What is Internet Personality Scanner?
Four questions, one of eight archetypes, about thirty seconds — think less 'official psychological assessment' and more 'accurately roasts your posting habits'. Eight scan modes set the flavour, from Wholesome and Gamer to Internet Goblin and NPC, and each question you answer carries a hidden tag; whichever tag you pick most often decides your archetype. Your answers are processed entirely in your browser and never transmitted, so retaking it with deliberately different answers costs nothing and reveals the machinery quickly. It is built to be shared and compared — the real fun tends to happen afterwards, arguing about whether someone's result matches how unhinged their group chat behaviour actually is.
The most famous personality test was built by two people with no training in psychology
The Myers-Briggs Type Indicator sorts people into sixteen four-letter types and is used by a great many employers. It was created by Katharine Cook Briggs and her daughter Isabel Briggs Myers, neither of whom had any formal training in psychology, working from Carl Jung's 1921 typology — which Jung had arrived at through observation rather than measurement, and which he never intended as a sorting system.
The deepest problem is not the origin, it is the retest. A test that measures something stable about you should give you the same answer twice. Studies have repeatedly found that a substantial proportion of people — in some, around half — come out as a different type when they take it again a few weeks later. Nothing about them changed; the boundary they happened to fall near did.
That is the structural flaw. The traits it splits on are continuous, so almost everyone sits somewhere in the middle of at least one of them, and a test built on dichotomies has to shove those people to one side or the other. The nine points either side of the line get reported as two different kinds of person.
What a real personality model looks like
The model academic psychology actually uses is the Big Five, and it is worth knowing about because it is built the opposite way round.
It began with a genuinely odd idea called the lexical hypothesis: if a personality trait matters to people, languages will have evolved words for it. In the 1930s Gordon Allport and Henry Odbert pulled roughly 18,000 personality-describing words out of an English dictionary, and decades of statistical work since kept finding the same five clusters underneath them — openness, conscientiousness, extraversion, agreeableness and neuroticism.
The crucial difference is that it does not give you a type. It gives you five positions on five sliding scales, which is the honest shape for something that varies by degree. Nobody shares their Big Five result at a party, and that is precisely the trade-off: it is more accurate and far less fun, because 'moderately high on conscientiousness' is not an identity.
Why you shared the result anyway
On 21 December 2013 the New York Times published a quiz that mapped your dialect from the words you use for a carbonated drink and a small road-side insect. Within eleven days it was the most-visited piece the paper had published all year, beating every news story on the site, despite appearing in the last week of December.
Almost nobody shared it to inform anyone about American dialects. They shared it because the result was a small, portable statement about who they were that they had not had to write themselves. A quiz hands you an identity claim with a plausible-looking method attached, and passing it on is a very cheap way to say something about yourself.
Which is also why the accuracy barely matters to the sharing. The question a good result answers is not 'is this correct' but 'would I be happy for people to think this about me'. Archetype quizzes have understood this for a long time, and this one is not pretending otherwise.
How this one actually decides
There is no hidden model here and no scoring you cannot see. Every answer you can choose carries one hidden tag. Across the four questions you accumulate four tags, the tally is counted, and whichever tag you picked most often selects your archetype.
Being honest about the edge case: with only four answers, ties are common. In about a third of runs no single tag has a clear majority, and the rule that breaks the tie is simply whichever of the tied tags you picked earliest. So in those runs your first answer matters more than the rest, which is arbitrary — it is just not hidden from you.
That is also why the earlier question shuffle mattered enough to fix. It used a sort with a random comparator, which is not a uniform shuffle: measured over 200,000 runs it left the first question of the bank leading 32% of the time against 12% for the second. Since the first answer breaks ties, a skewed order skews results. It now uses a proper Fisher-Yates shuffle, so every question is equally likely to lead and equally likely to be the one left out.
What's in the scanner
Eight scan modes set the tone: Standard, Chaos, Wholesome, Gamer, Internet Goblin, Mysterious, NPC and Main Character. Behind them sit five question banks of five questions each, so Mysterious, NPC and Main Character run the Standard questions with their own styling rather than their own set.
Each run asks four of the five available questions, drawn fresh, which is why taking the same mode twice does not replay the same quiz. There are eight archetypes to land on, each with a rarity badge, and an aura assigned at random alongside the result purely for decoration.
Nothing is transmitted. Your answers, your result and your scan history exist only in the page and are gone when you reload it — there is no account and no server involved. The site as a whole loads analytics and advertising scripts, which the cookie banner and privacy policy cover, but none of them receive your quiz answers. And the page says it plainly at the bottom: the results are 100% fake. There is no psychology in here at all.