Estimating radiative forcing is about measuring relative to a baseline. Here, the baseline is a world with no contrails. When you introduce contrails, you're introducing cloudy volumes predominantly made of ice crystals and occurring very high in the atmosphere. On the balance, these clouds re-emit more long-wave radiation (e.g. what' emitted by the Earth's surface) than they allow to escape the atmosphere.
Hence, these clouds have a small but positive net radiative forcing - meaning that aviation, by the way it leads to contrail formation, has at least this small radiative forcing on climate.
> As any object traveling through the atmosphere at that altitude, disrupting air, is going to form condensation and cloud trails. The more moisture in the atmosphere, the more trails.
Actually - it won't. We rigorously started studying contrail formation back in WWII when meteorologists tried to anticipate when bomber flights returning from mainland Europe might induce contrails and leave a path for intercept fighters to follow and shoot them down. As the science and understanding of vertical atmosphere thermodynamic structure and cloud microphysical structure has advanced in the ensuing 80 years, we have a much better understanding of when contrails are likely to form, versus when they aren't.
But don't take my word for it. Look up at the sky any time you hear an aircraft - sometimes you'll see a contrail, sometimes you won't. Contrails aren't a given when a jet flies high in the atmosphere.
(that's actually the entire basis for the Contrails/DeepMind team's work - avoid areas where contrails _are_ likely to form, to avoid that radiative forcing from the first part of this comment)
> Are we suggesting changing flight routes and wasting more fuel (which pollutes more) to protect the ground from these 0.0000001% reduction in light cloud trails? Seriously. I want to know the science behind how this plays out.
The science is pretty well developed at this point. You'd probably hit it in an undergraduate-level physical meteorology class. The missing detail that the Contrails team helped solve was improving forecasts of the key parameters involved here from weather models.
The whole point is that this is _another_ lever that flight planners could use to optimize their route planning. It's just one factor. It has trade-offs - although those trade-offs aren't always net negative (e.g. it's not a given that the "less contrail-y" route is also the "more fuel burn-y" one).
