Directory · U
New here? Start with the premise →
Underestimation
3 min read · 683 words
Underestimation is the reading that something — a task, a difficulty, a duration, a cost, another person — will require less than it actually does.
The hardware does this systematically. The inhabitant underestimates how long projects will take. How much energy difficult conversations will require. How much capacity recovery from setback will consume. What other people are capable of, both in difficulty and in capacity. The pattern is consistent enough across inhabitants that it has been documented in research. The system runs a reliable bias toward optimistic estimation of what its own operations will require.
It would be nice to think that one is the exception. One is not.
THE MECHANISM
The estimate runs from a model that excludes the obstacles the actual operation will encounter.
The estimate assumes things will proceed roughly as planned. The actual operation includes unexpected complications, the inhabitant’s own variable state across the duration, conditions that did not cooperate as the model assumed. The estimate based on smooth-execution produces consistent underestimation of the messy-execution reality. The model is not lying. It is simply incomplete in a predictable direction — and the direction is always the same one.
OPERATING FROM ESTIMATES WITHOUT CORRECTION
The inhabitant who has consistently taken twice as long as estimated, and continues to estimate from the same model, gets the same results.
The bias persists across life unless deliberately corrected. This time will be different. It will not. I’ll be more efficient. The model has been registering the inhabitant’s intentions for years and producing the same gap regardless. The pattern continues because the inhabitant did not change the operation that produced the pattern. They changed the resolution and re-ran the operation.
CHECKING ESTIMATES AGAINST PAST ACTUAL OUTCOMES
For the recurring categories of operation the inhabitant runs, how do estimates compare to what actually happened?
The honest data usually surfaces the bias in specific directions: tasks take longer, difficulties last longer, costs are higher, energy required is greater. The pattern is information the inhabitant can use to correct future estimates — provided the pattern is actually examined rather than dismissed as a string of unfortunate exceptions, none of which represent the underlying baseline.
The underlying baseline is the pattern.
APPLYING A CORRECTION FACTOR
The inhabitant who has consistently taken twice the estimated time can apply a doubling to current estimates and produce more accurate planning.
The correction does not feel natural. It produces estimates that the optimistic system reports as excessive. Six months? Surely it can’t take six months. It usually takes six months. The estimates often turn out to be closer to accurate than the uncorrected ones.
Trusting the math over the feeling is the discipline. The feeling will lobby against the correction every time.
OUTSIDE VIEW VS INSIDE VIEW
The inhabitant estimating their own operation runs the inside view, which is subject to the optimism bias.
The inhabitant looking at similar operations done by others, or done by themselves in the past, runs the outside view, which is usually more accurate. The combination — using outside view to ground inside view — produces better estimates than inside view alone. The question to ask: how long has this kind of thing actually taken, the last several times anyone did it?
The honest answer is rarely the one the inhabitant wanted to hear, which is part of why the question gets skipped.
UNDERESTIMATING OTHERS
The inhabitant who underestimates other people’s capabilities. Who underestimates the difficulty of the conditions another person is in. Who underestimates the time recovery requires for someone going through difficulty.
These underestimations damage relationships and produce miscalibrated responses. They should be over this by now is often the underestimation talking. So is anyone could have handled this about a situation the inhabitant has not actually been in. The honest examination extends to estimation of others, not only estimation of the inhabitant’s own operations.
The model is consistently optimistic. The inhabitant who knows this corrects for it — and the inhabitant who refuses to know this keeps being surprised by the same thing for sixty years.