The research behind this bias starts with a clean little thought experiment about gambling.

The bias got its formal start with economist Daniel Ellsberg in 1961, a decade before he published the infamous Pentagon Papers. In a paper on decision-making, he laid out a thought experiment built around two jars of colored balls and predicted how people would choose.

The first jar held exactly 50 red and 50 black balls. You could see the split. The second jar also held 100 balls, but the mix was hidden. It could have been anything from all red to all black, with no way to know.

Picture being offered a bet on pulling a red ball. Which jar would you choose?

Think about what’s actually in that hidden jar. It might be stacked 99 black and 1 red, which would be awful for you. It might be 99 red and 1 black, which would be great. With no information pointing either way, every lopsided possibility is balanced by an equally likely opposite. Your best guess for the hidden jar lands right back at 50/50, the same as the jar you can see. On paper, the two bets are worth the same.

So why do most of us still choose the visible jar?

Because that one hands you 50% for certain. The hidden jar gives you 50% on average, but the real odds are out of sight, and they might be terrible. That unseen possibility, the chance you’ve badly misjudged the whole thing, is what people are running from. Not bad odds. Unseen odds.

That’s essentially the ambiguity bias. People take a guaranteed coin flip over a mystery, even when the mystery is, on average, exactly as good a deal.

Almost everyone reaches for the one they can see. Ellsberg predicted exactly that, and the researchers who tested his idea in the years afterward found he was right.

This held up across decades of follow-up work. Researchers ran variations on Ellsberg’s setup, and the preference for the known option kept showing up. Even when they explained the logic to people and walked them through their own reasoning, the pull toward the visible jar shrank but never fully went away. Humans are just sort of stubborn that way.

Here’s an important distinction, and it’s worth holding onto for the rest of this episode. Risk is when you know the odds, and they might go against you. Ambiguity is when you don’t even know the odds at all. Those are two different things, and the second one bothers us a whole lot more. We can often stand to deal with a known long shot more easily than we can stand to deal with an unknown.

Why is the unknown harder to stomach? Because missing information gets processed less like a math problem and more like a threat. When we can’t fill in the gaps, an older part of the brain assumes they’re hiding something dangerous. So we default to the option we can fully picture and steer clear of the one we can’t. It’s a remnant of our old hunter-gatherer brains.

That’s also why the effect gets stronger as the stakes go up. A trivial guess doesn’t bother us all that much. But when the outcome matters, the discomfort with the unknown gets sharper, and the instinct to retreat to the familiar gets stronger. Researchers have even linked stronger ambiguity aversion to anxiety and a broader preference for safe, predictable choices, which fits the pattern. The more threatened we feel, the more we crave a clear picture, and the more we’ll give up to get one.

Long ago, in a world where the unknown genuinely could be dangerous, avoiding the unknown was a decent survival bet. The trouble is that the same reflex follows us into the modern day, into meeting rooms where the unknown is a new idea, an untested market, or a person we can’t read yet. The people most exposed to it are the ones whose jobs are built around making hard decisions with often incomplete information: designers, PMs, engineers, and researchers.

The art of our craft is making decisions before everything is clear, which means ambiguity bias is quietly influencing every one of those decisions.

A team weighing the roadmap has a familiar feature for a known market with clean benchmarks competing for time and resources with a new feature for a new market with fuzzy projections and real upside. The familiar one is crystal clear. It comes with a story you can tell leadership. So it tends to win, not because it’s the better bet, but because the other one is the hidden jar and nobody wants to reach into it. The ambiguous opportunity gets filed under “too speculative” and quietly shifts below the line.

A genuinely novel design pattern has no precedent to point to, no pattern library entry, no competitor already doing it. That absence is seen as a negative, even when the idea is strong. The safer, more conventional direction wins because it’s easier to explain, and “easy to explain” gets interpreted as “more likely to work.” Novelty gets penalized for the crime of being unfamiliar.

A proven engineering framework with years of Stack Overflow answers feels safer than a newer tool that might be a better fit, simply because the unknowns are visible in one and hidden in the other. The hidden unknowns feel scarier than the known limitations, so teams stick with what they can fully see, sometimes long past the point where it serves them.

So far, this is all about ambiguous things like a market, an idea, a tool. When the unknown is a person, more biases stack on top and amplify it.

When the unproven idea comes from someone you don’t know well, you’re reading two blanks at once: the idea is unproven, and so is the person behind it. A trusted teammate floats a half-formed concept and gets the benefit of the doubt. A newer colleague floats the same concept and gets a wall of questions. The idea didn’t change. Your ability to read the source did.

We tend to feel this one a lot with new leaders. A new director, VP, a new executive, or even a new PM joins, and the team can’t read them yet; they don’t have a track record, there’s no sense of how they handle pressure, no shared history. That’s ambiguity in human form, and the brain does its usual thing and fills the blank with caution instead of giving them the benefit of the doubt. Trust has to be earned, not because the new leader did anything wrong, but because “I can’t read you yet” quietly defaults to “stay guarded.”

And that’s the second bias worth naming, so we don’t blur them together. The new leader isn’t just ambiguous. They’re often part of the out-group, not yet one of “us.” In-group bias already shorts the trust we extend to outsiders, so the two effects stack: the person is hard to read and not part of the tribe. That’s a big reason leadership transitions feel so destabilizing, and why an outside hire spends months earning trust that an internal promotion would’ve been handed on day one.

But notice what’s really driving all of this. It isn’t that the new market is bad, or the new idea is weak, or the new leader is untrustworthy. It’s that we can’t read any of them yet, and we treat “can’t read” as “probably bad.”

That’s the thing I want you to think about here. We don’t sit neutral in a blank space full of unknowns. Our brain wants to fill the space, and so it fills it with the least generous reading it can. The new PM’s pointed question becomes territorial instead of curious. The unfamiliar tool becomes a liability instead of an upgrade. The untested market becomes a risk instead of an opening. We haven’t uncovered any clarifying information yet, so our brains supply a placeholder, and the placeholder skews dark, and it pushes us to avoid it in favor of the known quantities.

Don’t start thinking that the fix is pretending everything’s clear or trusting everything and everyone blindly. Instead, notice when “I don’t have a read on this yet” is starting to shift into “this is probably a problem,” and ask whether anything actually justifies the jump, or whether your brain is just doing what it always does when it doesn’t have all the facts.