UPDATE Nov. 16, 2020: This research has been published in the journal .
Just in time for Election Day, has examined how large social networks, such as blocs of U.S. voters, navigate making tough decisions鈥攅specially when time is of the essence.听
The team, which includes Zachary Kilpatrick from the University of baby直播app Boulder, has discovered that a mix of decision-making strategies might be the secret to success: Groups that are made up of both hasty and more deliberative choosers, he said, seem to have the best odds of coming to the right conclusions.听
鈥淲e wanted to look at how you can tune a group to have a good amount of these hasty decision-makers so that they鈥檙e providing a bump of information to the group, but not so much that they鈥檙e ruling the decisions of everyone else,鈥� said Kilpatrick, an associate professor in the Department of Applied Mathematics at CU Boulder.
Not everyone, in other words, makes decisions at the same speed鈥攁nd that might be a good thing. People who make decisions quickly could give their more sluggish peers the nudge they need to pick an option. Too many hasty decision makers, however, might drive communities to make the wrong choices.
Kilpatrick noted that the research, which will be published shortly听in the journal Physical Review Letters, doesn鈥檛 delve into the behavior of actual voters. Instead, it tackles a mathematical concept called 鈥渞ational agents鈥濃�攈ypothetical figures that always make the right selection based on the evidence they have on hand. 听
But, he said, scientists can still learn a lot by examining how these more idealized predictions line up鈥攐r don鈥檛鈥攚ith the decisions of flesh-and-blood humans.听
鈥淚f people are grossly deviating from this optimal behavior, which they can, then we can learn something about what biases they have,鈥� Kilpatrick said.听
Gathering evidence
In his latest study, Kilpatrick and his colleagues explored those kinds of decision-making processes by creating what was, essentially, an election-year environment鈥攅xcept this one was entirely captured in mathematical equations.
Using numerical simulations, the team set up scenarios under which groups of rational agents, which ranged in size from 100 agents to nearly 100,000, had to pick between two options. As the simulations progressed, the agents received data that pushed them toward one choice or another鈥攖hink of it like getting information by watching a presidential debate or reading the day鈥檚 news.听
鈥淲e have a continuous update, almost like a stock ticker, of their bias in favor of one choice or another,鈥� Kilpatrick said. 鈥淭hey only make a decision when that stock ticker hits an upper bound or a lower bound.鈥�
There was just one twist: Like voters, Kilpatrick鈥檚 agents did not all have the same threshold for making a choice. Some of them didn鈥檛 require a lot of information before landing on the option they preferred. Others needed to take their time and weigh a lot more evidence before deciding.听
The right balance
The balance between those two kinds of decision-making strategies seemed to make a big difference, Kilpatrick said.
Pretend your own social network contains a lot of hasty deciders. These seat-of-their-pants people are decisive, but that decisiveness can lead them, and the people who listen to them, into jumping into wrong choices. With too few hasty decision makers, however, the members of a group can get mired in indecision鈥攆ailing to make any kind of choice at all.听
鈥淭he hasty decision-makers are sniffing out the environment, and they pass that information on to the more deliberate decision makers,鈥� Kilpatrick said.
Still, he added, a few rash people can go a long way. The team calculated that networks that contain roughly 2 to 3% hasty decision makers were the most effective: These groups consistently made the best majority decisions in the least amount of time.听
鈥淲e were surprised that you only needed a small group of hasty decision makers to maximize this process of making decisions as a group,鈥� Kilpatrick said.听
He sees a lesson in his study for U.S. voters agonizing over their own upcoming choices. In other words, you always want to be wary of how or where the people in your social group are getting their information.
鈥淭here鈥檚 a lot of novices out there posting things on social media and they have not used a lot of evidence to come to their conclusions,鈥� Kilpatrick said. 鈥淧eople who listen to them value them as experts, and it can really cloud the decision making of a group.鈥�
Bhargav Karamched of Florida State University is the lead author of the new study. Coauthors include Kre拧imir Josi膰, William Ott and Megan Stickler of the University of Houston and Benjamin Lindner of Humboldt University in Germany.
