appears so differently in the graph above suggests that it might be possible to identify future hoaxes by looking at how interest in them differs from related topics.
Putting this all together, we see a number of interesting trends. For scripted events like a political press conference, Twitter is not actually any faster in its coverage than news media and the two offer a very similar picture in terms of overall attention. However, much of the Twitter conversation in this case was limited to users forwarding links and other people’s tweets, rather than contributing their own thoughts. This suggests Twitter’s use as a mechanism for understanding popular reaction to policy events may be more limited than we have previously believed. Search volume grew slowly over time, reminding us of the difference between production and consumption of information. Search interest also allows us to see which narratives are resonating the most with the public.
In the end, perhaps the biggest takeaway is that despite social media increasingly becoming the tool du jour of the political class, the graphs above show that to truly understand reaction to legislative proposals, we still very much need to keep an eye on traditional news media and on search trends.