Search This Blog

Nov 28, 2015

Testing people-knowledge

One of the most difficult things about adjusting to a new large and complex organization is learning who is who. Fundamentally, it is a task of building predictive models for each key person – how would she or he behave in such and such situation. One small subset of it is who can you trust. Once I had a colleague, who lied all the time. You could not trust anything she said – about anything. It is an extreme case, but it still took me a couple of months to figure that out. It is not something people put on their resumes, and you do need evidence. Another person would sabotage any project – not out of ill will, but because he so terribly ineffective and disorganized. To find that out, you need at least one failed project; as evidence. Some people are just awesome at everything they do – not many, but they do exist. I am definitely not one of them, alas. Those are the most valuable and the scarcest asset of any organization, and you can ask them to do anything – up to the point when they are overwhelmed and can do no more.

However, usually it is a lot more complicated: most people are good at something, and bad at something else. Placing them in right combinations into the right tasks – THAT is an art form. Let us call it the managerial casting. To do it, we need those mental predictive models. What will happen if we ask so-and-so-do this-and-that with such-and-such? Figuring all this stuff out takes many months, even years. And even then people are unpredictable, they get tired, get cranky, but they sometimes learn and get better at certain things. They quit drinking, find God, resign to their fate – who knows.

But I am interested in the people-learning process. What are those people-models, cognitively speaking? How do they form? How much information is needed to build one? Can it be helped? We definitely get better at this game with experience, because we recognize – oh, this person is just like someone I knew before. We assemble a cast of characters, which serves as a map for all future colleagues and acquaintances. It is quicker that way, but also vulnerable to errors, for two people may have a similar set of traits, but expressed to various degrees, and the mix can be very different.

Can we test how accurate are those models? I am thinking of a study where a manager is asked to predict how his or her colleagues will behave in certain situations. And then we can either actually check in real life (by putting people in such situations), or, easier and cheaper – administer a self-reporting survey to the people. That would be a decent test of leadership abilities; easy to do, hard to fake. I also wonder how I would score, for I have been wrong about people quite a few times. But I have not seen anyone who was very good at it either. It is an empirical question – is there some variance?

Or perhaps we can measure the size and the richness of one’s mental library of characters? I am not sure how, but perhaps by giving someone some vignettes and asking if they knew someone like that, and put those imaginary people in new situations? It could be a test without the follow-up with real colleagues, cheaper still.

Are there any adventurous researchers out there in need for a new project? I will help. Or perhaps such a test does exist already? –let me know.


  1. Fascinating. But not everyone judges people by analogy, Sasha. For me it starts with a first impression (which is occasionally biased) and accumulating evidence of how a particular individual behaves. I am certainly old enough to have a cast of characters in my head, but I am just not wired that way. Most people I know are so unique. And then there are the immeasurable intangibles.

  2. Leadership/talent models tend to suffer from confirmation bias, that is, calcified ideas of leadership/talent drive inquiries that tend to reconfirm those same qualities. Your talent pool, even if competent, tends to be homogenous – in education, for example, the tendency to see extroversion as indicative of competency, a notion proven wildly inaccurate again and again. Hyper-competitive industries have had to combat this in ways not generally encountered in education because in those arenas the discovery of talent is so vital to their immediate survival - think Google or Tesla or other industries where paradigms can change on a dime when someone has a better idea (distributed learning is now giving education a taste of that kind of paradigm-changing competitiveness).

    First impressions, for example, are biased observations (generally) that often impede the discovery of talent, so much so that Google has designed ways to minimize them during interviews (

    The answer to proper leadership with regard to managing people is the ability to align talents with needs, and to understand that talent takes many shapes and that needs can be fulfilled by a variety of strategies. That's a very difficult managerial model, but essential to quality leadership. Most leaders think the opposite – talent takes this particular shape and this particular need requires this singular solution. People, though, bring a complex set of tools to the workplace. The leader's job is to properly harness and foster those individually to most benefit the larger organization.