Diverse views, aggregation and incentives key to making crowds 'wise': Morgan Stanley
New Delhi, August 8
Diverse views, effective aggregation of information and incentives for participants to act on their knowledge are the three key conditions needed for crowds to make better decisions, according to a Morgan Stanley report.
"Wisdom is accompanied by specific conditions," it said, identifying cognitive diversity, aggregation and incentives as the three key factors.
The report examined how collective intelligence works across prediction markets and financial markets. It noted that crowds can be highly accurate when these conditions are present, but can also make poor decisions when one or more of them break down.
According to the report, cognitive diversity is not simply about differences in social characteristics but about differences in participants' information, knowledge, approaches and mental models. Such differences allow a group to draw on a wider range of information and perspectives.
The report said collective error tends to be smaller when participants are cognitively diverse and their views are properly aggregated. It identified diversity as the condition most likely to be violated in financial markets, where investors can influence one another and develop correlated views, potentially contributing to market booms and busts.
Aggregation is equally important because having diverse information alone does not ensure a good outcome. The report said markets use different mechanisms to combine information and arrive at a measure of the likelihood or value of an outcome.
It noted that even simple methods such as averaging forecasts can improve accuracy, while weighted averages and other techniques can provide further gains.
The third condition, incentives, refers to rewards for being right and penalties for being wrong. In markets, these incentives can be measured through monetary gains and losses. The report said participants with a well-founded view that differs from the market can have an incentive to act on that information.
Morgan Stanley also examined prediction markets, saying they can provide real-time probabilities that may be useful for forecasting. It noted that such markets have generally been found to be more accurate than polls, although the advantage is not always clear-cut.
In one example, the Iowa Electronic Markets had an average forecast error of 1.5 percentage points one week before US presidential elections, compared with 2.1 percentage points for the final Gallup poll in the elections studied.
The report further added that these principles of collective intelligence could have applications beyond markets, including hiring, employee training, meetings and organisational decision-making. It also noted that large language models could reach parity with superforecasters at some point in 2027.
— ANI
Reader Comments
"Even simple methods such as averaging forecasts can improve accuracy" — this is a great reminder that we often overcomplicate things. In our Indian families, we already do this when making collective decisions, like combining everyone’s opinions on what to cook for a wedding! The serious point here is about prediction markets. We don't really have a robust prediction market in India, unlike the Iowa Electronic Markets they mentioned. If we could institutionalize something like that for Indian elections and policy outcomes, it might give us better real-time probabilities than the opinion polls we see on news channels every week. Just a thought.
The part about large language models reaching parity with superforecasters by 2027 is fascinating and slightly scary. But I wonder — if we all start using the same AI tools to make forecasts, won't we lose that cognitive diversity that the report says is so crucial? In Bangalore's tech ecosystem, we already see everyone relying on the same platforms and APIs, which creates homogeneity in thinking. The wisdom of crowds might become the wisdom of one large model amplified across millions of users. That's a potential problem the report didn't fully explore. 🤔
As someone who works in capital markets in Mumbai, I can totally relate to the section about incentives and penalties. Indian markets definitely reward correct information — just look at how quickly some traders spotted the HDFC merger opportunity and profited before the official announcement. But the flip side is that when incentives are misaligned, you get pump-and-dump scenarios and penny stock scams. The report's framework of diversity + aggregation + incentives is actually a good lens to evaluate SEBI's various investor protection measures. Are they genuinely encouraging diverse views or just adding compliance burden? Worth thinking about.
The Iowa Electronic Markets example is telling — 1.5
We welcome thoughtful discussions from our readers. Please keep comments respectful and on-topic.