Three weeks ago, Sen. Elizabeth Warren (D-Mass.) experienced a frustration familiar to anyone who has been the oldest child at a wedding or family reunion. She got stuck at the kids’ table. By simple luck of the draw, she was placed in the first night of the Democratic primary debate, taking the lectern between Sen. Cory Booker (D-N.J.) and former representative Beto O’Rourke (D-Tex.), while her real peers, such as Sen. Bernie Sanders (I-Vt.), Sen. Kamala D. Harris (D-Calif.) and former vice president Joe Biden, ended up debating on the second night.

Warren’s status as the lone top-tier candidate probably hurt her in that debate. Almost 18.1 million Americans watched the second night of the debate, while only 15.3 million watched Warren’s appearance. Harris, not Warren, got the chance to press Biden on his long, baggage-heavy policy record, create a viral moment and surge in the polls. Warren instead had to try not to punch down too much as her lower-tier opponents vied with each other for a breakout moment.

This debate lineup wasn’t a game changer for Warren — she’s still a leading candidate. But the episode exposed a problem with the design of these early debates: They rely on randomness when they’re looking for fairness. And CNN, which is hosting the next debate, may end up unintentionally making the same mistake.

AD
AD

It’s not hard to look at the basic design of these debates and understand what the Democratic National Committee is doing. Candidates only have to hit one percent in three separate polls and raise money from 65,000 donors with at least 200 in 20 different states to qualify for the debate, and the DNC allows 20 candidates total on stage. The goal of these early debates is to be inclusive and give little-known candidates a chance to shine.

But things got hairy when NBC tried to determine who would be on which stage. The network divided the field into two groups: a top tier of eight candidates who were polling above two percent nationally and a bottom tier of 12 candidates who weren’t. Then it randomly assigned four candidates from the top tier and six candidates from the bottom tier to each debate night. NBC wanted two well-balanced, interesting nights of debates, and it used a random component to remove any inkling of bias from the final decision.

But introducing that random piece of the puzzle created new risks. Random processes, while fair in the long run, are often capricious and weird in the short term. To see what I mean, take a coin out of your wallet or purse, start flipping it and write down whether you get heads or tails. In the long run, the coin will come up heads about as often as it does tails. But the flips likely won’t alternate perfectly in a heads, tails, heads, tails pattern. You’ll probably see some stretches where the number of heads outnumbers tails or vice versa. The coin might give a fair, 50/50 outcome in the long run, but it can’t be trusted to do so in the short run.

AD
AD

If NBC was running thousands of consecutive Democratic debates, its process would be fair. Warren would have gotten her shots at Biden and Sanders. Some of the debates would have had the even spread NBC was originally going for, and other leading candidates would have ended up spending time on Beto O’Rourke’s Island of Misfit Toys. But NBC only had one random draw, and Warren happened to be disadvantaged by it.

CNN, the host of the next debate, has said it will use a random process to assign candidates to each stage, but it hasn’t released the details yet. It could improve on NBC’s process by dividing the candidates into smaller tiers — maybe by putting two of the top four candidates in the national polls in each night, dividing the next four, including O’Rourke, Booker, Pete Buttigieg and Andrew Yang randomly between the two nights and so on. It should resist the temptation to use a “pure” process where every name is drawn out of the same hat. An unrestricted process like that could lead to some seriously unbalanced, or just flat-out weird, divisions of candidates. And the best method of all might be non-random. CNN could simply apply an alternating pattern to the national polling data: Put the poll-leader (Biden) in the first night, put the second-place candidate (Warren) in the second night, put Sanders (who is polling third) on the first night, put Harris in the second, etc.

CNN will perform its draw live on air on Thursday night, and its random process may end up producing a reasonably fair, well-balanced pair of debates. I just wish that it wouldn’t leave something this important up to chance.

Read more:

AD
AD