Last updated: 23 September 2026
Socionics is a theory with a lot of secondhand material online, much of it translated, copied between sites and drifting from the original a little more each time. This page explains how Socionics Insight tries to avoid adding to that: where the data comes from, how it's checked, and what to do if you find something wrong.
It's a page about accuracy to the theory, not proof of the theory. Whether Socionics itself holds up is a separate question, and I've tried to answer it honestly in Is Socionics Scientific?
One source for every type and relation
Every function stack, quadra, club and intertype relation on this site comes from a single dataset: socionics-core. It's open source under the MIT licence, so anyone can read it, check it against their own understanding of Model A, and use it in their own projects.
Pages don't restate this information from memory. The Model A diagrams, the famous-people type strips and the compatibility tool are generated from it when the site is built, and every table on the type profiles, relation pages and comparison pages that restates it is checked against it, cell by cell, on every build. If a relation direction or a function position is wrong in the data, it's fixed in one place; if a page disagrees with the data, the build fails.
The site follows classical Model A as set out by Aushra Augusta. Where a page draws on another school, it says so.
Function claims are checked on every build
Prose is where errors creep back in. A sentence like "the ILE's creative function is the SEI's weakest" can be written confidently and still be wrong.
So the site runs an automated check every time it's built. It reads the written claims on the comparison pages and the famous-people pages (which function sits in which position for each type, and which side of an asymmetric relation experiences what) and verifies each one against socionics-core. It currently verifies 777 claims. If any claim fails, the build fails and nothing is published until it's fixed. The prose on the type profiles and relationship pages isn't covered yet, so for now those rely on a careful read.
When the check was introduced in September 2026, it found real errors on comparison pages that had been live for months, including Benefaction pages that credited the wrong function with the benefit and Extinguishment pages that put each type's leading function on the other's weakest position. An audit of the famous-people pages at the same time found Model A strips with functions in the wrong positions, and a read of the Dual comparisons found two that described the partner's most needed function as their most draining one. Those were corrected. Comparison pages that hadn't been through the full audit were taken offline rather than left up unchecked, and their URLs now point to the compatibility tool for the same pair.
I'd rather tell you that than claim the site was always right. It wasn't, and the check exists because of that.
What's original and what's attributed
Most English-language Socionics writing traces back to a small number of translated sources. Where this site uses classical descriptions, they're attributed to their source and treated as background, not as the main content.
The analysis on the type, relation and comparison pages is my own. It's written to explain how the model works in practice, not to repeat what other sites already say. Where I'm offering an interpretation rather than stating what the theory says, I try to make that clear.
I use AI tools to help draft and edit some pages, and to build and test the site. Every page is reviewed by me before it's published, and the automated checks described above apply to every page they cover, however it was written. The analysis, the judgement calls and any errors are mine.
How famous people are typed
Typing a public figure from interviews and public behaviour is an interpretation, not a fact. Each famous-people page gives the reasoning behind the typing, and where a figure is commonly typed differently elsewhere, that's worth knowing too. The hub explains the method in more detail. A typing that doesn't hold up is a correction like any other: tell me, with your reasoning.
Typing people, including you
The free test gives a working hypothesis from 12 behavioural questions and says plainly what a short test can't see, including when your answers leave the result open between several types.
The written typing reports work differently. They're built from a questionnaire answered in your own words, and every report shows its reasoning, including the close calls. A report is still a hypothesis, and says so.
Who writes this site
I'm Spencer Stern, a London-based product and experimentation analyst and the author of the Socionics Made Simple series, 16 books covering each type in depth. I also built Socion, a matching app based on the intertype relations, which is designed to collect evidence on whether the relations predict anything.
My own typing history is part of why this page exists. I self-typed as LII for years, was later typed as SEI, and now type as ILE. It's easy to be confidently wrong in Socionics, which is exactly why I don't want the site to rely on my confidence.
Found an error?
If something on this site contradicts Model A, misstates a relation, or gets a fact about a real person wrong, please tell me: the contact page, choosing "Type correction or feedback" as the subject. Include the page and, if you can, your reasoning or source.
Factual and theoretical errors are corrected as soon as they're confirmed. Where a correction changes a page's meaning, the page notes what changed and when. Disagreements about interpretation are welcome too, but they're judgement calls rather than errors, and I may not agree.