In the 1880s a French agricultural engineer named Max Ringelmann had students pull on a rope wired to a dynamometer, alone and then in groups, and measured how hard each one pulled. He expected a group to equal the sum of its people. It did not. One person pulled an average of 85.3 kilograms. Put seven people on the same rope and the total never reached seven times that; per person the force fell to about 65 kilograms, and in groups of fourteen it dropped to 61.4. More bodies, less of each one. His results, finally published in 1913, gave us the first measurement of a pattern that still quietly drains teams today, now usually filed under the name social loafing.
If you manage people, you have watched this happen without having a word for it. A task that two people would own completely gets handed to six, and somehow it moves slower and nobody quite feels responsible. The instinct that caused it is almost always generous: the work looked hard, so you added hands. Ringelmann’s rope is the warning that hands are not the same as effort, and that the two can move in opposite directions.
The numbers you have probably heard are wrong, and the real story is more useful
Nearly every article on this topic cites the same figures: that effort drops to 93 percent with two people, 85 with three, and all the way to 49 percent in large groups. Those numbers are not Ringelmann’s rope data. As one careful review of the original sources lays out, they come from a table Ringelmann labeled as summarizing many findings without saying what task produced them, and the rope-pulling attribution traces to a 1927 paper by Otto Moede, not to Ringelmann at all. The real documented decline was milder, roughly a quarter, not a half.
That correction matters for a manager, because the clean version of the experiment is where the useful finding lives. In 1974 Alan Ingham and colleagues rebuilt the study properly with a rope and a dynamometer and found individual effort fell to about 91 percent in pairs, 82 in trios, and 78 in groups of six. A steady, predictable erosion. Then they did the clever part. They blindfolded participants and had them pull while confederates only pretended to pull behind them. The subjects believed they were in a group but were actually pulling alone, and their effort still dropped to around 85 percent. Nothing was physically interfering with them. They simply pulled less once they believed others were sharing the load.
Two different problems wear the same mask
That blindfold result split one blurry phenomenon into two separate ones, and only one of them is a management problem you can engineer around.
The first is coordination loss. When several people pull a rope, or ship a feature, or run an incident, they get slightly out of sync, step on each other, duplicate effort. Coordination loss is real and it is a process problem. Better handoffs, clearer sequencing, and smaller working groups reduce it.
The second is motivation loss, and that is what Ingham’s blindfolded subjects revealed. They lost no coordination with anyone; there was no one to coordinate with. They just tried less because they believed effort was being shared. Bibb Latané, Kipling Williams, and Stephen Harkins nailed this down in a 1979 study where people shouted and clapped as loudly as they could while blindfolded and wearing headphones playing masking noise. Told they were part of a larger group, they got quieter, even though each person was physically alone and could have shouted exactly as hard. The researchers gave the effect its modern name: social loafing, the motivation half of the Ringelmann problem.
This is the distinction most managers miss. You can design your way out of coordination loss. You cannot design your way out of motivation loss by adding more structure, because it does not come from structure. It comes from whether a given person believes their own effort is visible and matters. Throw more people at a struggling task and you often make the motivation half worse while trying to fix a problem that was never about capacity.
What it looked like on my own floor
Running IT operations at a large telecom, I learned this the expensive way during a major incident. A core service was degrading, customers were calling, and my reflex was to flood the bridge call with talent. Within twenty minutes there were a dozen people on the line, including most of my strongest engineers. I had assumed throughput would rise with headcount. Instead the call went slower and strangely passive. People waited. Suggestions trailed off into “someone should probably check that.” The engineers I most relied on said the least, because with eleven other capable people listening, surely one of them had it.
Later that year a smaller outage got handled by three people who happened to be on shift, and it was resolved faster and cleaner. The difference was not talent. The three-person room had nowhere to hide. Each person knew the outcome rode visibly on them. On the twelve-person bridge I had diluted ownership until it evaporated, then mistaken the quiet for calm. I had added capacity and subtracted accountability in the same move, which is a bad trade I have since watched plenty of other managers make with full confidence. It connects directly to why span of control has limits that do not bend just because you are busy: past a point, each person you add makes the group a little less than the sum of its parts.
The conditions that decide how bad it gets
Social loafing is not a fixed tax. It swings widely depending on conditions you influence. Steven Karau and Kipling Williams pulled the whole literature together in a 1993 meta-analysis of 78 studies and more than 15,000 participants and found the average effect was moderate, around half a standard deviation of lost effort, but highly dependent on context. Their Collective Effort Model explains the swing with a simple logic: people invest effort when they believe the effort is visible, that it matters, and that it leads to an outcome they actually care about. Knock out any one of those beliefs and effort leaks.
So loafing runs highest exactly where you would least want it: when individual contributions cannot be told apart, when the task feels unimportant, and when the group is large. It shrinks, sometimes to nothing, when the opposite is true. That is not a counsel of despair. It is a list of levers.
The levers you actually control
Make each person’s contribution visible. This is the single strongest finding in the whole body of research. In a 1981 follow-up, Williams, Harkins, and Latané showed that when people’s individual output was identifiable, the group-size effect disappeared entirely. This is not a case for surveillance, which breeds its own problems. It is a case for ownership. Every meaningful piece of work should have exactly one name attached to it, known to the group, so that effort and its absence are both legible. “The team owns this” is often a polite way of saying no one does.
Keep the team small on purpose. J. Richard Hackman spent decades studying teams and concluded the effective size for most is four to six people. His reason was arithmetic as much as psychology: as a team grows linearly, the communication links among members grow by the formula n(n-1)/2, so a five-person team manages ten links while a ten-person team manages forty-five. Jeff Bezos reached the same place from the other direction with Amazon’s two-pizza rule: if a team cannot be fed by two pizzas, it is too big. Adding a seventh or eighth person to a working group does not just risk one more loafer; it adds links and dilutes visibility for everyone already there.
Say why the work matters, and mean it. Task meaningfulness was one of the strongest moderators in the research for a reason. People do not loaf on work they believe in and are seen doing. A manager who can connect a piece of work to something a person genuinely cares about has already recovered a chunk of the lost effort before assigning a single task. This is slower than reorganizing a chart and it does not feel like management, but it moves the number that reorganizing never touches.
Shrink the group for thinking, widen it only for doing. The most practical habit I kept from that incident is to keep the deciding group small and the executing group as large as the work genuinely needs, never larger. Three people can own a decision. Thirty can carry it out once it is made. Trouble starts when you put thirty people in the room where three should be deciding, and then wonder why nothing lands. If you want to understand what actually moves a team’s output, this is near the center of it, and it is almost entirely within a manager’s control.
None of this means lean teams are always right or that you should starve real work of people. It means the reflex to answer a struggling effort by adding bodies deserves suspicion, because the thing you are adding is not the thing that is missing. Ringelmann measured it on a rope a century and a half ago. The rope does not care how many hands are on it. It only responds to how hard each one actually pulls, and that is a function of whether the person on the end believes anyone would notice if they stopped.