Below is an essay I submitted to the 2016/17 William F. Buckley, Jr. Ideas Forum and Contest. I was not a finalist, but I appreciated the opportunity to make the conservative argument for algorithmic democracy.
Algorithmic Democracy and a Free Society
by C.R. Krenn (email@example.com) (11/12/17)
In 1950, William F. Buckley Jr. was frustrated by the state of university education in general and by Yale education in particular, and he was not alone. Today, I am frustrated by the state of democratic government in general and by the U.S. Congress in particular, and I am not alone. Regularly, more than 75% of voters polled disapprove of how Congress is handling its job . We should be frustrated that the current polarization in Congress seems to prevent any meaningful discussion or compromise on issues such as immigration, campaign finance reform, tax reform, gun violence, or climate change. We should be heartened that, although the fraction of voters who disapprove of Congress’s job performance is growing, this has not yet significantly changed the voting rate . And, we should be relieved that, although the electorate is becoming more polarized, there is still significant overlap in political values between self-identified Democratic and Republican voters . We all can and should have a louder voice in our government and in the future of our country.
These data mean that yes, voters are frustrated and have become more polarized, but most have not given up on democracy. We can take advantage of these facts by harnessing voters’ faith in democracy and their willingness to compromise to pressure Democratic and Republican officials to more often act in the interests of their constituents. How could we do this? Imagine a secure online system where a large fraction of citizens ages 10 and up regularly complete a customized electronic survey of their political views, legislative goals, and positions on current issues, and where these citizens all choose decision-making delegates for national, state, and local issues . This system, let us call it “Algorithmic Democracy 2025”, would dynamically monitor agreement between its users and both their delegates and elected representatives. It would predict voter support for current legislation and even produce outlines of new legislation with the support of broad majorities of voters . As will become more clear below, supporting the development and implementation of Algorithmic Democracy 2025 is a policy solution that should significantly advance the ideals of the free society William F. Buckley, Jr. championed throughout his life.
This proposed algorithmic democracy system is a new hybrid of direct democracy and representative democracy. It is not direct democracy, where each eligible voter has a say on every major governmental decision. Direct democracy was used in Athens and is used still in some small Swiss towns, but modern voters do not have the time, interest, or ability for this level of participation. Algorithmic democracy is also not representative democracy, where groups of eligible voters in geographically distributed districts elect representatives by majority vote. Representative democracy has been reasonably effective for 200 years, but its current implementations, particularly in the U.S., fail to be responsive to large diverse populations of many tens of millions of voters. The perception and the reality is also that well-funded special interests are much better represented than average citizens.
Algorithmic democracy has applications at federal, state, and local levels, but implementation should begin at the local level using existing open-source software for polling and consensus building . Issues addressed could include recreational department budgeting or city planning for cell phone tower placement. As the software and interface are improved, algorithmic democracy could be used to develop and advocate city and state propositions. Algorithmic democracy is not intrinsically conservative or liberal, but its interface would enable both conservatives and liberals to take their arguments directly to voters. A poll from early 2017 found that 67% of Americans identify “big government” as a larger threat than big business or big labor, so it is likely that algorithmic democracy would also support movement towards more limited government and free enterprise . Algorithmic democracy could be supported by the government, but success is more likely with support from a grass roots organization. Modest funding would also accelerate development and implementation.
Implementation details of Algorithmic Democracy 2025 will affect the utility of this system greatly. Nonpartisan groups of policy experts and policy makers should be involved in poll creation, and it should include aspects of deliberative democracy which provides for education and debate as part of the polling process . Algorithmic Democracy 2025 needs to be secure against manipulation by hackers and activists from the far left and right, to protect minorities, to manage privacy concerns, and to maintain the principle of “one citizen one vote” . Any legislative proposals produced by Algorithmic Democracy 2025 need to be constrained by budget and by law. It also needs broad participation, needs to be easy and quick to use, and should offer obvious benefits to the user.
There are some existing electronic democracy systems in development, but they all suffer from some flaws. Two (brigade.com and countable.us) are private firms supported by venture capital and currently lack the transparency required for widespread trust and adoption, and they lack the motivation to respect the users’ time. Democracy.earth is a non-profit foundation and aims to develop a “decentralized democratic governance protocol”, but it has not yet considered the importance of broad deliberative discussion or the importance of collecting users’ opinions as well as votes . Algorithmic democracy does not need to be developed independently from these systems, but progress would likely be more rapid if it were.
William F. Buckley Jr. was a visionary conservative, but he was also a pragmatist who was willing to consider compromise on issues such as gun control, campaign finance reform, and climate change. Congress is not currently willing to make these essential compromises, but algorithmic democracy can make it more likely to do so. Algorithmic democracy can document differences between the voters’ will and government actions and thus hold our elected representatives more accountable to their constituents. It can enable better laws to be proposed and passed. It will pressure members of government to support the interests of all of their constituents, and it can responsibly shrink and decentralize government while maintaining sensible protections for its citizens’ lives, liberty and property. Algorithmic democracy deserves your support.
 See http://www.pollingreport.com/CongJob.htm (2016).
 U.S. voting rates are available at http://www.fairvote.org/voter_turnout (2016). And, the average turnout in presidential elections from 1916 to 2016 (55%) was identical to the turnout in 2016. (https://en.wikipedia.org/wiki/Voter_turnout_in_the_United_States_presidential_elections).
 Polarization data is from http://www.people-press.org/2014/06/12/political-polarization-in-the-american-public (2014).
 The age of ten corresponds to the age of the author’s son when the author started working on this proposal. It is negotiable but should be less than 18.
 Algorithmic prediction of voter support and outlines of new legislation would be provided by machine learning tools similar to those used by companies such as Netflix, Pandora, Spotify, and the data science startup Kaggle (http://kaggle.com/) that is owned by Google.
 See, for example, https://www.limesurvey.org/ and https://pol.is/.
 See http://news.gallup.com/poll/201629/americans-big-government-top-threat.aspx.
 A “nonpartisan” group can contain partisan members, but questions should be reviewed and corrected for bias. See (James S. Fishkin, “Democracy and Deliberation: New Directions for Democratic Reform”, Yale University Press, 1991 or http://cdd.stanford.edu/ for an introduction to and history of deliberative democracy polling.
 Necessary safeguards would include demographic weighting of self-selected populations, validation against random subsets of the U.S. Postal Service’s Delivery Sequence File, and periodic paper ballot validation.
 See http://brigade.com, http://countable.us, and http://Democracy.earth.