Refactoring Democracy

Part 5: The Future of Representative Democracy

This is Part 5 of a six-part series on our Representation in the United States.

In Part 1, we reviewed the history of American apportionment of seats in the House of Representatives. In Part 2, we examined how the Act of Permanent Reapportionment dealt a fatal blow to American Democracy. In Part 3, we evaluated various methods for calculating optimal Representation. In Part 4, we covered key challenges to increasing our House size. Today, let’s imagine more innovative solutions to representation limitations.

Limitations Inherent in Representative Democracy

Even in the event we miraculously convince, sue or inspire state legislatures to coerce Congress to overcome any logistical, physical, and cost excuses shying them away from raising House membership above 435, we will continue to have an upper limit on representation in America. Applying the cube root law to bring us up to 675 seats appears the most realistic, effective and rational apportionment method, but our Representative Attention Threshold (R.A.T.) would only increase to 17–24 seconds per citizen per year. Still barely enough time to say hello to your Congressperson.

Even maxing out membership to our constitutional limit of 10,291 Members would only earn every American 4–6 minutes of dedicated attention per year with a Representative. We cannot get a whole lot argued or accomplished in 4–6 minutes each, that attention can still be bought by Special Interests, and more seats cannot guarantee officials will perfectly mirror our values. Even if we double the dedicated staff budget for each of the 10,291 Representatives, at best we’ll ever get two hours per year of attention by an office. You might be able to get to know a person in that time or even get one small thing accomplished, but that’s not enough for those of us who want to participate more than tweet, protest, write blog series or call an office full of staffers.

How then can we achieve optimal Representation in the United States?

Remember way back in part 1 when I alluded to Athenian direct democracy and despaired that today’s citizens had neither the time nor space nor voice to all participate at once in legislation?

Well, what if I told you that might actually be possible?

The House of Representatives was intended to serve as a proxy to citizens. In John Adams’s perfect world, “It should be in miniature an exact portrait of the people at large. It should think, feel, reason and act like them.” Bipartisan politicians hardly think, feel, reason and act like the rest of us. How can they? It takes certain skills, influence, and sleezeballing to navigate politics today. Representing Americans in the House is a career and a way of life very much separate from the careers and ways of life we each live.

The only way anyone could think, feel, reason and act just like us is if they are us. And I don’t just mean our neighbors. I mean direct extensions of us.

You know where this is going…don’t you?

…Clones.

Just kidding. But sort of.

Connecting Constituents to their Representatives

As illustrated above and in part 4, Representatives can only hire 18 staffers to help them do all the things including keep up with incoming feedback from constituents. Like their Representatives, these staffers only have so many hours in a day or year to help their boss serve the district. Asynchronous communication like letters and email help a bit, but they’re still cumbersome to weed through and respond to. The Internet empowered everyone with a voice, but it has not yet provided leaders with ears to effectively listen.

The Obama Administration offered a very primitive implementation of all that is possible herein by launching a White House Facebook Messenger Bot.

The bot allowed you to write the President through Facebook, but the whole experience felt like a one way street with no guarantee you’d ever really get read by anyone. The bot did not really engage with you and there’s no evidence the White House’s implementation of the bot applied any machine learning or analysis to understand the scope of things people were saying.

Imagine instead the ability to text, email, message, or chat with what felt like one of your Representative’s staffers at any time of day and get an immediate response from a bot driven by artificial intelligence.

For now, let’s call this A.I. “Sam.”

You could ask Sam questions, object to a bill, raise an issue, or learn more about the Member representing you. In return, Sam could answer your questions, ask you questions of its own, share pertinent news on issues you care about, solicit your feedback on a bill, offer legislative reports, and learn as much as it could about the needs of everyone in your district. Sam would share all that it learns at scale anonymously for all to see.

I could see Sam sending me lightweight polls on a regular basis to ask me how I feel on issues, especially issues pertaining to bills ready for vote. In the short period I tried online dating, I found OkCupid’s multiple choice match questions an especially smooth and even addicting way to help the software help me find my honest place in that community. The added dimension of “importance” to a question could help Representatives better understand what issues we actually care about beyond issues championed by extreme partisan agendas. If we could tell Sam how little or often we wanted these questions asked, we could participate as much or as little as we like.

OkCupid’s match question format

To hold our Representatives accountable, many of these polls could be triggered by other constituent moderators or automatically by the bot itself in anticipation of a legislative agenda. Results from these mini polls could then be directly compared to the way a Representative voted on an issue, demonstrating clearly whether the Representative voted in line with values of the district or not. More than any other mechanism currently available, the juxtaposition of measured constituent values and Representative activity would offer voters context and visibility into their Rep’s performance.

Depending on user adoption, Representatives could not afford to avoid Sam for long. Candidates running for office could use Sam to their advantage in elections and incumbents would need to fight fire with fire. Representatives and staffers could launch their own polls, review high level summaries of constituent feedback, collect campaign contributions through Sam, and even jump in to engage directly with voters in Sam’s place. Before long, our elected officials would need to fall in line with the will of their constituents lest a vivid rift appear and voters decide next election to vote another way.

A platform leveraging artificial intelligence and machine learning delivered in a conversational format could help the public and officials alike connect like never before. Through such a powerful implementation, we could have doors to a town hall open 24/7 and participate from anywhere in the world.

Crowd-Sourcing Legislation

As the fields of machine learning and artificial intelligence evolve, Sam could help facilitate the process of listening to the concerns of the people, check with others to see if they share the concerns, solicit solutions to the problems raised, collect feedback on the best solutions proposed, and come out the other end with brief proposals for legislation if not bills themselves. It would be Sam’s job to delicately figure out whether points people make are valid and dynamically spread ideas that gain the most traction.

Depending on the complexity of an issue, Sam could hear about a problem raised at breakfast, confirm that many others share the problem by lunch, and triangulate the favorite proposed solution to the issue by dinner. Before everyone goes to bed, a majority consensus could be reached on how best to deal with a problem specifically defined and present a bill for vote before morning. All of this could be handled through a system of conversational multiple choice questions dynamically generated and distributed by an A.I.

All of it would be made possible by an artificial intelligence unbounded by time, space, or the conventional one-on-one conversations required in political collaboration. How freaking cool would that be?

The House of A.I. Representatives

Whoa, whoa, whoa. That escalated quickly! A.I. House of Representatives?!

Hear me out before you reach for your pitchfork.

If, and only if, (but really “when”) our A.I. becomes powerful enough to listen to the will of the people, draft legislation, and hold Representatives accountable to voting along constituent lines to the point where Members become purely ornamental, someone will raise their hand and ask why we still need human Representatives at all. By then, bots will be driving our cars, flying our planes, preparing our meals, and dramatically maximizing the productivity of the human race. We’ll be addicted to bots. And it will only make sense that we delegate the complex task of listening to crowds and making sense of all that noise to artificial intelligence.

More than just a tool to make sense of things, bot Representatives have many advantages over human Representatives:

  • Bots don’t need to fit you into a packed schedule to pay attention. They are on call anytime of day or night for however long you want and on your own terms. The binding limit of a Representative Attention Threshold (R.A.T.) goes out the window when a bot can represent you and everyone else, simultaneously, full time and anytime on demand.
  • Bots don’t need to take recesses, sleep, or go home on weekends. Bots have no home, do not need to meet in Washington, and can always be available to constituents in their district or abroad. Instead of meeting less than 140 days per session, bots can legislate around the clock 365.2422 days a year and never stop working for the American people.
  • Bots do not have scarce time or attention that can be bought by lobbyists or donors. They do not need to fundraise or take a salary. If engineered openly and securely, they cannot be bribed or corrupted. Lobbyists will need to lobby the American people instead of a few politicians or make way entirely for brands acting more responsibly.
  • Bots behave. Bots do not have secrets to keep, tax returns, conflicting interests, dick pics, or scandals. Bots can be held accountable through algorithms and code. If we’re not happy with a bot’s performance, we can vote at any time to change its source code to fix it.
  • Bots have no need to brand themselves by a political party. Since we will no longer need a shorthand to understand what a candidate stands for since the “candidate” would dynamically conform to our values, we no longer have benefit to identify by political parties. Bots will take shape to embody the values and issues a local group of people care about. Our political identities will stem from and go by names of our communities.
  • Bots have no single socioeconomic, racial or sexual identity. Bots will be a reflective mishmash of all participating constituents and their values, conditions, and dreams. Bots can fractionally represent all demographics in their constituency no matter how big or small.
  • Bots can be fair. Bots could leverage participation data through unbiased algorithms to automatically gerrymander contiguous districts. Bots cannot be selective or prejudicial to whom they receive or listen to. No bot will have any advantages over another, more resources to leverage than another, or more experience than another.

The list goes on and on. Without question, we will eventually delegate our Representation to bots. It only makes sense and it’s only a matter of time.

Many Open Questions

This little bit of science fiction opens a ton of questions I’ll only be able to scratch the surface on today, but now’s the time for us to ask before the system takes shape on its own or by someone else’s hands without our input.

Couldn’t it get hacked?

Let’s just get this question out of the way first. Where there’s a will, there’s a way…especially with technology. Nothing is impregnable. That said, we’re starting to see a ton of revolutionary and transformative technologies taking shape like the blockchain that may help us address the issue of keeping our governing code bases secure. I’m not a computer engineer or security expert, but I can say that the American people will not endorse such an impactful technology unless it was built by reputable computer engineers and security experts we all trust. Moreover, we should plant bounties buried in the code for hackers to chase so that people have incentives to regularly test the system. With any luck, the good guys will find our vulnerabilities first.

Would A.I. Representatives even be constitutional?

Well, no. While the Constitution does not explicitly say “humans” must serve as Members, clause 2 of Article I Section 2 outlines age, citizenship, and residential requirements to serve which imply humanity. Article I Section 6 also outlines compensation, which our bots won’t need. Our Constitution would need to be thoroughly vetted and amended to address the implications of a House full of bots. Fortunately for those of you vehemently opposed to this idea, it will take one hell of a lot more influence on Congress or state legislatures than the issue of House size to affect change herein.

One big bot or many?

Every community has different perspectives that deserve to be acknowledged in the national debate. The entire catalyst for this blog series revolves around how few perspectives we allow on stage at the fault of our system today. Therefore, we cannot rationally roll up all national debate into one lone bot. We would need to break the United States into districts (as our forefathers envisioned) associated with diversified bots (probably not as our forefathers envisioned) that each develop unique identities and assumptions based on their constituents. Without knowing exactly how it will work yet, each district’s unique combination of answers to a myriad of variables could condition each bot to ask smarter questions, make assumptions, and reach conclusions faster for a given group of people. On one level, the bot could have a personalized relationship with you. On a higher level, the bot will have a communitized relationship with the community.

The tricky question remains the same: how many people should one bot represent? An ideal solution would keep the ratio narrow enough to reflect American diversity most accurately while wide enough to ensure enough regular human participation to function properly. Without amending the Constitution, the answer would be 30,000 people. If we need to amend the Constitution anyway, let’s consider rolling groups of 22,500 up to one bot according to the more defensible Dunbar Squared method.

What about the Senate? The Presidency? Supremes? All offices?

I have more faith in machines than most people, but I also know how the sausage is made and sometimes it’s not pretty. We need a human element to counterbalance the digital element, just like our founders envisioned an upper house to balance out a lower house. Remember that our House of Representatives is the “People’s House,” designed explicitly by our forefathers to proxy the American people. The offices of the Senate, President, and Supreme Court? They were designed to hold the People’s House accountable. While I cannot promise you that our descendants will eventually prefer all of these offices be filled by artificial intelligence, we should all see value in some human element holding our bots accountable. I for one vote to keep those 100 humans in the Senate around to keep the hundreds of millions of us in check. They may very well serve as our sole line of defense against runaway populism onset by crafty bots.


Will we be represented by artificial intelligence in our lifetimes? It really depends on how badly the American people want a solution to the problems explored in this blog series. Will we see an app or platform like Sam that helps leaders connect with their constituents? You bet your ass. When corporate America eventually figures it out to serve their mighty leaders, we can apply the same learnings to the public sphere. If not before.

In the next and final post of the series, we’ll summarize all that we’ve learned about modern representation in the United States and evaluate possible next steps for building a future we want to see in our own nation.

This is Part 5 of a six-part series on our representation in the United States.

Next: Part 6: Long Story Short