Musing: Project deliverable dates

Working on a roadmap for work, pondering deliverable dates and 'accuracy'

Not sure if I've read this anywhere, but does anyone remember a planning strategy along the lines of "however long away a release is predicted; expect plus/minus that value"?

If someone tells you something should be available in a week, expect it in a week +/- a week. It couldn't be before today; but it could be another week away or longer - but another week is a good buffer. If a deliverable is 12 months away; expect it in 12 months +/- 12 months. Given how deliverables happen... that actually seems born out in practice?

Or maybe I'm thinking less expect, more of "it's likely". So it's likely to need a buffer based on however long it's expected to take. Because, honestly, the 'expect' would be "if it's in 12 months; expect it in 12months +6months". So expect is spoken deliverable, plus half again. Likely is within same period tolerance?

This thought came about when doing roadmaps and dividing commencement/ deliverable dates into year quarters. In the near future, that's a reasonable target. In the further future... well deliverable's more likely to be in an expected half of a year. Further future again, we're looking at variances fluctuating to a full 12 months. I feel that having that sort of chart gradience in a roadmap to show that sort of uncertainty would help understand how predicting the future is a charlatan's game.

A chart with three years, each year broken into four quarters


A chart with three years, the first broken into four quarters, second year is two halves, last year is just year

I'd love thoughts please.