A running record of significant changes to the site. Updated when something worth noting happens.
There is an RSS feed of new pages if you would rather not keep checking, and the icon for it is in the top right of every page.
October 2026
7 October 2026. New: a Live Typing Conversation. If you'd rather talk than write, book a 45-minute one-to-one call with me over Google Meet and I'll work out your type from how you think and engage as well as what you say, then send you a short written summary of the reasoning. It's linked from the typing reports page and the Get typed hub.
6 October 2026. Type Finder results now end with a link to Socion, the free app for meeting people by Socionics type and relation, in every result rather than only some.
6 October 2026. Every type, relation, function and group page, and every article, now shows when it was last reviewed or published, next to the author's name. Pages that draw on published Socionics books, such as Kovalenko and Zvonaryova's lecture course or Meged and Ovcharov's subtypes, list them in a short Sources line at the foot. See the foot of Conflict for an example.
6 October 2026. New tool: You and Them. Pick your type and someone else's and see the whole picture: the relation between you, where each of your functions sits for the other person, and how you differ on planning, explaining things, disagreeing and working under pressure. It fills in your type if you've saved one, and gives you a link to keep or send. The compatibility tool now links to it from every result.
6 October 2026. The romance styles page has a new section on what each style jokes about freely in public and what it finds most embarrassing, from Kovalenko and Zvonaryova's lecture course. It explains why complementary styles find conversation so easy, and why some first dates go quiet at the same topic every time.
6 October 2026. Two more small groups: quasi-quadras, on what each type needs settled before it will give ground, with notes for negotiations; and perception groups, on the kind of explanation each type takes in best. The supervision rings page now has a second published description of Gulenko's forms of thinking, and the all 35 systems page counts fourteen with names.
6 October 2026. Three articles now cite Kovalenko and Zvonaryova's 2020 lecture course. Is Socionics Scientific? has a new section on what Russian Socionists say about their own field, its dissertations and journals as well as its weaknesses. The Reinin dichotomies article explains how each trait connects to Model A, flags the two the authors consider least reliable, and adds further reading. Valued and non-valued functions explains why some schools call them verbal and laboral instead.
6 October 2026. Three new small groups, all aimed at work: stress resistance, on how each type handles pressure and switches between tasks; implementation groups, on who invents a method, refines it, adapts it and scales it; and planning styles, on how each type sets goals and handles time. The all 35 systems page now counts twelve with names, and each careers page links to all three.
6 October 2026. All fourteen intertype relation pages have a new section, "Where the functions meet". It shows where each of your eight Model A positions lands in the other person's, and explains what those contacts do: why Duality's help comes without fuss, why Kindred can feel prickly up close, and why Super-Ego and Conflict are hardest in company. See it on Duality.
6 October 2026. The Model A page now covers all seven characteristics that make up each position, not just strength and information flow: the mental and vital rings, vertness, inert and contact, evaluative and situational, verbal and laboral, and accepting and producing. Each one shows which Reinin dichotomies it builds. There is also a new section on dimensionality, and an interactive Reinin cube that shows how the eight positions fit together.
4 October 2026. On a phone, the row of type links at the top of every type profile now sits in two rows of eight, so all sixteen fit on screen. Before, everything after LSI was cut off. Wide tables, such as the Model A table on each profile, now scroll sideways instead of losing their last column.
4 October 2026. The typing report page has a new, optional way to help test the Typing Interview. If you opt in when you order, you take the interview after your questionnaire. I write your report without looking at it, and once the report arrives you see what the interview made of you and how the two compare. It takes about 15 minutes and costs nothing extra.
2 October 2026. The Typing Interview now has its own page, explaining how it will work, what it is and isn't, and what happens to your data. The waitlist has moved there from the Get typed page, whose card now links to it.
1 October 2026. There is now an Instagram link in the footer, between TikTok and YouTube.
1 October 2026. When your quiz result comes out close or uncertain, the email box under it now asks "Not sure if you're LII or LSI?", naming the two types your own answers came out closest to. Leave your email and the result arrives with a link to the tie-breaker: a short set of questions written to tell those two apart in the Type Disambiguator. The link above the box starts the same questions straight away, without an email address. Your result is shown in full either way.
1 October 2026. The Reinin dichotomies article has been revised. A new table shows, for each intertype relation, which of the fifteen dichotomies the two types share and which they split, and for six relations it shows how that depends on the pair. There is a new section on Reinin dichotomy tests and why I would use the traits as a cross-check rather than a way to find your type, and the definitions of the four basic dichotomies are now more precise.
1 October 2026. The Type Finder page now invites you into the pilot instead of starting the unsaved test version, so every run until 23 October counts towards making the final version shorter and more accurate. A test you already started can still be finished. If the pilot doesn't find your email address, you can now go back and try a different one.
1 October 2026. "Copy share link" under a Type Finder result now gives a short link, such as socionicsinsight.com/r/k7x2m9Ab, instead of a long one. When it is posted on Discord or elsewhere, the preview names the type and the top three, and opening it shows the full result. Older long links still work.
1 October 2026. Every type profile now has a section on its two subtypes, such as ILE-Ne and ILE-Ti: Meged and Ovcharov's lead and creative subtypes, named after the type's two strongest elements, with a short portrait of how each one talks, decides and comes across. A note of caution sits underneath, and a line on how they differ from Gulenko's DCNH subtypes.
1 October 2026. The Type Finder pilot is now open to anyone. Join it from the Type Finder card on the Get typed page; if you are not already on the mailing list, it signs you up as you start. The pilot runs until 23 October and works best if you already have a type in mind.
September 2026
30 September 2026. The relation pages and each type's relationships page, such as ILE relationships, now have a short box after the section on where the friction shows up, offering Five Texts That Start Arguments: a free one-page sheet of five messages that often land badly between partners, with one calm question to ask for each. On those pages the bar at the foot of the page now offers the free guide, as it does elsewhere on the site.
29 September 2026. The homepage's introduction now shows the eight information elements with Aushra Augusta's original symbols (filled for extraverted, outlined for introverted), each with its SLIDE System name and a link to its function page. The Start here list beside it now also points to Model A and to Socion.
29 September 2026. You can now read your type's chapter from Your Social World Explained free: the full profile of your type and the chapter on Duality, as a PDF in the second edition's full-colour layout. Choose your type, enter your email, and it arrives in your inbox, with a direct download on the page too. Each type profile, your test result, the Duality page, the Dual Finder and the Compatibility tool now link to it with your type already chosen. So does the homepage: from the Start here list, and from a line under the quadra grid once you have set your type. The books page now starts with Your Social World Explained and then the type-by-type series.
28 September 2026. There is a new Get typed page, in the main menu in place of Typing Reports. It sets out the four ways to find your type side by side, from the twelve-question quiz to a written typing report, with a short guide to which one to start with. It also has a waitlist for the Typing Interview, a conversational way of getting typed that is coming soon.
27 September 2026. Once you have set your type, the type profiles, function pages, relation pages and famous-people pages now offer your 16-relation map by email: every one of your type's relations, including which side of Benefaction and Supervision you are on. It is shown on the page as soon as you ask for it. The form sends your email address and your type and nothing else, as the privacy policy explains. While it is on a page, the bar at the foot of that page stays away, so there is only one offer at a time.
27 September 2026. When the quiz result is a close call, with at least one axis coming out 2 to 1, the email box under it now names the two types your answers sit between and offers the tiebreaker by email. Above the box is a link that opens the Type Disambiguator on that pair, so you can settle it straight away without giving an email address. Your result is shown in full either way.
27 September 2026. The comparisons page has a new 16 by 16 grid of every pairing. Find one type down the side and the other along the top. Pairs with a published comparison link to it, and the rest show their intertype relation until their comparison is written. An old link to one of those unwritten comparisons now opens the page for its relation, such as Supervision, instead of the compatibility tool.
27 September 2026. Your Social World Explained: A Complete Guide to Socionics, the 16 Personality Types and Relationship Compatibility is out in a revised second edition, redesigned throughout with new diagrams for every type, relationship and small group, and now in full-colour hardcover as well as Kindle, paperback and audiobook. The books page and the article sidebar show the new cover.
25 September 2026. The relaxation groups page had the two rational groups wrong, and so did I. Reinin's formula for a square is duality, semi-duality and kindred, but the rational list I had followed, his own, puts LSE, EII, ESE and LII together, and those four are related by duality, business and mirage. The rational squares that match the formula are LII, ESE, LSI and EIE, and LIE, ESI, LSE and EII. The page now explains the general pattern: a dual pair can join one of three others, the relations depend on rationality for two of the three, and blocking groups are the one that doesn't change. Thanks to a reader who pointed out the rational/irrational distinction.
25 September 2026. The Type Finder now asks how familiar you are with Socionics before the first question. It doesn't affect your result. If you email yourself your result, your answer is sent with it, as the privacy policy explains.
25 September 2026. The comparison grid on each of the sixteen type profiles now links every published type comparison. 58 cards, including every Dual pairing, were showing as plain text although their pages were live.
25 September 2026. How to Find Your Socionics Type has been rewritten. It now narrows the field with the small groups first, then confirms a type from four positions in Model A: what you can't switch off, which function is the method, where criticism hurts, and what feels like relief. Why You're Probably Not the Type You Think You Are has two new mistyping patterns, the Mirror trap and being typed as your own Dual, and a worked example of how I got my own type wrong twice.
25 September 2026. Each dichotomy in the Reinin table on the sixteen type profiles now links straight to its own section of the Reinin dichotomies article. Each profile also has its own sharing image, with the type's code, name, MBTI cousin notation and quadra, for when a link to it is posted elsewhere.
24 September 2026. The typing report page has a new section explaining how the report is made. I read your answers and decide your type myself, AI helps me draft the written report, and I check and edit every word before it is sent. A short table sets this against an automated report, where a test score decides the type. The rest of the page now describes the process the same way.
24 September 2026. The definition on the Introverted Sensing (Si) page has been updated so it covers both halves of Si: physical comfort and wellbeing, and a sense of when things drift from what is known and trusted.
24 September 2026. The article on DCNH subtypes in Socionics has been revised. It now opens with a two-sentence answer, credits the system clearly to Victor Gulenko, marks the subtype descriptions as attributed material, and drops an incorrect account of which Model A block each subtype strengthens.
24 September 2026. The quadras page and the four quadra pages, Alpha, Beta, Gamma and Delta, have been rewritten. Each now opens with a short answer, sets out the quadra's values using the SLIDE System names alongside the classical ones, shows which type carries each value and where, and ends with further reading. The overview page has a new table showing what each quadra shares with the other three. Five related articles have new titles, and two small errors about which types supervise which, and about Gamma's functions, have been corrected.
24 September 2026. The intertype relations page has been rebuilt. It now opens with a full 16 by 16 relationship chart showing what every type is to every other, followed by a short entry for each relation and a new section on how compatibility works. Benefaction and Supervision have been rewritten, each with a table showing your benefactor and beneficiary, or supervisor and supervisee. The old Benefaction shortcut gave the wrong answer for rational types, and that is now fixed. Three older articles on the same ground have been folded into these pages.
23 September 2026. The bar at the foot of the page now offers something that fits the page you are reading. On the relationship pages for each type, the intertype relations, compatibility, the conflict tool and the Dual Finder, it offers the free one-page sheet on the texts that most often start arguments between partners. Everywhere else it still offers the free guide. The comparison between the free guide and the type test has finished. A new one compares two timings for the bar, one that waits longer and one that appears sooner. The privacy page describes what is kept in your browser for it: which timing you were given, and nothing that identifies you. If you closed the bar in the last week, it stays closed.
23 September 2026. The Open Source page now covers the latest release of socionics-core, the free data package behind the site. Version 1.1.0 adds all fifteen dichotomies, the four Jungian ones and Reinin's eleven, with every type's pole on each.
23 September 2026. The Hub has been refreshed. It now leads with the free things: the tools, the reference library and the free guide, then the app, the AI assistant, the Discord and the paid services. There is also a new card for socionics-core, the open data behind the site.
23 September 2026. A new Methodology page sets out where the site's type and relation data comes from, how it is checked on every build, what is original and what is attributed, and how to report an error. It is linked from the footer, and every published type comparison now says its function claims are checked and links to how. Adding the checks also turned up one error, now fixed: the LSE page gave its stimulus group as "Esteem" rather than "Esteemed".
23 September 2026. Quiz results can now be shared. Once you finish, the address bar holds a link to your result and there is a button to copy it; whoever opens it sees which type it is, with links to the profile and the Dual, and an invitation to take the test themselves. The link carries only the type and which answers were close, nothing about you. Close results also have a clearer next step: when your answers fit two or four types, the result now names them and opens the Type Disambiguator with that shortlist already filled in. How the quiz scores you has not changed. See the quiz changelog.
23 September 2026. The Type Disambiguator now has questions for all 120 pairs of types. 13 more were written pair by pair, which covers every pair that differs on a single letter of the type code, so whenever the quiz says your answers fit two types there are questions written for exactly those two. The remaining 77 are written the same way, pair by pair from the two types' function stacks. All the sets are drafts and will be refined.
23 September 2026. A new article, Is Socionics Scientific? An Honest Answer: what it would take for Socionics to count as science, where it currently falls short, and what a fair test of its predictions about relationships would look like.
23 September 2026. A tidy-up of the famous people pages. The four-function strip at the top of each page is now drawn from the same type data as the rest of the site, which corrected it on more than half of them, and the directory now says how many figures have a full write-up (the others are listed by name, and an old link to one of them opens that type's profile). Beyoncé has a published page; Kevin Spacey's has been withdrawn. Carl Jung and Donald Trump have been partly rewritten, and page titles now follow one format.
23 September 2026. The 63 type comparisons marked "not yet available" have been taken down until they are written to the same standard as the published 57, so every comparison you can reach is one that has been checked. An old link to one of them now opens the compatibility tool on that pair instead. The relation pages now list only the published comparisons for each relation and say how many of its pairings are covered so far.
23 September 2026. An accuracy pass on every published type comparison. Each claim about where one type's functions fall in the other's was checked against the model, and the errors are fixed: several pages had the benefit in a Benefaction relation coming from the wrong function, the Extinguishment pages misplaced each type's leading function, and a few mentioned Dual pairs across quadras, which do not exist. Repeated summary sections are gone, the "how each sees the other" notes are now written about both types rather than in their voice, and each comparison links to both types' relationships pages. Every one also has a new section on where each type's functions land in the other's: which strengths meet a weak spot, what each needs that the other can or cannot give, and what tends to help.
The comparisons page now says how many of the 120 pairings are published so far, and pairs not yet available are clearly marked.
23 September 2026. All eight Dual comparisons have been tightened up: ILE vs SEI, ESE vs LII, EIE vs LSI, IEI vs SLE, ILI vs SEE, ESI vs LIE, IEE vs SLI and EII vs LSE.
Each now has a proper section on where the pair rubs: which of each type's strengths the other finds hard to take, how their rhythms differ, and what each does under stress. Older overview text that repeated itself, and in two places misdescribed how the exchange works, has been removed, each page links to both types' relationships pages, and each also looks at how the pair develops over time and how it works outside romance.
23 September 2026. Longer comparisons for all eight Dual pairs: ILE vs SEI, ESE vs LII, EIE vs LSI, IEI vs SLE, ILI vs SEE, ESI vs LIE, IEE vs SLI and EII vs LSE.
Each now explains what the two types supply for each other, function by function, including how each one covers the other's weakest spot, and how to tell duality apart from the two relations it is most often confused with. All eight Dual comparisons are now listed on the comparisons page, and each type's relationships page links to its Dual's comparison.
22 September 2026. Relationship pages for the four Delta types: IEE, SLI, LSE and EII.
That completes the set: every one of the sixteen types now has a relationships page covering what they need from a partner, where they get hurt, how to talk with them and what their Dual brings, with all sixteen of their relations laid out underneath.
22 September 2026. Relationship pages for the four Gamma types: SEE, ILI, LIE and ESI.
As with Alpha and Beta, each covers what the type looks for in a partner, where they are most easily hurt, how to get through to them and what their Dual brings, with all sixteen of their relations underneath. Delta is the last quadra still to come.
22 September 2026. Relationship pages for the four Beta types: EIE, LSI, SLE and IEI.
Like the Alpha pages before them, each sets out what the type needs from a partner, where they tend to get hurt, how to talk with them and what their Dual brings day to day, followed by all sixteen of their relations. Gamma and Delta are still to come.
22 September 2026. Four new pages on how the Alpha types behave in close relationships: ILE, SEI, ESE and LII.
Each one goes further than the relationships paragraph on the type's own profile: what the type needs from a partner and cannot make for itself, where it tends to get hurt, how to talk with it, and what its Dual actually does for it on an ordinary day. Each also lists all sixteen of the type's relations, from the most supportive to the most difficult, and for the four one-way relations it says which side the type is on. The other twelve types will follow.
22 September 2026. The small group finder now tries to settle disagreements instead of just reporting them.
When your temperament, club and quadra answers do not meet on one type, it asks one or two extra questions from other small-group systems: cognitive styles, communication styles, stimulus groups, argumentation styles or romance styles. Each time it picks whichever system best splits the types that are still level. The options are labelled A to D with no group names, because a name like "Aggressor" tells you how an option sounds rather than whether it fits. Every question also has a "Not sure, it's between two of these" option now, and each of the two picks counts for half.
The result says how confident it can be. Consistent means all three core answers agreed, and nothing has changed for anyone whose answers do. Leaning means the extra questions favoured one type over another. Unresolved means they did not, and it points you to the tools that can go further.
22 September 2026. A new page listing every small-group system in socionics, all 35 of them.
Quadras, clubs, temperaments and the rest look like an open-ended collection, but they are not. Any two of the fifteen dichotomies bring a third one with them, and together the three split the sixteen types into four groups of four. There are exactly 35 ways to do that. Eleven of them have names and pages on this site. The other 24 have never been described by anybody, and the page lists them as exactly that rather than making something up.
For each system the page also shows the relations inside every group, worked out from the data rather than assumed. In 31 of the 35, all four groups are built from the same relations, the way every quadra holds two dual pairs, two activation pairs and two mirror pairs. The table can be filtered by dichotomy.
22 September 2026. Two new pages on the groups that run in one direction: supervision rings and benefit rings.
Every small group on the site so far has been a set of four types. These two are chains. ILE supervises LSI, LSI supervises SEE, SEE supervises EII, and EII supervises ILE, which closes the loop, and saying it the other way round describes four relationships that do not exist. There are four such rings for supervision and four for benefaction, and each page sets them out in the direction the data gives, checked on every deploy.
The part I would read if you only read one section is where each step lands. Every supervision pair puts the supervisor's strongest function onto exactly the thing the supervisee is worst at, and all of the contact happens in the conscious half of the model. Every benefaction pair is the precise opposite: all of the contact happens in the unconscious half, and the benefactor's second function feeds the very position a dual feeds. That is why supervision is the relation people can describe without knowing any theory, and why benefaction is the one nobody can name from inside it while it quietly tilts the room.
Both pages carry a second layer from Viktor Gulenko, whose cognitive styles and project groups land on exactly these eight sets of types, and both treat that layer as a claim rather than a fact. Where a group has no name in any source I could reach, it is described instead of being given one.
22 September 2026. The comparison behind the bar at the foot of the page now counts people rather than pages.
The bar is being shown in two versions, one offering the free guide and one offering the type test, and until now the only count behind that comparison was of the bar appearing — so someone who read six pages counted six times and someone who read one counted once, which cannot say whether the two versions reached the same number of people. One anonymous count is now recorded the first time you are given a version, once, whether or not the bar is ever shown to you. The privacy page describes it and the small mark kept in your browser to make sure it happens only once: which of the two versions you were given, and nothing that identifies you.
22 September 2026. Closing the bar at the foot of the page now stays closed even when the comparison behind it is restarted.
The bar is being shown in two versions, one offering the free guide and one offering the type test. Each time that comparison was restarted — three times in a day last week, as the counting behind it was fixed — everyone who had closed the bar got it back, and everyone in the comparison was silently moved to whichever version a fresh coin flip gave them. Neither was intended. Restarting the comparison now only re-labels the figures: it does not reopen a bar you closed, and it does not change which version you are being shown.
Closing it has been reset once, today, on purpose, so that the restarted comparison begins with everyone on the same footing. If you had closed the bar before today you may see it once more; closing it again will hold for a week, on every page and in every tab, and will keep holding through any further restarts.
Two smaller things came with it. The bar was being counted twice under two different names each time it appeared, which is now one count under one name. And the link in each version now says which version it came from, so that a visit to the guide or to the test can be told apart — the privacy page covers what that does and does not record, and the answer is unchanged: which of the two versions you were shown, the date you last closed it, and nothing that identifies you.
21 September 2026. Closing the bar at the foot of the page now closes it everywhere, for a week.
It was meant to already, and mostly it did, but the counting behind it was re-sending itself: a record of a page being closed could arrive again on the next page you opened, which made it look as though the bar were being closed over and over and made the figures behind the current test unreadable. The counting is fixed, and the bar itself now keeps quiet for a week from the moment you close it, on every page and in every tab, rather than for a fortnight in principle and less than that in practice. Coming back to a page with the browser's back button no longer brings it back either.
One new thing is kept in your browser as a result, and the privacy page describes it with the rest: the date you last closed the bar, in a key of its own. It holds nothing that identifies you, and the comparison the bar is part of has been restarted so that the muddled figures are not carried forward.
21 September 2026. The bar at the foot of the page was cutting its own sentence in half, and now it wraps.
If you read this site in a browser window narrower than about 1140 pixels — which is most laptop windows that are not maximised — the strip at the bottom offering the free guide was ending mid-word. It read "a timing, requests and b" and simply stopped. The sentence it was trying to show is the one that says what is actually in the guide, so the offer was being made and then cut off at precisely the point where it told you what you would get.
The line now wraps onto a second line rather than being trimmed, and onto a third on narrower screens. Below about 680 pixels the bar stacks the way it has always done on a phone, which hands the sentence the full width of the bar to use. If you have your browser set to a larger text size than the default, that truncated it too, even on a wide screen; it now takes another line instead of losing the end.
One thing is worth saying plainly, because it is why this was worth fixing the same day. The bar is currently being shown in two versions, one offering the guide and one offering the type test, and the privacy page sets out what that keeps in your browser — which of the two you were shown, and nothing that identifies you. Only the guide version carried the over-long sentence, so the two were being compared against each other while one of them was unreadable for a good share of people on a laptop. The comparison has been restarted from scratch rather than carrying the muddled figures forward.
21 September 2026. The glossary had four things wrong about the Reinin dichotomies, and the type pages had all of them right.
A reader pointed out that the glossary's entry for Aristocrat and Democrat did not match what the type pages say, which it did not. It named Alpha and Beta as the Aristocratic quadras and Gamma and Delta as the Democratic ones. It is Beta and Delta who are Aristocratic, and Alpha and Gamma who are Democratic. The page on blocking groups says so correctly in the very sentence that links to the glossary for the definition, so anyone following that link landed on the opposite of what they had just read.
Checking the rest of the glossary against the type pages turned up three more. The entry for Static and Dynamic listed the wrong information elements: it had Ne and Se down as dynamic and Te and Fe as static, when it is the other way round. The static elements are Ne, Se, Ti and Fi. The entry describing the Reinin dichotomies as a set said there are fifteen of them beyond the basic four; there are eleven, which is how many each type page lists, and fifteen is the total once the original four are counted in. And two of the traits it offered as examples are not what this site calls them — one is listed on every type page as Yielding and Obstinate, and the other is a naming dispute rather than a trait, so the entry now simply names all eleven as the type pages do.
None of the type pages changed, because none of them was wrong. If you have used the glossary to work out where you fall on any of these traits, the tables on the type pages were right throughout, and the glossary now agrees with them.
While all this was open, the glossary also gained the eight Reinin traits it had been naming without explaining. It defined three of the eleven; it now defines all of them — Judicious and Decisive, Subjectivist and Objectivist, Process and Result, Carefree and Farsighted, Yielding and Obstinate, Tactical and Strategic, Constructivist and Emotivist, and Asking and Declaring — in the order the table on each type page lists them, so the two can be read side by side. The entries also say which traits line up with the quadras and which do not: only three of the eleven do, and expecting the other eight to follow quadra boundaries is the most common way to get them wrong.
The article on the Reinin dichotomies was checked at the same time and had four problems of its own, all now fixed. It was missing a section for Tactical and Strategic — it counted that trait in its summary and in the table on every type page, but never explained it, so there was one dichotomy the site listed everywhere and described nowhere. It has a section now. Its note on "Merry and Serious" said that was not a real Reinin trait; in fact it is an older name for the pair this site calls Subjectivist and Objectivist, so the note now explains the two names rather than dismissing one. The same section named the wrong functions for that split — it is introverted logic and extraverted ethics on the Subjectivist side, extraverted logic and introverted ethics on the Objectivist side. And the title said the article covered 15 traits beyond the basic four when it covers 11, which is what its own opening paragraph has always said.
Nothing in the article's type lists changed, because they were correct throughout — all 22 of them were checked against the 16 type pages. As with the glossary, every error was in the explanation rather than in who is what.
20 September 2026. New pages are now marked as new, the small group pages point at each other, and there is a feed you can subscribe to.
Most people arrive here from a search engine, land on the one page that answered their question, and leave from it without ever seeing the front page. The strip added earlier today does nothing for them. So the same information is now where they already are.
A page added in the last month carries a small New mark: on the small groups page, on the Reference, Discover and Tools pages, and on the sitemap. On the section pages the mark also appears against a whole section that has grown, so Small Groups is marked on the Reference page because eight pages appeared under it this month, not because that page itself is new. The mark goes away a month after the page went up.
The thirteen small group pages also have a panel of their own now, listing the four most recently added, so a page in that section tells you the rest of it exists. It used to show the latest essays instead, which is not what somebody reading about blocking groups is there for.
And there is an RSS feed. The icon is in the top right of every page, next to the search and theme controls, or under More in the menu on a phone. Subscribe to it in any reader and new pages arrive without you checking.
20 September 2026. The front page now tells you what has just gone up, whatever kind of page it is.
There has been a line of links under the opening blocks for a while, labelled Latest articles, and it did exactly what it said: it listed articles. Everything else that appeared on the site appeared silently. Four new pages on the small groups went up this month, and the four quadras each got a page of their own, and the homepage said nothing about any of the eight. The only place they were written down was this page, which is linked once, from the bottom of the site.
The strip now reads Latest updates and lists whatever is newest, so a new reference page or a new tool reaches the front page the day it goes up, exactly as an essay does. At the end of it there is a link to this page, which is where the rest of them are. The articles have not lost their way in: they have a card of their own in the grid just above.
Pages that were here before the site started keeping records are not dated and do not appear in it. Their real dates are not recoverable, and a made-up one would be worse than none.
20 September 2026. There is a page for blocking groups, which completes a pair.
A few days ago the site got a page on relaxation groups: four types made of two dual pairs taken from neighbouring quadras, who relax when you put them together. Blocking groups are the same idea with one word changed. Take the two dual pairs from opposite quadras instead, and you get a group where everybody still has their ideal partner in the room, and nothing whatsoever gets done. Those two are the only ways of putting two dual pairs together, so the set is now complete.
The interesting part is that you can say exactly why. Every member of a blocking group is being addressed on six of their eight functions, one each, and the two that are left out are the two they are best at. Their dual covers the things they are weakest at and most want help with, which is why the room is comfortable and nobody leaves. Their super-ego partner is aimed squarely at the two functions they defend, which is why every suggestion dies the moment it is made. If you have ever sat in a committee that everyone enjoys and that has achieved nothing in two years, that is the shape of it, and the page ends with how to break it.
20 September 2026. Five of the relation pages listed the wrong type pairs, and they have been corrected.
Extinguishment, super-ego, quasi-identity and activation each named the wrong eight pairs in their questions section, while the table further down the same page had them right, so each page disagreed with itself. Business said the relation holds between types in the same club, which it never does: business partners are in different clubs, as that page's own summary at the top already said.
If you read any of those five and worked out your own pairings from the questions section rather than the table, the table was the one to trust. All five now agree, and there is a check on every deploy comparing those lists against the source data so they cannot drift again.
20 September 2026. Two more small groups have pages: health groups and bouquet groups.
A health group is four types whose leading function is the same kind of thing: the four intuitive types who lead on possibility, the four sensing types who lead on the present, the four ethical types and the four logical ones. Socionics literature reports that four people assembled this way tire each other out less than almost any other group of four, and the page sets out why the structure would predict that, which is that nothing any of them does lands anywhere the others have to defend. It also sets out plainly what that claim is worth. The source is a lecture by Grigory Reinin, who gives no sample size, no method and no control for it, and who says himself that the small groups are barely studied and that this effect does not follow from the theory. The page says all of that, and nothing on it is medical advice.
Bouquet groups are the temperaments under a second name, and the page exists because the second name is attached to a different question. The name, incidentally, is not about flowers: Reinin called them bouquets because he thought each group shared a bouquet of diseases, a claim the page reports and declines to pass on. The temperaments page tells you what an EP or an IJ is like. The bouquet page tells you what happens when you put four of them together, which is that the room goes flat: they share their weak spots as well as their strong ones, so there is nobody present who holds what the others are short of. If you have ever been in a team where everyone gets on and nothing moves, that page is about it, and it ends with the fix.
20 September 2026. There is a new page on the relaxation groups, the small group made of two dual pairs from neighbouring quadras.
Reinin called them square groups, and the claim attached to them is that four people put together this way relax without anybody arranging it. The page sets out all four, what they have in common, and why the structure is unusually easy to be in: everyone has their dual in the room, there is no relation that puts one person above another, and none of the four difficult relations appear at all.
It also corrects something. The description of these groups in general circulation, including on the older Socionics Demystified site, says all four contain the same three relations. They do not. The two made of irrational types are held together by semi-duality and kindred; the two made of rational types by business and mirage. Both are comfortable, but not in the same way, and the page says which is which. There is also a section on where the relaxation claim looks weakest, written as a prediction that could be shown wrong rather than as settled fact.
The semi-dual and kindred pages now point to it from the part of each that explains the cross-quadra structure, and prev/next now runs in a fixed order across all ten small-group pages instead of an arbitrary one.
20 September 2026. Corrected the questions at the foot of the kindred page, which had the relation backwards.
The answer there said kindred partners come from the same quadra and have opposite extraversion. Both are the wrong way round: kindred pairs always bridge two neighbouring quadras, and they always share extraversion. It also said the difference from duality is that kindred partners share rationality, when in fact dual pairs share rationality too. The main text of the page has always been right, so if you read the page itself you were not misled, but the questions are what search engines quote and that is what they were quoting.
All three are fixed, and the rest of the relation pages have been checked for the same mistake. There were none.
20 September 2026. The nine cards on the homepage now each carry an icon, and two of them look right in dark mode for the first time.
The grid is the main way into the test, the compatibility tools, the articles and the rest, and nine blocks of text set out identically gave you nothing to aim at. Each card now opens with a small drawn mark for what it leads to — a clipboard for the test, a grid for the compatibility matrix, two overlapping circles for the Dual, a bolt for the conflict tool — so the one you want is findable at a glance rather than by reading all nine.
Two dark-mode faults went with it. The Articles card had a border so faint it read as a gap in the grid, and the Socionics World card was the only one whose description changed colour when you moved the mouse over it. Both now match the other seven. Nothing has moved and nothing in light mode has changed.
20 September 2026. There is now a sitemap listing everything on the site, linked from the bottom of every page.
The footer covers the main sections, but most of what is here sits underneath one of them: 52 type comparisons, 28 famous people, 25 articles, the 16 type profiles, the eight functions, the relations, the small groups and the rest. The sitemap puts them all on one page, grouped exactly as the footer is, so you can see the whole site at once instead of working down through it a section at a time.
It builds itself from the site rather than being kept by hand, so a new page appears on it the day it goes up, and if a page moves between sections in the navigation it moves here too. There is also a check that runs on every deploy and refuses to publish if the sitemap and the machine-readable one behind it ever stop agreeing, so neither can quietly fall out of date. The link sits next to the copyright line at the very bottom of the page.
20 September 2026. Every page now shows the trail back to the homepage, and there are three new pages that gather the site into Reference, Discover and Tools.
Until now the line at the top of a page named the section it was in but gave you nowhere to click except home. The section is now a page of its own: Reference for the guide, the types, the functions, the relations, the small groups, the comparisons, Model A and the glossary; Discover for the articles, the typings, the MBTI bridge and the reading; Tools for the test, the compatibility tools and the rest. Each one says in a line or two what every page under it covers, which is the thing the footer could never do.
Two things have moved in the footer. Type comparisons are now under Reference rather than Discover, since they are built from the type data rather than written as articles. And the link to the resources page now reads "Resources", which is what the menu at the top of the page has always called it. Nothing has changed address, and every link that worked yesterday still works.
Two tools that were easy to miss are now listed with the rest: the type disambiguator, for when you have narrowed it to two types and cannot get past it, and your people list, which keeps the people you know and their types in your own browser. Both existed already and neither was in the footer, so you had to come across a link to find them.
The four headings in the footer are no longer just labels: Reference, Discover and Tools now take you to the page listing everything under them. About has no such page, so it stays as it was.
The resources page is shorter. It was a long list of links to everything, and now that Reference, Discover and Tools each carry that list, it says what is where in a few lines and points you at whichever one you want. Nothing has been removed from the site, only from that page.
20 September 2026. The logo is now at the bottom of the page as well as the top.
When the site got its logo last week, only the name in the top left was replaced. The footer carried on showing the old set-type version of the name, so the quadra circle appeared once per page rather than at both ends of it. Both now use the same drawn lockup. Nothing has moved and it still takes you home.
20 September 2026. There are now links to the TikTok and YouTube channels in the footer, under the credits on the left.
Both have been running for a while and neither was linked from anywhere on the site, so the only way to find them was to already know they existed. They are two small icons rather than a row of buttons, sitting with the rest of that column, and nothing else in the footer has moved.
The videos cover the same ground as the site, in shorter form: individual types, how two types get on, and the questions that come up most often when people are working out their own type. If you would rather read, nothing here changes, and the written material remains the point of the site.
20 September 2026. The site now counts how its search is used, and the privacy policy has a new section setting out exactly what that means.
Search still runs entirely in your browser — the index is downloaded to your device and queried there, and what you type is never sent anywhere to be answered. That has not changed and is not going to. What is new is that four things are counted: that search was opened, the words searched for once you stop typing along with how many results they returned, which of the suggested searches was picked, and which result was clicked.
The middle one is the only place on this site where something you typed is recorded, so it is worth saying how it is handled. It is sent once the search settles rather than as you type. It is lowercased and trimmed, so the same search counts once. Anything over 64 characters is not sent, and neither is anything containing an @ or a run of six or more digits — an email address or an order number is not a plausible search here, and those are the shapes a personal detail takes when it goes into the wrong box.
The reason for any of it: searches that come back with nothing are the clearest signal of something you expected to find here and did not, and those are what tell me what to write next. Nothing is tied to you, no cookies are set and no profile is built. The privacy policy has the full detail, and any tracker blocker stops the counting while leaving search working exactly as it does now.
20 September 2026. Search now takes you to the part of the page you were looking for, rather than just to the page.
Search for duality and the Duality relation still comes up first, but underneath it you now get the sections within it — how the dynamic works, how it feels in practice, what it looks like at work — each one a link straight to that part of the page. On long reference pages that is usually what you wanted. Nothing about how you search has changed: the button in the header, ⌘K on a Mac or Ctrl+K elsewhere, the suggestions when you open it, and Escape to close.
It should also feel quicker everywhere else on the site. The search tool used to be downloaded by every visitor on every page, whether or not they ever opened it — about 131KB of it. It is now fetched the first time you actually search, and not at all if you never do.
20 September 2026. There is a guide on the downloads page again. The 16 Intertype Relations is ten pages on how any two types stand in relation to each other: an introduction to the three families the relations fall into, then the relations themselves — what the structure underneath each one is, the dynamic it tends to produce, and one line you can hold onto about how it usually goes, from Duality at "most complementary" to Conflict at "most difficult". Fourteen entries cover all sixteen, because Supervision and Benefaction are each described from both sides.
For a while this was the thing you got in exchange for an email address. A longer book took its place, the offer moved with it, and the guide quietly stopped being available anywhere — it has been sitting on the server since, with nothing on the site pointing at it. It is now a direct download like the three reference cards: no email address, no account, nothing to fill in first.
The page has been rearranged around it. The guide is at the top with a summary of what is inside; the three one-page cards — the 16 types, the relations and the four quadras — are below it under their own heading. The cards remain free to use and share with credit; the guide is drawn from Your Social World Explained and is for personal use, so if you want to pass it on, send people the page and they can take their own copy.
One thing worth separating, because the page now mentions two guides: the relationship-patterns guide at the foot of that page, the 76-page one, is a different book and still arrives by email. The relations guide is the direct download.
20 September 2026. Each of the four quadras now has its own page: Alpha, Beta, Gamma and Delta. Until now there was one quadras page giving each of them a paragraph, and then five separate articles that each went through all four again from a different angle: their social philosophies, how each one decides what is true, how each relates to economic life, and so on. If you wanted to know about Delta you had to read five pages and assemble it yourself.
Now you read one. Each quadra page covers what that quadra values, who is in it, two people who illustrate it, its characteristic social philosophy, what it treats as good evidence, what it wants from working life, its stage in the way ideas move through the socion, and the two clubs inside it. The quadras page itself keeps the things that are genuinely about all four at once: what a quadra is, the two axes they divide along, and why people from different quadras so often talk past each other.
The five articles that were folded in now take you to the relevant part of the quadras page, so any link or bookmark you have still works. Nothing was thrown away: every distinct idea in them is on one of the five pages, just not repeated on four of them. One article did not move, your quadra and club together, because it answers a question the other pages do not.
One thing was corrected along the way. Two of the retired articles described Gamma and Delta's core themes differently from the reference data the rest of the site is built on. The pages now use the canonical wording, and the site's own tests check all four quadras against that data on every build, so they cannot drift again.
Two other things were put right at the same time. Your quadra and club together described the function each pair of types shares as the one they both lead with. Only one of the two leads with it in every case, and the article now says which, and which of them uses it as their creative function instead. And when the test or the group finder tells you your quadra, the link now takes you to that quadra's own page rather than to its paragraph on the overview.
19 September 2026. The Model A diagram on each of the 16 type pages now does two things when you touch it.
Tap or click any of the eight rows and it lights up along with its partner — the two halves of a block always belong together — while the rest of the diagram steps back, and a line appears underneath saying what that position actually is. Tap it again, press Escape, or tap anywhere outside to put it back. It works from the keyboard too: tab to a row and press Enter or Space. The links on the function names are untouched and still take you to the function's own page.
Underneath there is a button that says "Show your Dual's strengths", naming whichever type that is — SEI on the ILE page, LIE on the ESI page. Press it and positions 5 and 6 are picked out and relabelled as your Dual's leading and creative functions, because that is what Duality is: the two things you most need from someone else are the two things they lead with. There is a line under the diagram saying so, with a link to the Dual relation if you want the longer version.
Which type is whose Dual is read from the same relations table the rest of the site uses, and the site now refuses to build at all if that table and the function stacks ever stop agreeing with each other — so the diagram cannot quietly start telling you something that is not true. The diagram itself is unchanged: same shape, same colours, same links, and if you have JavaScript turned off you get exactly the page you got before, with no button and nothing missing.
19 September 2026. You can now have your test result emailed to you. Under the result there is a short form asking for nothing but an address, and what arrives is the result itself plus a check over the following week on whether it is right: the small groups to test it against, the types yours is most often confused with, and how to tell them apart. It is optional in every sense. The result is shown in full on the page whether or not you use it, nothing on that screen is held back behind it, and the page works exactly as it did before if you ignore the form entirely.
What is recorded alongside your address is your result and the facts that go with it: your type and its name, your Dual, the two types yours is most often confused with, and how many of the four axes came out close. All of those except the last are the same for everyone who gets that result, and they are there because the emails are about them. Your individual answers to the twelve questions are not recorded and are not sent. The privacy policy says the same thing at more length, and it has been updated for this. If you use the form and then retake the test, the page says your result was already sent rather than asking again.
The form that used to sit in that spot, the one offering the free guide, has moved off the result screen rather than been removed. It is still on the relations, comparison and MBTI pages, and the guide page is unchanged. Two forms asking for the same address on one screen was one too many.
19 September 2026. The hub now lists everything it says it does. It had no mention of the typing reports or the practical guides, and both now have a card there. The count of articles on it was also out of date, and is now taken from the articles themselves so it cannot drift again.
19 September 2026. The press page has been brought up to date. It had not been touched since it was written, and it was still describing the Socion app as something that had not launched yet, still counting 372 pages and 136 type comparisons, and still saying nothing at all about the typing reports, the practical guides, the free tools or the published quiz results. The numbers are now the real ones: 347 pages, 120 comparisons, 86 public figures typed. There is also a new section of logo files, the author photo and the sharing image, all free to use by anyone writing about the site.
19 September 2026. The wording around the ebooks is the same everywhere now. A few places still introduced them by the name of the old shop, or sent you off to browse a collection that was never called that anywhere you landed. They are the practical guides, and that is what every line and every link calls them. The panel on the books page and the resources page was still headed "the books" while describing the guides underneath it, which is now fixed. Nothing has moved, nothing has been added or taken away, and the guides, the prices and the page itself are unchanged.
19 September 2026. The practical guides page reads a little more plainly. The link to it at the foot of every page is now called Practical Guides, so it cannot be mistaken for the Amazon books listed a line above it. The page now says who wrote the guides: they are by Spencer Stern, who writes the Socionics Made Simple series and runs this site. The button on each one says "View the guide and get it on Whop", which is what it does: it takes you to that guide's own page, where you can read about it and buy it. The guides, the prices and the order they are listed in are unchanged.
19 September 2026. The Guides menu item now opens a page on this site: practical guides. It used to send you straight out to a shop. Each guide is now described here in full before you go anywhere, with the cover, who it is written for, the situation it deals with, what you will be able to do by the end, how long it is and what it costs. The button on each one takes you to that book's product page on Whop, where the guide is sold and delivered as before. Nothing about the guides themselves has changed, and nothing on this site has moved behind a payment.
18 September 2026. There is a new tool: the group dynamics mapper. Everything else here that deals with relations deals with two people. This one takes up to eight. Add the people in a team, a household or a group of friends, give each one a type, and it shows every pair at once: a diagram with a coloured line for each pairing, the full list underneath, how the group splits by quadra and by club, which pairs are in one of the relations this site files as difficult, and who holds the most supportive relations of anyone in the group.
Four of the sixteen relations do not run the same way from both ends, and the tool never flattens those into one label. A Benefaction pair is shown as both sentences, so you can see which of the two is the benefactor rather than being told the pair is "Benefaction" and left to work it out. Every relation in the table links to its own page.
Names are optional, and if you leave one out the person shows as Person 1 and so on. Nothing is sent anywhere: the whole thing runs in your browser, and the only place a name is ever written down is the share link, if you choose to copy one. That link carries the group, so you can send someone the same map.
18 September 2026. The compatibility matrix has had its wording corrected, and the fault was worst for anyone using a screen reader. Each of the 256 cells carries a spoken description, and those descriptions were being built out of the shortened text that fits inside a cell. So a cell was read out as "ILE is ILI's Extingmt", which is not a word. On the 64 directional cells, the ones this page exists to sort out, it said "ILE is LSI's Supervises", and on the other 192 it said things like "ILE is ESI's Conflict". That last one is the interesting error: saying someone is your Benefactor makes sense because it tells you which end of the relation you are standing at, but twelve of the sixteen relations read the same from both ends and there is no end to stand at. Those now read "ILE and ESI: Conflict", and every relation is given its full name. The grid itself, the colours, the legend and the filters are unchanged.
18 September 2026. The site has a logo. The name in the top left has been set type since the site was built; it is now a drawn mark and wordmark together — the circle to the left of the name is the four quadras, with one segment picked out in gold. The same mark is the favicon, so the tab now shows something other than the two letters SI, and it comes with the icon set a phone uses when the site is saved to a home screen. Nothing has moved: the logo is the same size as the name it replaces and still takes you home. The picture that shows up when a page from here is shared on Facebook, LinkedIn, Slack or X has the logo on it now too; it is otherwise the same card it was. The one visible knock-on is that the menu now folds into the hamburger button on narrower laptop and tablet screens than before — below about 1000px rather than 820px — because the logo is wider than the name it replaces and the row of links had nowhere left to take the space from. Nothing has been removed from the menu, and every link is in it either way.
18 September 2026. The Books item in the menu is now Guides, and it goes to the guides rather than to the books. Those are two different shelves and the site had been blurring them. The Socionics Made Simple series and Your Social World Explained are on Amazon, and the books page is still where they are described, with every cover and every link it had before. The practical guides are a different thing: short guides that take one situation each, a message that keeps being misread or a playground friendship to repair, and work through it start to finish. They are read online as soon as you buy. The front page now says which is which instead of describing one shelf and linking to the other.
You will also see one guide suggested by name in the few places where it fits what you are already looking at: under a result on the conflict tool, at the foot of the Duality and Conflict pages, and after you ask for the free guide. It gives the title, what the guide gets you and what it costs, so you can tell whether it is worth a click before you make one. No page shows you more than one. Everything on this site is still free to read, and nothing has been moved behind a payment.
18 September 2026. The test now gives you a different next step depending on whether it actually worked. Each of the four axes is decided by three questions, so it comes out either 3 to 0 or 2 to 1, and a 2 to 1 is one answer away from a different type. If any of your four came out 2 to 1, or if you tell the page the result does not feel right, the block under your result says so and offers the written typing report, which reads your own words rather than a score, with the free two-type check on the button directly below it as the option that costs nothing. If all four came out 3 to 0 and you have not said otherwise, there is nothing left for twelve questions to separate, so that block offers the short volume for your type instead, and the button for separating two types is not shown at all, because at that point it would be arguing with the line above it.
What the uncertain version says depends on how open it is. One close axis names both types outright, because there are only two and naming them is more useful than a count. Two or more says how many types are still consistent with your answers, which is four at two close axes and all sixteen at four, and then names the types yours is most often confused with rather than listing every code. Everything below that block is where it was: the small groups, your full profile, the "if this result feels wrong" list, Socionics World and the retake. Answering the "does this feel right?" question switches the block in place, with no reload and nothing else on the page moving.
The description of the test in search results also said it was a "function-based questionnaire", which it is not. It measures the four dichotomies, and now says so.
17 September 2026 — The site has a dark theme. There is a new button in the header, to the left of the search icon, and it has three settings rather than two: it follows your device by default, and you can override that to light or to dark. Left alone it does the sensible thing, which is to be dark if your phone or laptop is dark and to change with it when your device switches in the evening — so if that is what you wanted, there is nothing to press. If you do press it, the choice is remembered in your browser and nowhere else; the privacy policy covers the one word it stores, and cycling the button back to the default setting removes it.
It is the site's own colours rather than an inversion. The navy the header has always used is now the page behind everything, the cards sit above it, and the gold is the same gold. Everything is included — the test, the compatibility matrix, the people list, the tools, the forms — and not just the articles.
Several things that were wrong in the light theme are fixed at the same time, because building the dark one is what found them. The gold buttons inside articles were rendering their text in link blue on gold, which is close to unreadable and had been for a while. The primary buttons on the tools — the guestbook form, the small group finder, the disambiguation tool — were white on gold, or white on the quadra colour of whichever type page you were on, and none of those combinations was legible enough. And the small count pills on the articles index were white on blue. All of them now have dark text. Nothing else about the light theme has changed: I checked it page by page against what was there before, and on most pages the result is identical to the pixel.
17 September 2026. The test now publishes what it has been producing. What everyone else got is the full distribution of every completed run since June: all sixteen types ranked by how often each comes up, the same results split by quadra, and the honest reading of both. The headline is that introverted types take 82.3% of results and the top two, ILI and IEI, take nearly a third between them, which says a great deal about who finds a Socionics quiz and finishes twelve questions and nothing whatsoever about how common those types are in the population. I would rather put that in writing on the page than have the numbers quoted as though it did. The more useful figure is the second one: 94.7% of results came out a single answer away from a coin flip on at least one of the four axes, which is what a twelve question forced choice instrument does and is the best argument there is for treating your result as a starting point. The oddity, which I cannot explain, is that the four quadras come out close to level despite the type level spread being twenty to one from top to bottom. Nothing on the page identifies anyone: the only thing the test records on finishing is the type code, whether you arrived by taking it or on a shared link, and whether the result was close.
17 September 2026 — The test result screen has been rebuilt around one next step instead of nine. Under your result there is now a single button, and it names the two types your answers are most likely to have confused — "Not sure? Separate ILE from LII" — going straight to the tool that works through the questions separating that specific pair. Everything else that used to compete with it for your attention is still there and is now below: your Dual, your full type profile, the identical-pair comparison, Socionics World and the typing report sit together as a row of equal options, and the small group finder, the book for your type and the retake control are behind a "More from this result" heading you can open. Nothing has been taken away. The reason for the change is that a twelve-question test cannot separate types that share three of their four letters, so the honest next step after a result is the instrument that can, and a screen offering nine equally-weighted destinations was not pointing at it.
17 September 2026 — The free guide is now offered where you are already reading about the thing it covers, with the email field on the page rather than a click away. It appears on duality, conflict, romance styles, every type comparison and every MBTI bridge page, and on the test result under your result. One field, your email, and a tick box. The guide's own page is unchanged and still where it is sent from.
The description of it has changed everywhere at the same time, and that is the more useful half. It is a 76-page book with two worksheets in it — a three-cycle mapping worksheet and a timing, requests and boundaries checklist — and nowhere on the site said so. It was described as "a free guide to using Socionics to make sense of communication, attraction and conflict", which is true and makes a 76-page book sound like a leaflet. To be clear about what it is not: it does not give you a reading of your own type's functions, strengths or blind spots — that is what the Socionics Made Simple books are — and it does not reason from anything you have told the site, which is what the typing report does. It is about patterns between people and how to read one you are actually in.
17 September 2026 — The typing report is now linked from the places where you would plausibly want one, and each of those links says what it costs and what you get rather than "expert typing". You will see it at the end of the disambiguation tool when it has either separated your two types or told you it could not, at the foot of each of the 16 type profiles, phrased around the two types that profile is most often confused with, and in the row of options on the test result. Every price on the site now comes from one place, so when the half-price offer on the Standard report ends, it ends everywhere at once rather than lingering on whichever page was missed.
17 September 2026 — Two link fixes. A link in the Reinin dichotomies article was written relative rather than absolute and sent readers to an address that does not exist; it and one other of the same kind are fixed, and a sweep of every link on all 350 pages now finds none of that pattern left. Two internal working documents were also being published as pages by accident and are no longer on the site. Neither was linked from anywhere or listed in the sitemap.
17 September 2026 — The test asks, once it has given you a result, whether the type feels right. Answering "Not sure" used to be the one answer that led nowhere: it told you the three small groups were the quickest check and left you to scroll down and find them. It now offers the small group finder, which asks those three groups as questions instead of handing you three pages to read, and says so when its own three answers disagree with each other. "Not sure" is the ordinary answer after twelve questions rather than an unusual one, so it was the wrong place to stop.
16 September 2026 — If you have set your type, the comparison pages and the famous people pages now say what they mean for you, the way the type, function and relation pages already did. On a comparison page the line above the two types tells you where you stand relative to both of them — and if one of the two is your own type, it says so and names the other as your Dual, your Conflictor or whatever it happens to be, with a link straight to the tool that works through the questions separating those two specific types. That is the situation those pages exist for: most people reading a comparison are not weighing up two other people, they are trying to work out which of the two they are. On a famous person's page it tells you what that person's type is to you — that Adam Sandler, typed IEI, is your Benefactor, and which end of that you are on, since four of the sixteen relations do not run the same way in both directions. This takes the personalised line from 38 pages to 244. Nothing has been removed or moved to make room for it, the pages read exactly as they did if you have not set a type, and as everywhere else your type stays in your browser and is never sent anywhere. There is a control to clear it on every strip.
16 September 2026 — The guestbook is now findable. It has been reachable only from the privacy policy, which is nobody's route to anything, so it is in the footer with the rest of the pages about the site, and there is a line on the contact page for the messages that do not need a reply — if you just wanted to say the site was useful, or what you would like to see on it, that is a note rather than an email, and the guestbook takes it without an address. It is also offered at the end of an article if you have said the article answered what you came for, which seemed like the moment somebody might actually want it.
16 September 2026 — You can now leave a note in the guestbook yourself. It has been there for months with a handful of entries and a line asking you to email instead, which almost nobody was going to do. There is a form on it now: a note, and optionally your name and your type. It asks for no email address and there is nothing to confirm afterwards.
Two things are worth saying plainly, and the form says them above the box. Nothing goes up automatically — every note is read first and added by hand if it suits the page, so yours will not appear straight away and may not appear at all. And because anything published appears as you wrote it, please do not put anything private in it: an address, a phone number, or someone else's name. If you want a reply rather than a note on a page, the contact page is the right place, since nothing written in the guestbook reaches you back. The privacy policy now covers both forms, including how to have an entry removed.
16 September 2026 — The compatibility matrix had a sentence telling you to read it backwards, and it has been corrected. The table maps all 256 pairings, and each cell says what the row type is to the column type — so the ILE row in the LSE column reads "ILE is LSE's Beneficiary". The note under the table always said that correctly and the cells were always right. The paragraph above it said the opposite: that a cell showed what the column type was to the row type. For twelve of the sixteen relations that makes no difference, because they read the same from both ends. For the other four — Benefactor and Beneficiary, Supervises and Supervised — it inverts the meaning entirely, and those 64 cells are the ones the table exists to make clear. If you have used this page to work out which way round a supervision or benefit relation runs, it is worth checking again. Every cell now also says what it means in full when you hover it, rather than just naming the relation.
The page has also been given the thing it has been claiming to do. It has said "click to pin" since it was built, and clicking only ever rewrote one line of text that the next click replaced. You can now pin as many cells as you like; they stay marked in the grid, they are listed underneath with a link to the relation and to the full comparison of the two types, and they are not hidden when you filter. Your pins and filters live in the address bar, so reloading keeps them, the back button works, and a bookmark comes back to the same view. Nothing is stored on your device for this.
16 September 2026 — Each of the articles now asks, at the end, whether it answered what you came for. Yes or not really, and nothing else — there is no comment box and nothing you write goes anywhere, because there is nothing to write. If you say not really, it asks which of three things was the problem: you needed the basics first, there was a term you did not know, or it simply was not about your question. Each of those is a link to the page that should help — the beginners' guide, the glossary, or the contact page — so the answer is useful to you as well as to the site. The reason for asking is that there has never been any way of telling the difference between an article that answered someone's question and one they opened and gave up on. Which of the three reasons people pick is the part that says what to do about it: an article that keeps failing readers who needed the basics needs a different opening than one failing readers who hit an unexplained term. As with the typing votes, your answer is remembered in your browser so you are not asked twice, is counted in aggregate, and is never published or attached to you.
16 September 2026 — You can now say whether you agree with a typing. Every one of the famous people pages argues that someone is a particular type, and until now there was nowhere to put it if you thought the argument was wrong. There is a line under the case on each of them: agree, disagree, or name the type you would give instead. The third of those is a list of the other fifteen types rather than a box to write in, which is deliberate — this is a count, not a comment section, and there is nothing in it to moderate or publish. If you do name a type, it offers you the comparison page setting that type against the one the page argues for, which is where the two readings actually part company. Your answers are remembered in your browser so you are not asked twice on the same page, and they are counted in aggregate for working out which of these write-ups need another look. Nothing is published, nothing is attached to you, and the privacy policy covers what is stored, as it now does for everything this site keeps in your browser.
16 September 2026 — There is a new tool: the small group finder. The site has been saying for a long time that the fastest way to check a type is the small groups — recognising yourself in a group of four is a far easier task than choosing between sixteen — and the test result has linked to all three of them since it launched. As three pages to read. This asks them instead. Three questions: how you pace yourself, what you pay attention to, and which sort of room you would rather be in. Each has four answers, and between them they name a type. The part worth knowing is that two of the three questions would have been enough on their own, and the third is there to disagree with them if it is going to. When all three land in the same place you get a type and a reason to believe it. When they do not, you get told so, along with which type each pair of your answers pointed at and, where the two candidates rule each other out completely, that they do. A tool that always agrees with itself is not checking anything, and this one is meant to be a check. It sits alongside the test rather than replacing it: they work from different parts of the theory, so two of them agreeing is worth more than either on its own.
16 September 2026 — The disambiguation tool now has questions for 30 of the 120 type pairs, up from three. More usefully: it now covers every one of the 29 pairs the test actually points people at when it says two types are easily confused, where before it covered two of them. If you have been sent there from a test result, from a comparison page, or from your people list and been told there were no questions for your pair, there is a good chance there are now. The sets are written one pair at a time from the function stacks that separate those two types specifically, not generated, and they can still tell you that the questions did not separate the two — that remains a real answer rather than a failure. The pairs that still have no set say so plainly, as before.
16 September 2026 — The "Share on X" link has been taken off the test result screen. It had been there since the result screen was built and nobody had ever used it: of 2,072 tests completed between March and September, not one person opened a result someone had shared with them. Your result is still in the address bar when you finish, so you can still copy or bookmark that link and it will still open on your result — the button that offered to write the tweet for you is simply gone, and the row under your result is two links instead of three.
16 September 2026 — The privacy policy now sets out what this site keeps in your browser, which it should have done from the moment the first of those features shipped. Three things are stored there: the type you have set, your people list, and whether you have closed the email bar at the foot of the page. The new section names all three, says exactly what each one holds — down to the fact that the names on your people list can be up to 40 characters and that there can be up to 50 of them — and states plainly that none of it is ever sent anywhere. It is not transmitted to this site, is not visible to the analytics tools, is not accessible to the site owner, and is not shared with anybody. It also says how to remove each one, that it is per-browser and per-device so setting a type on your phone does not set it on your laptop, and that every feature using it is optional and the site works without it. Nothing about what the site does has changed here; this is disclosure of behaviour that has been in place for months and was not written down. The one thing that was actively misleading has been fixed too: the analytics section said the two analytics tools use no cookies and no local storage, which is true of those tools and was easy to read as a statement about the whole site. It now says which it means and points at the new section.
16 September 2026 — The test has had its biggest change since it launched, and is now version 3.0. Two things are different. The three Extraversion/Introversion questions have been rewritten, because the first six months of results showed that axis coming out 83% introverted while the other three were close to balanced — every introvert type ranked above every extravert type. One axis that lopsided next to three that are not is a fault in the questions, not a fact about the people answering them, and looking at them it was clear enough why: one offered "work the room" as the only extraverted answer, which few people will claim about themselves; one asked how you feel after a full day of back-to-back meetings, which is draining for almost everyone whichever way they lean; and one asked about a difficult problem, which pushes everybody towards going quiet. All three now give you two answers that are equally worth picking, and ask which direction your energy goes rather than how much you can take. If you have taken the test before, those three answers may come out differently now. That is the intention.
The second change is that the test is more honest about how sure it is. Each of the four axes is decided by three questions, so a score is either 3–0 or 2–1 — which means the test could never produce a tie, though the old result page offered "came out tied" as one of the things it might tell you. It also called a result "borderline" whenever any single one of the four came out 2–1, which was true of 87% of results: the warning meant to mark the unusual case was being shown to almost everyone. The result page now counts how many axes were close and tells you what that actually costs you, because each close axis doubles the shortlist — one leaves two types that fit your answers, two leaves four, and four close axes leave all sixteen, which is the test admitting it has not separated anything. You get the close axes named and the number of types still in play, instead of one word that meant the same thing whether you were one answer from certain or the test had told you nothing.
One consequence: the line added earlier today saying how often each result comes up is off the page for the moment. Those figures were measured on the old Extraversion/Introversion questions, so they describe a test that no longer exists, and publishing them against the new one would be quietly wrong. It comes back once enough people have taken version 3.0.
16 September 2026 — The test now asks you a question back. Once you have a result, there is a single line under it — does this type feel right? — with three answers: yes, not sure, no. Whichever you pick, your result does not change; what changes is where the page sends you next. "No" is the most useful answer you can give and gets the most specific reply: it takes you to the tool that works through the questions separating your result from the types it is most often confused with, because a twelve-question test mixes up a predictable set of pairs and yours has its own. "Not sure" points you at the three small groups below the result, which are the quickest way to check a type against yourself. "Yes" takes you to the full profile, to read for the parts you did not expect rather than the parts you did. Nothing about the answer is stored on your device, sent anywhere with your name on it, or added to your record — it is counted in aggregate and nothing else, which is also why the question is only asked after a test you took yourself. Opening somebody else's shared result link does not ask you to pass judgement on their type. The point of asking at all is that the test has never had a way of knowing when it is wrong: it has been handing out sixteen results since it launched with no signal at all about which of them people recognise and which they do not, and the honest thing to do about that is ask. There is a second, quieter addition alongside it: a line saying how often your result comes up — what share of people who take the test get it, and whether that makes it a common result here or a rare one — with the number of runs it is drawn from and the period they cover. It reads, for instance, "10% of people who take this quiz get LII — the third most common result here", across 2,072 completed tests between March and September 2026. Those are the test's own results and nobody else's, which is the only thing a line like this can honestly claim; it is not a statement about how common any type is in the world. Where two types have come out exactly level the line says they are tied rather than putting one above the other. It is a figure from a fixed sample rather than a live counter, so it changes when the figures are refreshed and not otherwise, and it stays off the page entirely if the sample behind it is ever too small to quote.
15 September 2026. The site has a new top menu and a clearer front page. The menu is now Learn Socionics, Types, Relationships, Resources, Books, Typing Reports and Socion. Nothing has been taken away to make room: the functions, the small groups, the articles, the comparison pages and the test all still have their place in the menu on a phone, in the footer, and on a new resources page that gathers everything free on the site in one list: the reference pages, the printable cards, the tools and the writing, grouped by what you are trying to do. The front page now says what this site is before it says anything else, and offers two ways in: start learning, which takes you to the beginners' guide, or have a look at the books. The Books item in the menu goes to the practical guides, a separate site where they are browsed and bought; it opens in a new tab so you do not lose your place here. A link to the guides now appears at the foot of the type profiles, the relations pages, the function pages, the small group pages, the comparison pages, the articles and the guide, one line, at the bottom. The guides are a different kind of book from the type-by-type series: each one takes a single situation and works through it, rather than explaining a type. The books page says which is which, so you can tell before you click. Everything on this site stays free. The reference material is not a sample of the books and nothing has been put behind a payment; the books are the same ground at greater length, for anyone who wants it that way. The books page is still here and still lists the whole Socionics Made Simple series with every link it had before.
13 September 2026 — Signing up for the free guide now happens in one place. The bar along the bottom of the page, the block at the end of each article and the strip on downloads used to each carry their own email box, supplied by our email provider. All three are now a button that takes you to the free guide page, where you enter your address once. They also now describe the right book: all three were still offering the intertype relations reference, which has not been what that page sends for some time. They offer the relationship patterns guide instead, around 16,500 words on using Socionics to make sense of communication, attraction and conflict. If it was the relations reference you were after, it has been a free download on the downloads page all along and needs no email address at all. The bar is otherwise unchanged: it still waits until you have been reading a while, still closes for three months when you close it, and still stays out of the way on the test, the typing pages and the legal pages. The practical difference is that no third-party form code loads on the site any more, which makes every page a little lighter and means the signup either works or visibly does not, rather than quietly failing somewhere we cannot see. Some people have been entering an address and hearing nothing back, and this is the first step in finding out why.
12 September 2026 — There is a new page for anyone who has ever wondered why the same relationship patterns keep turning up: Understand the Relationship Patterns That Keep Repeating. It offers a free guide, around 16,500 words, on using Socionics to make sense of communication, attraction and conflict — why you feel instantly understood by some people and consistently misread by others, and why the same argument follows you from one relationship to the next. It is written for people coming from MBTI and for anyone who has never met Socionics before; no prior knowledge is assumed and every term is explained where it appears. What it deliberately does not do is tell you who you are compatible with, or predict whether a relationship will work: it is a way of reading a pattern, not a verdict on anybody. Enter an email address and that is the whole form. The guide itself is still being finished, so the page tells you it will arrive by email when it is ready rather than pretending something has already been sent, and you can leave the list with the link at the bottom of any email. The privacy policy now has a section on what happens to an address you give the site — which it should have had already, since the signup forms in articles and on downloads have been there for months. A link to the new page appears on the home page, at the foot of the MBTI pages, the comparison pages, the beginners' guide, the compatibility, relations and functions pages, and on the test result screen once you have a type — and nowhere else; the bar that usually offers the relations reference stays out of the way there rather than covering the form.
6 September 2026 — The sixteen "on screen" pieces — one fictional character for each Socionics type — are now a single page: Socionics Types on Screen. Each character keeps its own section and its own address, so an old link to, say, the ILE piece still takes you straight to the ILE section rather than to the top of the page. The writing is unchanged apart from the repeated sign-off that used to close every one of them, which only made sense when they were sixteen separate pages. There is a new introduction, and a set of jump links grouped by quadra so you can go directly to the type you want. The sixteen videos no longer load until you click one, which makes the page considerably quicker to open than sixteen embedded players would be. Every type profile now links to its own character, which none of them did before.
6 September 2026 — The eight articles on the cognitive functions have been folded into the eight function pages themselves. Each function had two pages covering the same ground: a short reference page at, for example, Extraverted Intuition (Ne), and a longer article on the same function elsewhere. The longer piece now sits on the function page, below the definitions and the everyday examples and above the Model A position grid, so everything on a function is in one place instead of split across two addresses. Nothing has been thrown away: where the two overlapped, the better-written passage was kept and the repetition removed, and the old article addresses now take you to the matching function page. The reading order on each function page is otherwise unchanged.
4 September 2026 — Ten more side-by-side comparisons are now linked from the compare directory, where they previously appeared as plain text you could not click. They are EII vs ESI, ESI vs SEI, LII vs LSI, EII vs LII, EII vs SEI, ESI vs LSI, ILI vs SLI, IEE vs ILE, SEI vs SLI and ESI vs IEI, and each has the same full write-up as the others: both function stacks, what each type makes of the other, and the relation between them. The pages themselves are unchanged and were always there if you knew the address; they are now reachable from the directory and from the related-comparison links at the foot of every comparison page.
3 September 2026 — The email signup has been reworked. The bar along the bottom of the page used to slide up after twenty seconds on every page, whatever you were doing; it now waits until you have been reading for at least half a minute and have got a good way down the page, or until you look like you are leaving. It no longer appears at all on the test, the typing report pages, downloads, or the privacy, terms and contact pages, and it stays away if you arrived from one of our emails. Closing it now keeps it away for three months rather than for that visit alone, and if you subscribe it will not come back at all, which it previously could. The close button was too small to hit reliably on a phone and is now a full-sized target, the bar can no longer show you an offer and a confirmation at the same time, and it will not slide if you have asked your device to reduce motion. The signup box that sat near the bottom of the sidebar has been removed: on articles it was a second copy of the same form that already appears at the end of the piece, and on a phone it sat about fifteen screens down, where nobody was ever going to find it. The form at the end of each article, and the one on the downloads page, are unchanged and are now the only signup on those pages.
1 September 2026 — There is a new tool for when you have narrowed your type down to two and cannot get past it. Pick the two you keep going back and forth between and it works through a short set of questions written for that specific pair, from the function stacks that actually separate them rather than from the four dichotomies the test uses. Every question has a "neither really fits" option, and choosing it counts towards the result rather than being ignored. The tool can also come back and tell you the questions did not separate the two, which is a real answer and the one it will give rather than round a one-question lead up into a verdict; where that happens, both types stay on your record as an open question and your stored type is left alone. Question sets are written one pair at a time and three of the 120 pairs have one so far, so for most pairs the tool will say so plainly instead of inventing questions. Nothing is saved unless you press the button that saves it, and as everywhere else it stays in your browser. You will find the way in on the test result screen, on the personalised line on any type profile once you have set a type, beside anyone on your people list, on the typing report page, and in the article on why you are probably not the type you think you are.
1 September 2026 — You can now keep a private list of the people you know. Set your own type, then add anyone else with theirs, and the new people list tells you what each of them is to you: who your Dual is, who you supervise, who supervises you, and who you are likely to find hardest going. The asymmetric relations say which end you are on in plain words rather than leaving you to work it out from the name, and anyone whose type you are not sure of is shown as a maybe rather than a fact. The list stays in your browser: the names you type never leave your device and are never sent anywhere. There is a control to remove the whole list, and removing one person offers you an undo rather than asking you to confirm. You will find the way in at the end of the personalised line on any type profile, once you have set a type.
1 September 2026 — The home page now recognises your type. If you have set one, the quadra grid at the top of the page marks two of the 16: your own type, and your Dual. If you have not set one, there is a short line under that grid to do it, and the rest of the site adapts from there. As everywhere else, your type is stored in your browser only and is never sent anywhere. Nothing on the page has been removed or moved to make room, so if you have not set a type the home page reads as it did before, apart from that one line. The grid is a desktop feature and always has been, so the markers and the line appear on wider screens only.
1 September 2026 — You can now tell the site your Socionics type, and it will tell you what each page means for you specifically. Set it once, and the 16 type profiles, the 8 function pages and the relation pages each gain a short line near the top: on a type profile, the relation between your type and that one, including which side of it you are on for the asymmetric ones; on a function page, where that function sits in your own Model A and what sitting there involves; on a relation page, which type that relation actually names for you. Finishing the test sets it for you. Your type is stored in your browser only and is never sent anywhere, and there is a control to clear it in the strip itself. Nothing on any page has been removed or moved to make room: if you have not set a type, the pages read exactly as they did before.
1 September 2026 — The "If this result feels wrong" section on the test result screen is now complete. It lists three types for all 16 results, where it previously appeared for only four of them. Each entry names a type yours is commonly confused with, one line on what actually separates the two, and a link to the side-by-side comparison.
1 September 2026 — Your Dual now has its own line on the test result screen, directly under your type, instead of sitting in the row of small links further down. It names the type, what it is called, and links through to what Duality actually involves. Two things are new below the small groups: a short "If this result feels wrong" section listing the types yours is most often confused with, each with a line on what separates them and a link to the side-by-side comparison, and a card for the Socionics Made Simple volume for your type. The mistyping section is being filled in type by type, so it will not appear for every result yet. The book card was meant to appear on this screen and never has, owing to a fault in the page's styling; that is fixed.
1 September 2026 — The test result screen has been tidied up. It previously ended in a row of nine buttons of roughly equal weight, which made it hard to see what to do next. There are now two: read your full type profile, and retake the quiz. The links to your Identity and Dual comparison pages, to Socionics World AI, and to share your result on X are all still there, as a small line of links below the buttons. The "Find your dual on Socion" button has moved down the page, below the explanatory section about the test, where it reads as a next step rather than one option among many. Nothing has been removed except a duplicate: a second link to the typing service that pointed to the same place as the button in the panel above it. Also fixed a display fault on the same screen, where the words "Extraversion" and "Introversion" were being drawn over the score bar on phones instead of fitting beside it.
1 September 2026 — The half-price offer on the Standard typing report now shows across the site, rather than only on the typing page and the test result screen. Everywhere the service was previously advertised at "from £40" — the home page, type profiles, comparisons, relations, the MBTI pages, articles — you will now see £20 alongside the crossed-out £40, and the link takes you straight to the offer. The typing page's own title and search listing were quoting £40 as well, so anyone finding it through Google was told the full price before they arrived; those now carry the offer price too. Nothing about the offer itself has changed: still the Standard individual report only, still the first 25 people, still applied for you at checkout. Also fixed a display fault in the same panel, where the button's label was being cut off mid-word on phones.
1 September 2026 — The typing report page now states plainly that reports are produced with AI assistance, working from the Socionics framework developed here over 16 years, and reviewed before they are sent. Two earlier descriptions were wrong or misleading on this point and have been corrected — one on the test result screen described reports as written by hand, which they are not. What the service offers is unchanged: a personalised written analysis of your own questionnaire answers, with the reasoning shown.
August 2026
31 August 2026 — The typing report questionnaire now arrives the moment you pay, instead of being emailed over within 24 hours. It runs to 20 questions across six short sections, saves your progress so you can finish it in more than one sitting, and asks for a little more detail than before — around 1,200 words in total rather than 500, because short answers are the main reason a report has to hedge. For Couples reports, whoever books gets theirs immediately and a link for the second person follows by email.
30 August 2026 — The typing report Standard tier is half price — £20 instead of £40 — for the first 25 people to claim it. The discount is applied for you at checkout, so there is no code to type in; it covers the Standard individual report only, and once 25 have been claimed the price returns to £40. The typing prompt on the test result screen has been rewritten at the same time: it now says what is actually in the report — its length, the function stack analysis, the relations section, and that the reasoning is shown in full including where your answers were ambiguous — along with the refund terms, all of which previously took a click to find. Also fixed a long-standing display fault on that result screen, where every button was drawing its label in link blue rather than its own colour, leaving some of them hard to read.
14 August 2026 — Removed the Google AdSense script from every page. No ads were ever served through it, so it was loading on each page visit for nothing — pages should now load a little faster. The privacy policy has been corrected to match: it previously described advertising cookies and a consent banner that were never actually in use. It now states plainly that no third-party advertising is shown and no advertising cookies are set, and commits to updating the policy and asking for consent before that ever changes.
14 August 2026 — A small layout adjustment: the sidebar is now a little wider, and the page as a whole has a slightly wider maximum width to make room for it. The article column is exactly the width it was before, so reading is unaffected. On tablet-sized screens the sidebar now moves below the article a bit sooner than it used to, which gives the main text more room at those widths.
10 August 2026 — Four new guides, driven by what readers actually search for. Three answer the question in the direction the MBTI bridge didn't yet cover — you know your Socionics type and want the MBTI translation: IEI in MBTI, SEI in MBTI, and EII in MBTI, each explaining why the last letter of your code flips between the two systems and which MBTI result your type actually tests as. The fourth is a full relationship deep dive on the Beta dual pair: SLE and IEI — Duality in Practice, covering the Se/Ni exchange, the Aggressor–Victim romance dynamic, and where the pairing strains.
July 2026
30 July 2026 — The comparison and famous people sections have been consolidated so the site points more clearly at the material people actually read. The sixteen pages that compared a type with itself have been retired: pairing a type with itself is the Identity relation, which is covered properly on its own page, and those sixteen URLs now redirect there. Across both sections, pages that have drawn no readers from search over the past three months are no longer listed in search results. They remain on the site and every link to them still works, so nothing you have bookmarked will break. The comparison index now covers 120 pairings across 13 intertype relations.
22 July 2026 — The site has a proper email address at last: you can now reach me at a new address, revealed on the contact page. The contact form works as before.
22 July 2026 — The floating Feedback tab that appeared in the corner of every page has been retired — it saw very little use. If you'd like to get in touch, share a correction, or leave a note for the guestbook, the contact page remains the best way to reach me.
22 July 2026 — Two of the site's most-visited pages have been substantially expanded. The Socionics test now explains what each dichotomy measures, how the scoring works, how to read a firm versus borderline result, and what a short test can and cannot tell you. The Famous People directory gains an analytical introduction to each quadra's public figures and a new section explaining how public figures are typed from the behavioural record. The FAQ answers that previously existed only behind the scenes on the test and the 16 type profiles are now visible on the pages themselves.
16 July 2026 — Every type profile now opens with a substantial original "The SLIDE System™ View" analysis — covering each type's cognitive signature, strengths, blind spots, behaviour under stress, and relational needs — written from the ground up for this site using the SLIDE System framing. The classical description drawn from Weisband's Working Materials is unchanged but now sits below the new analysis as "A Classical Perspective." The result is a fuller, more original picture of each of the 16 types.
7 July 2026 — The Socionics test is now linked from the desktop header navigation — previously it was only reachable from the mobile menu and footer, despite being the most-visited page on the site. Also fixed a pair of broken links to the SEE vs SEI comparison and restored the SVG favicon.
2 July 2026 — Repo-only files (README, CLAUDE.md, CHANGELOG.md) no longer build into live pages — they were accidentally being published as unstyled URLs. Netlify now also skips the build entirely when a commit only touches those files, saving build minutes on documentation-only changes.
2 July 2026 — The Socionics Insight Shop (type mugs, prints, stickers) has closed due to lack of sales. Removed all shop links and mug widgets site-wide: header and footer navigation, the type-page sidebar, the hub, the roadmap, the about page, and the type quiz result screen.
May 2026
15 May 2026 — Corrected several factual and consistency issues across the relation and function articles following two catches from a reader.
On the relation pages: the Conflict page was describing Infantile as "Alpha Rationalists and Delta Irrationals" and Aggressor as "Beta Irrationals and Gamma Rationals" (the correct categorisation is by S/N within quadra: Infantile is Alpha and Delta intuitives, Aggressor is Beta and Gamma sensors). The same paragraph also implied that conflict's romance style pairings produce the same dynamic as duality's, when in fact they are inverted — duality pairs Infantile with Careful and Aggressor with Victim, conflict and super-ego pair Infantile with Aggressor and Careful with Victim. The Semi-dual page claimed every semi-dual pair shares the same romance style combination as duals; in practice, only the four irrational pairs do — the four rational pairs (ESE–LSI, EIE–LII, LIE–EII, LSE–ESI) pair Careful with Aggressor or Victim with Infantile instead.
On the function articles: the "What Si does" section of the Introverted Sensing article only mentioned two of the four Ego-block types and was missing both SLI (position 1, leading) and LSE (position 2, creative); the "What Se does" section of the Extraverted Sensing article was missing LSI (position 2, creative). Both now name all four Ego-block types. The Extraverted Intuition article did not name any types in its "What Ne actually does" section, breaking the pattern set by the other seven function articles; an Ego-block paragraph covering ILE, IEE, LII and EII has been added.
14 May 2026 — Added the expert typing service to the highest-intent surfaces it was missing from: the type quiz result screen — with a stronger prompt when a result comes out tied or borderline — and the 16 type profile sidebars. The typing prompt on type, function and relation pages now sits above the previous/next navigation rather than below it. Quiz analytics flag borderline results so quiz-to-typing conversion can be tracked against result confidence. Typing call-to-action copy reworked site-wide to lead with price and turnaround.
11 May 2026 — Added IndexNow submission to the build. Every production deploy now pings Bing, Yandex, Seznam.cz and Naver with the full URL list so updates are picked up within hours rather than weeks. Google does not participate in the protocol.
11 May 2026 — Added topical cross-linking to type comparison pages and famous people profiles. Each comparison page now links out to all other compares involving either type and to other pairings sharing the same intertype relation. Each famous person profile now links to other notable figures of the same type. Stronger internal linking to help readers explore related material and improve organic discoverability.
April 2026
29 April 2026 — Corrected the direction of asymmetric relations in several places where the active and passive roles were the wrong way round: six type comparison pages had Supervision inverted (EIE/SEI, EII/ILE, EII/SEE, ESI/IEE, LIE/SLI, LSE/SEI); the Beta vs Gamma article on the Apple–Microsoft rivalry had Supervision inverted in its core thesis and has been rewritten with the correct direction (Beta supervises Gamma, not the other way round); and the compatibility checker was inverting the Benefaction direction in its result text.
26 April 2026 — Added a new Hub page — a single landing page linking out to every Socionics project I run, including Socion, Socionics World, the books, the Discord, the YouTube channel and the shop.
March 2026
23 March 2026 — The original Definitive Socionics Index (definitive-socionics.wikidot.com) was retired after nearly two decades. All material has been carried forward, revised, and expanded here. This site is now the sole home of that research.
21 March 2026 — YouTube channel created at @SocionicsInsight. Videos coming soon.
18 March 2026 — Type quiz updated to v2 — several bugs fixed, results now shareable via direct link. See the quiz changelog for detail.
16 March 2026 — Full site search added. Every page on the site is now searchable via the header search button.
15 March 2026 — Published individual pages for 15 famous people, each with a typing rationale and supporting video.
14 March 2026 — Published all 136 type comparison pages — every possible pairing of the 16 types, with perspectives from both sides and a compatibility verdict.
9 March 2026 — Site launched at socionicsinsight.com with 284 pages covering the 16 types, 8 functions, 16 intertype relations, small groups, Model A, a type quiz, and an initial set of articles.
Changes to the type quiz are also tracked in the quiz changelog.