Tom Wujec: Surprising lessons about team performance from the marshmallow challenge

Tom Wujec: Build a tower, build a team | Video on TED.com or on The Marshmallow Challenge.
(video, 7 min)
Tom Wujec ran more than 70 experiments where teams of people competed with each other in building the tallest structure made of simple objects. Here are some of the surprising results.
Recent kindergarten graduates perform better than average adults, probably because kids make multiple successive prototypes, while adults design and build a single structure.
Recent business school graduates perform worse than average adults.
Excessive compensation dramatically decreases performance of inexperienced teams.

Below are the details. The Marshmallow Challenge [...] task is simple: in eighteen minutes, teams must build the tallest free-standing structure out of 20 sticks of spaghetti, one yard of tape, one yard of string, and one marshmallow. The marshmallow needs to be on top.
Recent kindergarten graduates perform better than average adults (average height of the structure: 27 in. vs. 20 in.).
Wujec notes that adults usually spend a long time planning, then building a single structure. When they place the marshmallow on the top of the lightweight structure, the whole thing often collapses. The incorrect assumption is that marshmallows are light - but compared to spaghetti, they are not. Kids build multiple successive prototypes, always with the marshmallow on the top, until they arrive at the best structure.

Does this remind you of some projects at your job? Plans, meetings, months of implementation, and then the final marshmallow that no one thought would be so heavy is placed on the top and - oops! down it goes... Think of kids making prototypes and compare this to agile software development, with its emphasis on feedback and iterative approach to the final results. Or think of scripting (Perl, Python, etc) and high-level (R, MATLAB, etc) programming languages, which, with their wealth of libraries, allow one to prototype ideas quickly.

Recent business school graduates perform worse (10 in.average height of structure) than average adults.
CEOs performs slightly better than average adults, but when an executive administrator is added to a team of CEOs, the team performance significantly improves (from 21 to 30 in.  average height), perhaps because executive administrators have special skills of facilitation.
Excessive compensation dramatically decreases performance of less experienced teams (from 20 in. with no compensation to 0 in. - yes, not a single standing structure! - with $10,000 prize for the tallest structure). I am not sure what level of compensation is excessive in the real world, but it makes me think that the highest compensation may not always lead to the best performance.
Tom Wujec concludes: Every project has its own marshmallow. How true!