Downgrading Data and Opening Up the Democratic Party
JOHN REMENSPERGER
University of Pennsylvania, USA
Abstract
The U.S. Democratic Party’s over-reliance on consultants and technocrats over the last several decades has created a chasm between the party’s leadership and its voters, the consequences of which have only begun to be revealed via the 2016 election. Through the adoption of digital technologies and fetishization of data above all, it has centralized its strategies in the hands of an elite class of overconfident political experts and number-crunchers whose practices are characterized by the rapid platformization and professionalization of party processes. Placing these changes in the context of the party’s long history of control, this manifesto critiques these practices, particularly the use of predictive analytics and the inability to see volunteers beyond tools to gather and input data. It argues that the party needs to focus on collaborative, grassroots party-building activities and leadership development of local volunteers. Its focus must shift from one of control to one that encourages and emphasizes contention and pluralism as positive aspects of democracy rather than as problems to be obscured or controlled.
Keywords
platforms, parties, technocracy, data, politics
The Democratic Party is in denial. It has spent a decade practicing a religion that has blinded it from the changing priorities of the electorate and from the needs of state parties. In service to this religion, it has centralized its strategies in the hands of an elite class of overconfident political consultants and number-crunchers who, like their financial brothers and sisters who ten years ago brought about the collapse of the housing market and started a global recession, have used complex models to create a political bubble that has now burst, bringing the Democratic Party into a precarious position that will take years to repair. This religion is political technocracy and its basic tenets are that the electorate is a puzzle to be solved and that it can only be solved by specialist experts. One need only compare the expansion of Democratic and Republican digital and data firms over the last decade to see the rapid overdevelopment of a class of elite consultants aimed at servicing the party with hundreds of data and analytics products[1]. The practices of these elites are characterized by both data fetishism – in which “techniques and numbers are…put on a pedestal for what they are rather than what they do”[2] – and the rapid platformization – in which practices are consolidated via software products that “become the nexus of an ecosystem of partners that are dependent on its product”[3] – of party processes. While these specific practices are relatively recent phenomena, they are in line with the long-standing tradition of political parties subverting their own democratic rhetoric with systems that prioritize the decisions of political insiders over those of voters.
A History of Control
The embrace by the Democratic National Committee (DNC) of technocratic practices – in which the decisions of elites and experts are valued above popular democracy[4] – grows out of wider party traditions of both over-centralization of processes and attempts to exert control over party members and partisans. Since the 1970s, the party has resisted popular attempts to wrest control of the party’s nomination process from the party elite which sees its own expertise as invaluable. Nowhere is this clearer than in the continued existence of superdelegates to the Democratic National Convention. Superdelegates – party insiders who vote at the Convention but are not bound to a popular vote – were created in 1980 to return some power over the nominations process back from primary voters to party elites, allowing them a technocratic failsafe in the event that they find a popular candidate to be undesirable. Despite the fact that superdelegates have only impacted the outcome of a nominations process in one election held over three decades ago, the party has not moved to get rid of them despite intense pressure from partisans and elected officials. And yet, in the current political and media environment, it is implausible to imagine the party elite replacing the candidate who enters the convention with the most pledged delegates with its preferred candidate. Such an act would undoubtedly be seen as antidemocratic and the party would incur a rapid exodus of angry voters. As such, superdelegates only exist as a symbol – a reminder to the party’s more ideological members of who truly controls the party. They serve no practical purpose other than to confer current and former elected officials or high-level party members with status. This status comes at the price of the controversies and legitimacy issues that emerged during the 2008 and 2016 primaries that, along with other issues like data fetishism, continue to threaten the party’s viability.
Data Fetishism
After Hillary Clinton’s 2016 loss to Donald Trump, party elites and pundits published numerous reflections, analysis, and postmortems about how the party lost a general election to an inexperienced demagogue who bucked his own party’s long-standing positions on issues related to trade, foreign policy, and social security, while simultaneously making platitudes to long-standing rival nations and fending off several accusations of sexual misconduct. Rampant finger-pointing emerged between the Clinton campaign who blamed the quality of the party’s data and the DNC’s current and former data gurus who defended the data and accused the Clinton team of building inferior statistical models with said data. Both these arguments depict a party engaging in data fetishism, in which data is seen as inherently objective and true. If the Clinton campaign did indeed ignore states like Wisconsin and Michigan[5], two states that proved decisive for Trump, because her analytics saw it as waste of resources, it is irrelevant whether the raw data was the problem or whether the statistical models left out key variables. The problem is that predictive analytics should not have been driving that type of decision. Predictive analytics is a misnomer, particularly in the context of a high-risk scenario like a nationwide election involving a high-variable population like the U.S. electorate.
Predictive analytics – in which statistics are combined with large data sets, machine learning techniques, and algorithms – has exploded in the last decade across a variety of fields including marketing, financial services, and risk management. According to the Harvard Business Review, predictive analytics has brought about significant improvements in fields like supply chain management and marketing.[6] But U.S. politics is not a supply chain. A failure in predictive analytics might lead an auto manufacturer to produce an excess of vehicles that need to be sold at cost. The losers of elections simply go home and their constituents pay the price.
In a 2012 article in Wired, statistician Nassim N. Taleb, the author of the book Black Swan[7], a popular work on predictive failure, argued that “big data, and the many variables involved, can easily lead to spurious relationships and prediction errors.”[8] The Obama campaign built a huge predictive analytics system that was used to drive fundraising targets, decide where to purchase advertising, and predict turnout. It is this last category, predicting turnout, that failed Clinton and, based on Taleb’s concerns, very well could have failed Obama. And yet, the Party’s failings have done little to change the minds of the technocratic political class.
While the DNC and the Clinton campaign have looked to blame each other’s data processes for the recent electoral failings, the Democratic consultant class has been busy writing think-pieces to shore up support for data-driven campaigns that might have diminished as a result of Clinton’s unexpected loss. A recent piece entitled “An Open Letter to DNC Chair Candidates on the Future of Tech for Our Party,” by two former DNC Directors of Technology who have since moved into the consultant class, ensures party leaders of the value of the party’s digital infrastructure and recommends the party “take a leadership role in figuring out the next generation of opinion research so Democrats at all levels have accurate intelligence on what voters are thinking.”[9] They also recommend the creation of a Democratic Tech Fellows program and an increased focus on promoting the party’s smartphone apps.[10] From their data fetishist perspective, data-related failures in the 2016 election were the failures of technological systems, and the only solutions to them are better technological systems.
A review of the DNC’s 2017 job postings makes it clear they are continuing to double down on data. The job description for a ‘Data Science Lead’ position states the Democratic Tech department’s mission “to optimize the future of the country.”[11] A recent interview with Raffi Krikorian, the party’s new Chief Technology Officer who was hired away from Uber, touts the party’s new investments in real-time analytics, artificial intelligence, machine learning, and data science tools.[12] The party is pruning the tree of its technocratic practices without realizing the roots are weak and rotted out. Its leaders are blinded by their continued faith in engineering and complex systems. As such, all of their solutions to the party’s failures involve creating systems and maintaining the philosophy that data must be the nexus of political campaigns. One of the primary ways in which they evangelize this philosophy nationally is via platforms.
Platforms
The processes of identifying and reaching out to voters via phone calls and door-to-door canvassing were historically overseen by the state party in concert with candidate campaigns. This was not without its problems. Campaign staffers who worked for John Kerry’s presidential campaign in 2004 explained how volunteers and staffers canvassing on election day were going to every house on the block attempting to turn out Democrats but likely turning out many Republicans in the process. By contrast, the turnout machine engineered that year by Republican strategist Karl Rove for George W. Bush and the Republican Party was lauded by pundits as the most sophisticated in modern history. As a response to this, the DNC adopted a nationwide data regime in 2005 that centralized in Washington many of the field practices and voter data previously left to state parties. The cornerstone of this project was contracting with the Voter Activation Network (now NGP VAN) to create Votebuilder, a nationwide voter database for tracking individual contacts with voters and generating maps and lists for contacting and turning out voters.
While it is without question that Votebuilder has resulted in more efficient campaigns, any digital platform “imposes specific constraints on the communication process.”[13] In the case of VoteBuilder, its affordances and its limitations are one in the same. It is inherently about narrowing the electorate and placing voters into baskets. One basket might contain seniors likely to be concerned about Medicare, while another contains millennials with concerns about student loans. Using these baskets in messaging campaigns can help campaigns connect with voters. But there are intra-party power dynamics in the process of organizing voters into baskets. A basket containing people who supported Obama in 2008 but say they plan to vote for Romney in 2012 is placed aside by the campaign, and regardless of whether people in it might still support other Democrats on the ticket, its voters are likely to be excluded from voter turnout efforts. As such, the path to victory of down-ballot candidates is secondary to the needs of the candidate at the top of the ticket whose campaign generally controls the “coordinated campaign,” a partnership of all party campaigns that come together for a large turnout effort in the final days of an election cycle.
Persuasion campaigns are also problematic. Campaign staffers and volunteers spend six months prior to elections going door-to-door querying voters about who they plan to support and what issues are of concern to them. Hesitant voters are re-contacted a few weeks before the election with a talking point ready for that voter’s previously stated concern. If the voter says they plan to support the candidate in question, the database is updated and the campaign’s data team sees this movement as a successful persuasion. Whether this voter would have ended up supporting the Democratic on their own is irrelevant. They have been “persuaded” by the low-level campaign staffer and are included in the aggregate metrics that campaign managers obsess over. If their stated concern was not one of the eight to ten pre-determined choices available to the field organizer, it likely never makes its way to anyone in the campaign concerned with public opinion or messaging. Even if the staffer writes it down, at best it ends up in the database’s “notes” section which is rarely, if ever, looked at in any aggregate way. This is both a problem of the platform – in that information is lost in the process of narrowing complex voter opinions into simple categories – and another example of data fetishism as managers ignore data that cannot easily be quantified and plugged into complex models. The task-centered aspects of the platform (generating phone calls and door-to-door canvasses), combined with the post-campaign exodus of the staffers who manage it, also fail to produce creative leaders ready to build and maintain the party between elections.
Party Building
Though the public shock of Clinton’s loss brought a barrage of scrutiny to the strategies of both the Clinton campaign and the Democratic Party writ large, its electoral problems pre-dated the 2016 campaign. The previous six years (2010 – 2015) saw the party take repeated losses in Congress while its control of both state legislatures and governorships reached their lowest points in almost 100 years. The thousands of volunteers that worked on Obama’s 2008 and 2012 campaigns were not melded into any kind of party structure between elections. The number of registered Democrats nationwide fell from 38% to 36% during Obama’s first year in office, reaching a low of 32% in 2015[14]. Effectively, Obama was able to win voters but unable to develop partisans.
Recent articles by a key member of Obama’s 2008 new media team[15] – tasked with creating digital content to tell the stories of Obama and his supporters – and one of his 2008 senior policy advisors[16] explain how Obama’s grassroots organization was “mothballed”[17] after the 2008 election, squandering opportunities for a greater digital-based movement to rally public support for Obama’s agenda. For them, the fact that the movement-style enthusiasm of Obama’s supporters petered out in the absence of any post-campaign direction is one of the great failures of Obama and the DNC under his leadership. It is also completely in line with the technocratic ideology of the party and its embrace of platforms.
In his recent works on “platform capitalism”, Nick Srnicek argues that “data is the basic resource that drives…firms, and it is data that gives them their advantage over competitors. Platforms…are designed as a mechanism for extracting and using that data.”[18] Applying this to the structure of political campaigns, volunteers are not people with individual identities and creativity, but rather an extension of the platform itself. Volunteers are widgets – applications or components of an interface that enable a user to perform or access a service[19] – that provide access to additional volunteers (widgets) via their personal networks and access to voters (data) via phone calls and door-to-door canvassing. In her book, From Widgets to Digits, Katherine Stone observed similar dynamics in the workplace as employers shift from a focus on employee longevity to one of employee flexibility, discouraging long-term attachment between employees and firms[20]. While firms might realize some tangible benefits from an increase in short-term or temporary employees, parties do not. They must seize the opportunity to channel some of the excitement and enthusiasm of campaign volunteers into building and expanding the local party structure.
Following the 2016 election, such an opportunity emerged around the progressive activists who are often referred to as the “Sanders Wing” of the Democratic Party. This group maintains an enthusiasm for Sanders and progressive politics but many of its members remain angry at the DNC for what they see as the rigging of the 2016 primary campaign toward Hillary Clinton. Limited efforts to repair the relationship between the party and Sanders’ supporters have included a “Unity Tour,” in which DNC Chair Tom Perez and Sanders toured the country doing rallies together, and a “Unity Commission” including representatives of both campaigns, set up by the national party to address his supporters’ grievances and recommend changes to party processes. This commission recently released recommendations about voter registration for primaries and transparency in elections for state party leadership, as well as a 60% reduction in superdelegates[21]. And yet, the Party has taken steps to neuter the efficacy of these projects.
In April, Perez dumped the bulk of Sanders’ allies from party leadership positions. The rules and bylaws committee that will eventually decide whether or not the recommendations of the Unity Commission are adopted is stacked with long-time party insiders who will likely be hesitant to adopt the commission’s reforms. Additionally, a recent report in Wired revealed that left-leaning Democrats, who have come together under the moniker “Justice Democrats” to challenge incumbent centrist Democrats in primary elections, have been refused access to the Party’s voter database[22]. An Illinois state party official was quoted as saying, “Why would you want to give it to outsiders who may or may not be Democrats?”[23] Democrats (and Republicans) have a long history of withholding resources from potential primary challengers that long predates voter databases[24]. Doing so in this case makes it clear that while party leaders may want Sanders’ supporters in their coalition, they want it on their terms only.
Towards a More Democratic Party
The Democratic Party needs urgently to exorcise itself of the technocratic culture that is creating blind spots in its perceptions of voters, discouraging new participation in local and statewide parties, excluding minority views, undermining its legitimacy, and losing elections. It needs to embrace contention and pluralism as inherent and positive aspects of democracy rather than as problems to be obscured or controlled.
Over the course of the last decade, the party has elevated predictive analytics and data science to the point where it plays an unsustainable role in party decision-making. Like the supposed whiz kids who crashed the financial markets in 2008, the Democratic Party’s data consultants have a faith in these systems that, given their continued evangelism even after the 2016 election, can only be described as hubris. The use of predictive analytics should be relegated to activities, like fundraising and making ad buys, where the level of risk is appropriately balanced with the potential fallibility of these practices. Predictive analytics should not be used to make large-scale strategic decisions. Critical safeguards, such as training of staff in the limitations of these systems, should be implemented to protect against future over-reliance on analytics.
The Democratic Party has generally failed at motivating its voting base in the absence of presidential elections. Votebuilder, the party’s platform for its voter contact programs, should not be the be-all and end-all of a party’s ground game. Throughout campaigns and between election cycles, it should be replaced with a robust effort to develop community leaders who engage with voters about local issues, a practice that is simply impossible to do well using campaign organizers who parachute into localities for six months and leave after election day. The party must encourage volunteers to develop into party activists rather than squandering their presence during election cycles by thinking of them only as widgets that can be set aside between campaigns. Following campaigns that draw new volunteer talent, the party should reach out to its newly committed volunteers with opportunities to get involved locally or to participate in online activities that expand their experience with the party. Encouraging nascent leaders to participate in such projects would help them develop as party members while also making sure the party is in touch with the views and priorities of its newest partisans.
This investment in party building should not solely be a conduit from localities back to the national party. In response to a long overdue need to build excitement and engagement for the party as a community institution rather than an organization that exists solely for its candidates and goes largely dormant between elections, the party should connect with nationwide and local interest groups and social movement organizations both outside of and during election cycles. Democratic Party offices should be re-envisioned as community spaces where party members can connect with one another, and get information not only about the party and its candidates, but about events and opportunities with partner organizations like Planned Parenthood Action, BlackPAC, Indivisible, and Everytown for Gun Safety. This will not be without contention. An organization as large as the Democratic Party has many entrenched interests that may be resistant to a more open and inclusive party. However, ignoring battles between party interests that are already playing out in party conventions and in the media will not make them go away. Allowing them to play out at the local level may promote greater understanding between partisans and prevent issues from lying dormant until large-scale events like the presidential nomination process expose them.
The Democratic Party must also reduce its control over the primary process at the presidential level. The party leaders on the rules and bylaws committee need to admit that it will never be politically feasible to substitute their will for that of the millions of voters who cast ballots during the presidential primaries. Also, while there are no superdelegates in congressional, state and local elections, the party practice of protecting incumbents from intraparty challenges by providing them with resources – endorsements, money, and data access – amounts to the same problem and should be stopped.
Engagement with voters in primaries will help counter this trend and keep people engaged. Allowing challengers access to Votebuilder will engender goodwill between intra-party factions, and, despite handwringing from elites, will likely do little to counter the built-in advantages of incumbency. Primaries are one of the very few formal processes through which partisans express their values and exert themselves democratically in relation to their own party. Given the systematic entrenchment of the two parties and the related lack of viability of third parties in the U.S., parties have a public interest responsibility to be open, transparent, and malleable to the changing concerns of their constituents. Intra-party primary elections will act as a way to keep the views of representatives in line with those of their constituents. If incumbents are in line with the views of the voters they represent they will be much less likely to face such a challenge. There is indeed precedent for a party that both has formal factions and functions properly. In the 1930s and 40s, some state parties had formal subgroups within the party that ran their own slates of candidates and pushed the party on specific legislation. Some current state parties already have such groups. For example, California has an active Progressive Caucus that, despite lacking a majority of state party members, earlier this year drafted and pushed the state party to adopt a resolution to outlaw “money bail” in the state court system due to its disproportionate impact on the poor. A subsequent bill is currently being debated in the California Legislature.
Many of these proposals require reconsidering the structure and practices of the Democratic Party. The party’s over-reliance on consultants and technocrats over the last several decades has created a chasm between the party’s leadership and its voters, whose true consequences have only begun to be revealed via the 2016 election. If the party wants to fight for the rights of those it claims to represent, it needs to abolish systems that exist to protect itself from its own voters, place leadership in the hands of diverse partisans at the state and local level, and create open systems that allow competing priorities to be represented by its members. Most importantly, it needs to reject its veneration of the technological and embrace democracy for what it is – collaborative, contentious, and unpredictable.
Notes
[1] Daniel Kreiss and Christopher Jasinski, “The Tech Industry Meets Presidential Politics: Explaining the Democratic Party’s Technological Advantage in Electoral Campaigning, 2004–2012,” Political Communication 33, no. 4 (2016): 544–562.
[2] Trevor J. Barnes, “Big Data, Little History,” Dialogues in Human Geography 3, no. 3 (2013): 297–302.
[3] David Kirkpatrick, The Facebook Effect: The Inside Story of the Company That Is Connecting the World (Simon and Schuster, 2011).
[4] Frank Fischer, Citizens, Experts, and the Environment: The Politics of Local Knowledge (Duke University Press, 2000).
[5] John Wagner, “Clinton’s Data-Driven Campaign Relied Heavily on an Algorithm Named Ada. What Didn’t She See?,” Washington Post, November 9, 2016, sec. Post Politics, https://www.washingtonpost.com/news/post-politics/wp/2016/11/09/clintons-data-driven-campaign-relied-heavily-on-an-algorithm-named-ada-what-didnt-she-see/.
[6] Jacob LaRiviere et al., “Where Predictive Analytics Is Having the Biggest Impact,” Harvard Business Review (blog), May 25, 2016, https://hbr.org/2016/05/where-predictive-analytics-is-having-the-biggest-impact.
[7] Nassim Taleb, The Black Swan : The Impact of the Highly Improbable /, 1st ed. (New York : Random House, c2007.).
[8] Nassim N. Taleb, “Beware the Big Errors of ‘Big Data,’” Wired (blog), February 8, 2013, https://www.wired.com/2013/02/big-data-means-big-errors-people/.
[9] Bryan Whitaker and Josh Hendler, “An Open Letter to DNC Chair Candidates on the Future of Tech for Our Party,” Medium (blog), December 1, 2016, https://medium.com/@joshhendler/an-open-letter-to-dnc-chair-candidates-on-the-future-of-tech-for-our-party-1afd14f77402.
[10] Whitaker and Hendler.
[11] “The Democratic Party – Data Science Lead,” Democrats.org | Work With Us, accessed December 19, 2017, https://jobs.lever.co/dnc/57223024-dc08-48fe-a912-39090665e277.
[12] “Raffi Krikorian: The Man with a Plan to Upgrade the Democrats,” The Armenian Mirror-Spectator (blog), December 14, 2017, https://mirrorspectator.com/2017/12/14/raffi-krikorian-man-plan-upgrade-democrats/.
[13] Ganaele Langlois et al., “Networked Publics: The Double Articulation of Code and Politics on Facebook,” Canadian Journal of Communication 34, no. 3 (2009): 415.
[14] “Party Identification Trends, 1992-2016,” Pew Research Center for the People and the Press (blog), September 13, 2016, http://www.people-press.org/2016/09/13/party-identification-trends-1992-2016/.
[15] Kate Albright-Hanna, “How the Cool Kids Killed Obama’s Grassroots Movement | Civicist,” Civic Hall, February 21, 2017, https://civichall.org/civicist/cool-kids-killed-obamas-grassroots-movement/.
[16] Micah L. Sifry, “Obama’s Lost Army | New Republic,” The New Republic (blog), February 9, 2017, https://newrepublic.com/article/140245/obamas-lost-army-inside-fall-grassroots-machine.
[17] Sifry.
[18] Nick Srnicek, “The Challenges of Platform Capitalism: Understanding the Logic of a New Business Model,” Juncture 23, no. 4 (March 1, 2017): 254–57, https://doi.org/10.1111/newe.12023.
[19] Roy Choudhury Barnali, Resource Optimization (UNESCO Publishing, 2015).
[20] Katherine VW Stone, From Widgets to Digits: Employment Regulation for the Changing Workplace (Cambridge University Press, 2004).
[21] Kevin Robillard, “DNC ‘unity’ Panel Recommends Huge Cut in Superdelegates – POLITICO,” Politico (blog), December 9, 2017, https://www.politico.com/story/2017/12/09/dnc-superdelegates-unity-commission-288634.
[22] Issie Lapowsky, “Progressive Democrats Fight For Access to the Party’s Voter Data,” Wired, December 11, 2017, https://www.wired.com/story/justice-democrats-denied-access-party-voter-data/.
[23] Lapowsky.
[24] Hans JG Hassell, “Party Control of Party Primaries: Party Influence in Nominations for the US Senate,” The Journal of Politics 78, no. 1 (2016): 75–87.
John Remensperger is a Doctoral Student at the Annenberg School for Communication at the University of Pennsylvania, United States. He researches the relationship between communication technologies and political institutions, including political parties, interest groups, and advocacy organizations. He researches how these groups use technology to organize their members, foster citizen engagement, and educate citizens to achieve policy change. His recent projects look at how organizational culture and intra-organizational dynamics impact, and are affected by, the adoption of new technologies.
Email: johnremensperger@gmail.com