To a late riser like me it seems that the days are already getting longer even though is is still almost a week before the actual winter solstice. But this is not an illusion as the sunsets really are starting to happen later here, even as the length of time from sunrise to sunset continues to shorten.
How can this be? Well, the sunrises are actually also getting later too because in addition to getting longer and shorter the days actually “move around” a bit (in the sense that the time that a regular clock gives as noon even on the exact meridean of a time zone is not exactly the time when the sun is highest in the sky). This is due to a fairly subtle effect of the earth’s tilt combined with a slightly smaller effect of its orbital eccentricity (the fact that it is an ellipse rather than a perfect circle). These effects cause the length of the entire (noon to noon) day to vary with the seasons so that the clock’s equal 24-hour days cannot all be exactly centred on solar high points (so noon on a sundial moves back and forth relative to noon on a clock). Popular accounts of this phenomenon frequently relate it only to the eccentricity which actually gives the smaller effect , but this article in The Atlantic is closer to the mark in that it gets the sizes of the two effects right. (It does however give a wrong impression of how the obliquity effect works though, because it implies that it’s like the difference between summer and winter when in fact it is between solstices and equinoxes. At both solstices the noon to noon time is about 20 seconds longer than average and at both equinoxes it is about the same amount shorter.)
This effect has been well understood for (at least a couple of hundred) years but first came to my attention when a colleague asked me to look at an essay in which one of his students seemed to have figured it out for himself!