So if you look into the eyes of someone that likes you, their pupils will be dilated. If their pupils are tiny and you're not standing in the sun, they don't like your face at all.
On the other hand, while learning how to recognize cues and body language well is certainly a good thing, you can't rely on tricks like this as if they're gospel, either. There's an amount of intuition and complexity involved that isn't picked up by simple rules, and people's behaviors can differ wildly across the board. Last thing we need are people who think their significant others don't like them anymore based on something as silly as the size of their pupils.
You're not supposed to take anything as "gospel". But there's a set of information there based on psychological evaluation and studies and generally applies so take what you can from it. It'll never apply in 100% of cases but if you apply it correctly you'll be able to read people better and act empathically with people, for good or evil.
And what you called silly is shown to be quite true by psychological scientific studies, by the way
I didn't mean that the fact itself is silly. Sure, I'm willing to believe that, in general, it's true. However, if you're in a situation like that, the most that fact will do is maybe allow you to guess slightly better, which shouldn't be necessary. The problem is when people take little tidbits like "people's eyes dilate when looking at you if they like you" and then assume that one is a necessary or sufficient cause of the other, which is bogus. Hell, this goes for most tips like that. Things like that are for enhancing your intuition and making better guesses about things, and that's about it. No single reflexive or behavioral quirk (and often, not even patterns of them) is ever a sufficient or necessary condition of anything as complex as "does this person like me?" or "is this person nervous?".
I'm not saying people actually act like this, but... okay, let's face it, some people
will. In the face of little behavioral and people-reading tips like the ones here, there are going to be people who allow those bits of trivia to override other common sense, reasoning, and intuition. Learning about human reflexes and impression can certainly
help, but they aren't total universal truths. Yeah, there are some gross similarities in how people act and react, but trying to judge and read people using only knowledge of those similarities, comparing their expression to some laundry list of "X means Y" comes off as extremely ridiculous. It's extremely distancing (ironically enough), the research is still young and controversial to begin with, and the variations and nuance of human interaction (and human personality) are so vast that it's foolish to apply some set of quasi-scientific "rules" to all interaction, especially since most actual emotional "reading" is done not by reading individual behaviors, but patterns of behavior, including the context in which they are done, and through a knowledge of the person involved. Yeah, it's nice to know things like "using distancing language makes a person seem insincere" and "a person fidgeting with their hands a lot seems restless", but those tips are only useful when applied to a more natural, holistic, and personalized understanding of how people operate.