We have all been in meetings where we felt nothing was getting done. In the corporate world, the cost of inefficient meetings has been recognized. According to a recent CBS news report, professionals lose four work days each month in meetings and that out of 11 million meetings that occur in the U.S. every day, half the meeting time is actually wasted. There have been a lot of efforts to make meetings more productive including efficient meeting templates, ground rules for meetings (pdf) etc. However, a scientific, data-driven approach to understanding “soft” phenomenon such as a meeting has until now been rare.
A paper “Learning about Meetings” (pdf) by Been Kim and Cynthia Rudin at MIT is one of the first such efforts to employ a data-driven approach on the science of meetings (in this case, meetings that are held to arrive at a decision and not to brainstorm) to learn more about how meetings are conducted. Meetings are difficult to assess as there are social signals and interpersonal dynamics that are difficult to capture. Kim and Rudin, using AMI data show evidence that it is possible to automatically detect when during the meeting a key decision is taking place, that there are common patterns in the way social dialogue acts are interspersed throughout a meeting, that at the time key decisions are made, the amount of time left in the meeting can be predicted from the amount of time that has passed, and, finally, that it is often possible to predict whether a proposal during a meeting will be accepted or rejected based entirely on the language used by the speaker.
Some particularly interesting take-aways are:
- As it turns out, the important parts of the meeting are characterized mostly by information and information requests, and very few offers, rejections, or acceptances. They hypothesize that at the important parts of the meeting, when the decisions have been narrowed down and few choices remain, the meeting participants would like to ensure that they have all the relevant information necessary to make the decision, and that the outcome will fit within all of their constraints.
- They found, somewhat counter-intuitively, that the next judgment following a negative assessment is almost never a socially positive act. The converse is also true, that a socially positive act is rarely followed by a negative assessment. For example, a negative assessment such as “But I thought it was just completely pointless” will rarely be followed by a positive comment such as “Superb sketch by the way.”
- Some words are more persuasive than others. Non-technical words that fit this profile are yeah (specially following up on someone else’s comment), give, start (opportunity for group members to agree before starting), meeting (usually used to postpone things to another “meeting”), discuss (related to organizational suggestions), and find (need for more information).
This work has particular significance for those of us in the development field for the following reasons:
(1) Meetings are a key instrument through which we engage with stakeholders. For example, meetings between development organizations, between development organizations and governments, between civil society and development organizations, internal meetings, sector specific meetings and international forums are some of the ways we reach consensus and come to decisions that affect the development agenda. This kind of an evidence-based scientific study of meetings provides actionable information that can help us to improve the effectiveness of such meetings.
(2) Beyond meetings, communication is key for the development community. While there has been work on identifying characteristics of persuasive speech and discourse, such work has rarely been data-driven. This is the first case where machine learning (study of systems that can learn from data) techniques were used to learn persuasive words, to rank and to compare them to gain insights.
(3) This study helps us learn about the dynamics of interactions that span both social and work aspects in meetings. Meetings are never just about work. Learning about how the social aspects interact with the work aspects will help us understand the complex dynamics of meetings using a data driven approach.
(4) Just as customized medicine will one day help provide better care for millions of people, customized communication will help the development community communicate messages more effectively
This paper demonstrates that even though meetings involve complex social signals and interpersonal dynamics, it is possible to detect when key decisions are taking place, common patterns in meetings, predicting timing of meetings and whether proposals will be accepted or rejected. This paper used a scientific data driven approach to a nebulous subject like meetings and came up with actionable results. This gives us hope that a scientific, data-driven approach to understanding another larger area of “soft” phenomenon i.e. “development work” can provide enormous benefits. This blog is written to start a dialogue and we encourage comments about whether you agree and what we can do to make this happen.
Photo Courtesy: luigi diamanti/freedigitalphotos
Follow PublicSphereWB on Twitter