The lesson I actually drew was more like so:
Hm, if you release a few particles into this system you'll quickly notice that they all seem to be traveling along the same rough lemniscate. The lemniscate is a tiny minority of the entire space, so learning the differential equations lets me substantially improve my predictions of where the particle will be at any moment. ("It's probably near the lemniscate" is much more precise than "it's somewhere in space".)
If the difference between the best winter and the worst is a 40% longer duration (admittedly, not pleasant) and 15 degrees centigrade... there's plenty of room left for real variation. ;)
But if you do allow for that larger variation, you will soon run out of people to observe it. Long-lasting supplies scale somewhat linearly, but if you want to have fresh meat at the end of the winter, too, you will need quadratically more fodder at the beginning of winter to feed the animals, for example.
In both series, high technology from the science fictional context eventually plays a disrupting role in the fantasy world. You don't see a lot of that these days. I would slap my thigh with unrestrainable jollity if GRRM ends ASOIAF that way.
Excerpt from the abstract:
To right that appalling wrong, here we attempt to explain the apparently erratic seasonal changes in the world of G.R.R.M. A natural explanation for such phenomena is the unique behavior of a circumbinary planet. Thus, by speculating that the planet under scrutiny is orbiting a pair of stars, we utilize the power of numerical three-body dynamics to predict that, unfortunately, it is not possible to predict either the length, or the severity of any coming winter. We conclude that, alas, the Maesters were right -- one can only throw their hands in the air in frustration and, defeated by non-analytic solutions, mumble "Coming winter? May be long and nasty (~850 days, T<268K) or may be short and sweet (~600 days, T~273K). Who knows..."
(I like arXiv.)