But, but, isn't it just that it's really hot in the sun and cold in the shade?
Sure. And as soon as you figure out what "it" is you'll be going places.
For a given point on a surface, the amount of heat it can radiate is determined partly by the radiation falling on it from nearby radiant sources—either emitters or reflectors. In the pure language of physics, this is one of those "nasty integrals." Just the geometric part (i.e., "per steradian" irrespective of "per wavelength") is heinous for a single radiant source, let alone all of them. This is what we call the "view factor."
Here's how engineering fudges it.
Imagine a point on the ground surrounded by a smattering of radiant sources. You want to know the total amount of radiant energy incident upon that point. Imagine a circle around your point, of comfortable radius. Now build a hemispherical dome of clear plastic using the circle as your footprint. Put a webcam at your point of interest and aim it in all the various directions. When you get to a radiant source, draw an outline of it in Sharpie on the dome as seen from the center. For the sun, say, you'll have a little circle. For the wall of the next door apartment building, you'll have a certain shape.
Now project all those Sharpie figures downward to the circle, the floor of your little dome tent. Imagine hanging a plumb bob from the outline of each one and then tracing on the floor where that figure lies. Figures near the top of the dome will project downwards in a fairly congruent fashion, since their orientation on the dome means the projection is very nearly perpendicular to the planar approximation of the figure. But for items on the walls closer to the ground, they project obliquely downward. Though they may have significant area on the dome, their projection onto the floor represents the fact that the floor sees them mostly edge on, thus they will be narrower. The area of each projected figure on the floor is a close approximation of its "view factor" to your point of interest—the effect of the geometric relationships (distance, orientation, angle of incidence, etc.). Later you'll come back and add values for intensity and wavelength and all the other physical values you need.
This is a great iterative solution that lends itself well to implementation on a computer without having to analytically derive any nasty integrals. You can set it up to adaptively subdivide the surfaces when it detects that nearby points have markedly different view factors to elements in the environment. Turn on the computer, take a long lunch, and when you come back you'll have a picture (literally) that approximates the radiant influx on each part of your object, including the cases where your object reflects light upon itself. This is why we don't let Frank Gehry build spaceships.
Now they actually did this back in the day for the Apollo lunar module, albeit without the adaptive subdivision. The computer they did this on was about as powerful as an old Apple II, so their results were crude by modern standards. But it was useful information that they wouldn't have been able to get previously at that early stage of design. Now the computer I use for this sort of stuff weighs more than a lunar module...
This kind of complexity is why I got drawn into the sciences. The more you learn the more you realise there is to learn, and that many 'solutions' really boil down to 'this is the best we can do and it still might go wrong so we need a contingency for that, which will introduce its own complications, so we create this system that as far as we know should work well enough to complete the thing it is meant to do if all goes well and we've anticipated the problems correctly'. And that that's okay. Too many hoax believers operate on the impossible notion that the designers had to find the perfect solution to every problem (or indeed that such a solution even exists).
You can forgive people for thinking Apollo had to be perfect because its designers were sort of driven that way too.
The CM's caution and warning system had an irreducible problem in that when you first turned it on, it would shriek and holler because all the things it was supposed to monitor had not yet "warmed up" to full operating status. The engineers went down a rabbit hole on this problem, trying to adapt the system to all that changing state. As the story goes, it was one of the astronauts who drew the parallel to an automobile. Cars at the time did the same thing; when you first turned them on, a lot of the warning lights came on and then cleared after a few seconds—and no one cared. The solution to the irascible C&W dilemma was to let the CM shriek and holler for a minute or so, and then push the reset button.
Similarly, the LM ascent stage was always slightly off balance. If you watch the 16 mm footage, the LM wallowed constantly in one direction and had to be corrected. This was deemed acceptable since the corrective control action wasn't very significant. But it's one of those things that could have been corrected by a design change, but wasn't. The existing design was "good enough." Hence being successful as an engineer means (among other things) realizing when Better is the enemy of Good, and emotionally letting go of the ghosts in existing designs. Every successful design will still harbor "We never fixed that" issues. And every engineer accumulates a list of, "If we had it to do over again..."
