3 times more ideas: start the brainstorm alone, then meet

I run hackathons and AI workshops, and the default opening is always the same: eight people, one whiteboard, a facilitator asking the room for ideas. Send those eight people to separate tables with the same question for ten minutes and they come back with more ideas, and better ones. The meta-analytic evidence has pointed that way for 35 years, and the mechanism is as dull as waiting for your turn to speak.

The same people score higher apart than together.

Brian Mullen, Craig Johnson and Eduardo Salas pooled 20 studies from 18 articles, 2,577 participants in 844 groups ¹. Each study compared a brainstorming group of n people against a nominal group: the same n people working alone, their lists merged afterwards with duplicates struck out. Nominal groups won on the number of non-redundant ideas (r = .57) and on idea quality (r = .56). The loss grew with group size, grew when an experimenter sat in the room, and grew when people spoke their ideas aloud rather than writing them. The authors call the popularity of the technique "unequivocally and substantively misguided".

Turn-taking does the damage.

Michael Diehl and Wolfgang Stroebe ran four experiments to separate three candidate causes ²: free riding, where you let the others carry the session; evaluation apprehension, where you hold back a weak idea; and production blocking, where one person talks at a time so everyone else parks an idea in memory or loses it. Personal assessment instructions raised output, so motivation plays a part. The group deficit survived conditions that removed any temptation to free ride. Blocking carried most of the loss.

That diagnosis points at the fix: give every person their own channel. Mullen's moderator analysis agrees from the other side, with written contributions losing less ground than spoken ones ¹. Darleen DeRosa, Donald Smith and Donald Hantula meta-analyzed the electronic brainstorming literature and found a crossover at group size: large groups typing in parallel beat nominal groups, while small nominal groups beat electronic ones ³.

Solo first, then together, triples the rate.

Karan Girotra, Christian Terwiesch and Karl Ulrich compared two ways of organising four-person teams on a product idea challenge ⁴. The team structure gave each group 30 minutes together. The hybrid structure gave each person 10 minutes alone first, then brought the group together. Hybrid teams produced about three times as many ideas per unit of time, those ideas scored higher on average quality, and their best idea beat the best idea of the team structure. They also judged their own output more accurately. One finding cuts against standard facilitation advice: teams that built on each other's ideas generated no more ideas, and the ideas built on earlier ones were no better.

Paul Paulus and Huei-Chuan Yang tested a quiet version with four-person groups ⁵. Each person wrote an idea on a slip, passed it to the right, read what arrived, added an idea, passed again. Over 15 minutes those groups produced 40% more ideas than nominal groups. In a second session where everyone worked alone, the people who had traded slips earlier produced 90% more than those who had not. Reading other people's ideas helps. Waiting for the floor does not.

Generating more does not buy you a better pick.

Eric Rietzschel, Bernard Nijstad and Wolfgang Stroebe had interactive and nominal groups both generate ideas and then select one ⁶. Nominal groups generated more ideas, and more original ones, though less feasible. On the ideas each group ended up choosing, quality did not differ, and selection landed no better than chance.

Brian Johnson and Christopher D'Lauro found the pattern behind that. Groups asked for their best idea picked one generated early, at a mean of 117 seconds into the session, and those picks rated feasible but unoriginal ⁷. Asked instead for the most original idea, groups reached deeper into the session, to a mean of 328 seconds. Groups that generated more ideas did not select better ones.

The session has two failure points, and a solo opening fixes only the first. Scoring needs its own design.

The room will rate the worse format higher.

Wolfgang Stroebe, Michael Diehl and Georgios Abakoumkin handed participants the pooled idea list and asked which ideas were theirs ⁸. People who had brainstormed in groups were more satisfied with their performance than people who had worked alone, while producing fewer ideas, and they could not reliably separate their own contributions from the ones they had heard. Hearing 40 ideas feels like having had 40 ideas.

I say this to a room before the first silent block, because ten minutes of silence among eight people reads as a facilitation failure. The complaint about energy arrives on schedule. Run the format anyway and compare the lists.

The protocol.

1. Open with 10 minutes of silence. One question on screen, everyone writing alone, nobody talking. That is the ratio Girotra's hybrid teams used, 10 minutes solo inside a 30-minute slot ⁴.
2. Collect in writing before anyone speaks. Each person submits their own list to a shared doc or a stack of cards. The first idea spoken aloud anchors the next ten; an idea collected first survives on its own.
3. Cap the discussing group at 4, or move it to keyboards. Above that size, parallel typing beats a room where one voice holds the channel ³.
4. Pass written ideas instead of discussing them. Rotate slips every 3 minutes: read, add one, pass right ⁵.
5. Send people back to their desks for a second solo round. Paulus and Yang measured 90% more ideas in the individual session that followed an exchange ⁵.
6. Score every idea blind, on 2 separate criteria. Strip the names, then rate feasibility and originality apart. Groups told to find "the best" fall back on the early, safe one ⁷.
7. Require one shortlist slot from the last third of the session. Original ideas showed up at a mean of 328 seconds against 117 for the feasible ones ⁷.
8. Delete "build on each other's ideas" from the brief. Teams that did it produced no more ideas, and the derived ideas scored no better ⁴.

For your next session: put the question in the calendar invite, book 60 minutes, and spend the first 10 with nobody talking.

Sources.

1. Mullen, B., Johnson, C., & Salas, E. (1991). Productivity loss in brainstorming groups: A meta-analytic integration. Basic and Applied Social Psychology, 12(1), 3-23. doi.org/10.1207/s15324834basp1201_1
2. Diehl, M., & Stroebe, W. (1987). Productivity loss in brainstorming groups: Toward the solution of a riddle. Journal of Personality and Social Psychology, 53(3), 497-509. doi.org/10.1037/0022-3514.53.3.497
3. DeRosa, D. M., Smith, C. L., & Hantula, D. A. (2007). The medium matters: Mining the long-promised merit of group interaction in creative idea generation tasks in a meta-analysis of the electronic group brainstorming literature. Computers in Human Behavior, 23(3), 1549-1581. sciencedirect.com/science/article/abs/pii/S0747563205000506
4. Girotra, K., Terwiesch, C., & Ulrich, K. T. (2010). Idea generation and the quality of the best idea. Management Science, 56(4), 591-605. doi.org/10.1287/mnsc.1100.1172
5. Paulus, P. B., & Yang, H.-C. (2000). Idea generation in groups: A basis for creativity in organizations. Organizational Behavior and Human Decision Processes, 82(1), 76-87. doi.org/10.1006/obhd.2000.2888
6. Rietzschel, E. F., Nijstad, B. A., & Stroebe, W. (2006). Productivity is not enough: A comparison of interactive and nominal brainstorming groups on idea generation and selection. Journal of Experimental Social Psychology, 42(2), 244-251. doi.org/10.1016/j.jesp.2005.04.005
7. Johnson, B. R., & D'Lauro, C. J. (2018). After brainstorming, groups select an early generated idea as their best idea. Small Group Research, 49(2), 177-194. doi.org/10.1177/1046496417720285
8. Stroebe, W., Diehl, M., & Abakoumkin, G. (1992). The illusion of group effectivity. Personality and Social Psychology Bulletin, 18(5), 643-650. doi.org/10.1177/0146167292185015