When you look at what appears to be checkered farmland with a magnifying glass that illusion disappears.
What purpose does a magnifying glass serve when examining digital images?
It appears more like little buildings, roads, pipes or something along those lines.
"Something along those lines." That's the indication of a weakly inferred identification.
The "identification" of cityscapes, industrial assemblies, or similar types of artificial construction passes through very broad goalposts because we are accustomed to a high degree of incomprehensible complexity and substantial variation in them. A lot of shapes qualify as a "building," and a lot of collections of lines qualify as "road system." There's even a screen-saver for Linux that generates what many people take for aerial city plans, but which is just a simple set of rendering algorithms that generate essentially Cubist tonal patterns and lines.
The question to ask yourself is not whether what you're looking at is "something along the lines" of industrial installations, but what test you could devise to prove it's not industry. That is, if the data had to remain the way they are, and you had to make a case for its not being industrial equipment, what examinations or interpretations of the data would you make? This is the "falsifiability" test of the hypothesis. If you can't figure out how your interpretation could be disproven, then you can't really hold it as a good explanation. That's where "looks like some kind of industry" shoots you in the foot: it's too general to be disproven, so it's too general to hold as a defensible interpretation.
The next problem is magnification. Whether you use optical magnification or simply make the pixels bigger in software, magnification removes context. It accentuates the abstract nature of the data, allowing you to "fit" more interpretations to it without your brain yelling at you.
And you're forgetting that these buildings and roads and oil refineries you see are not apparent in the original data. You have to artificially change the data in order for these figures to appear. That means you (or Skipper) bears the burden of proof to show that the manipulation reveals information instead of creates it as an accident of mathematics. Or stated differently, you bear the burden to prove that the contrast-amplified image is the proper state of the data. You can't get so excited with your magnifying glass that you forget that you grossly manipulated the picture.
Skipper simply supposes that "The Government" started with the data he sees in his manipulated versions, and then themselves changed it to give the data he downloads from the catalogue. His only evidence of any such process is the blatantly circular line of reasoning: since he "uncovered" these "artificial" constructs, then the government "must" have covered them up. Skipper's cheating; that's not valid reasoning.
I was certain as can be that it wasn't crops.
I was unclear; I meant that as a general example. When you photograph farmland from very high altitude it can look very abstract, almost like image compression artifacts. Because we see abstract patterns in regular photography, we don't feel that it's wrong to interpret abstract, regular shapes from other sources as "some kind" of artificial object (say, farmland).
|