Any idea why these tests rank video game characters so highly? I ran some random ones through just now, and Gordon Freeman scored 53%, John-117 scored 42% (including actions he took in the expanded universe), Pre-Other-M Samus scored 43% (Post-Other-M Samus scored an unbelievable 112%!!), and Jak/Mar scored 45%.

I suspect it's because video games usually aren't fun unless you're more powerful in some way than 90% of the mooks you run into (otherwise the game would be all boss battles, and the pacing would be terrible), and the character is actively being controlled by the player.

The thing with this test is that it detects character traits as opposed to actions.

What I mean by this is that Sir Large McHuge, suave young attractive philanthropist playboy multibillionaire, owner of a multinational high-tech product empire, and secret fighter of crime would be flagged as an enormous Suvian by the test you linked.

I've also described Tony Stark, aka the Iron Man (though my knowledge of the actual character is limited to The Avengers and the first Iron Man movie). Whether he truly is a Stu is hotly contested among certain fans.

However, the test would not catch Snivelly Uselesspants, a totally inept and hideous freak of a human being whose only purpose in life is to whine and complain about being stuck in a rut. That's an Anti-Suvian, and they're just as annoying to read as regular Suvians.

You see, it's about what the writers have their characters do that differentiates between Suvian and non-Suvian. If a character is shown to behave in accordance with his/her established personality, face logical cause-to-effect consequences for his/her actions (be they positive or negative), and experience character growth over a certain period of time (though admittedly difficult for "gag" characters or other one-dimensional types), then they are most likely not a Suvian.

In short, a character needs to be alive as opposed to a walking stack of random traits and clichés bound together by Rule of Cool and wish fulfillment.

I have used that test for characters as a way to try to figure out whether they have too many speshul traits. Cluttering up a character like that makes them (ironically) less interesting, so it can be quite helpful to answer those questions. My characters usually get somewhere from 10-25; the higher-powered characters do test in the low Mary-Sue range; the highest, an Abyssal Exalted character I played in a tabletop game, scored a 28. Exalted has a very high power level, and Abyssals by definition have a tragic past. And yet I don't think he was a Gary Stu. Yeah, he had the big sword and the necromantic power, he did save the world a couple of times, and he did manage to complete a nice little redemption arc (which is near impossible for Abyssals, considering their brains are basically linked to eldritch abominations)... but on the other hand, he started out as honestly cowardly (though he didn't stay that way), spent most of the game looking more like a living mummy than a glamorous vampire, and got manipulated by pretty much every ancient evil in the game until he finally broke free (and died shortly thereafter, because you don't flip the bird to the Neverborn without serious consequences).

So yeah, I think it's a useful tool, but you do have to look at the setting and the character's personality. If you're writing a high-powered character, and the character needs all of those powers to make the story work, you might ping as Mary Sue even though the character is an engaging, interesting character. Some Mary Sues don't show up as Mary Sues on the test simply because they're low-power self-inserts with little personality, who don't disturb the canon much but don't add anything to it either.

In-universe, Mary-Sue litmus tests are represented by litmus paper, which works like ordinary litmus paper except that it measures glitter level instead of pH. It has a tendency to burst into flames when a particularly bad Sue is around, but at least your agents won't destroy yet another CAD.