MB3I

MB3I
Type	Experimental text-pattern classification framework
Primary use	Broad MBTI-style family signal estimation from written text
Output families	NTJ, NTP, NFJ, NFP, STJ, SFJ, STP, SFP
Classification level	Family-level, not full four-letter type
Associated application	Reputation Flair
Related method	Parametric Text Tessellation
Primary data form	Written posts, comments, or text batches
Output form	Family scores, family wins, score rates, entropy, aggregate distributions

MB3I is an experimental text-pattern classification framework used to estimate broad MBTI-style family signals from written content. Rather than assigning a full four-letter personality type, MB3I groups outputs into eight type families: NTJ, NTP, NFJ, NFP, STJ, SFJ, STP, and SFP.

In application contexts such as Reputation Flair, MB3I is used as a supplemental classifier for aggregate analysis, contributor statistics, subreddit transparency views, and optional flair display. It is not designed as a psychological diagnosis, clinical personality test, or authoritative identity label. It is better understood as a statistical writing-pattern classifier that compares structural and behavioral features in text against family-level model patterns.

Overview

MB3I classifies text into one of eight broad families derived from the MBTI type system. These families preserve the final three-letter grouping while avoiding a claim about the full sixteen-type structure. For example, instead of attempting to determine whether a user is specifically INTJ or ENTJ, MB3I may classify the text under the broader NTJ family.

The purpose of this simplification is to reduce false precision. Full MBTI typing from short text is unstable because a short sample may reflect topic, role, mood, platform pressure, imitation, sarcasm, or the writer's temporary communicative mode. Family-level grouping is intended as a broader signal that can be interpreted more cautiously.

MB3I therefore emphasizes pattern distribution rather than fixed personal identity. It is strongest when used across many text samples or as an aggregate comparison layer rather than as a single-comment label.

Rationale for family-level classification

The framework avoids the introversion/extraversion dimension because written online behavior often weakens that signal. An introverted person may appear more extraverted online, especially in argument, analysis, roleplay, moderation, or community participation. Likewise, an extraverted person may produce sparse or reserved text depending on context. For this reason, MB3I collapses the sixteen-type structure into eight no-I/E families.

The result is a classifier focused on broad structural family resemblance rather than a complete personality-type assignment. The family label should be read as a model output about the text sample, not a fixed claim about the person who wrote it.

Type families

MB3I uses the following eight families:

Family	General grouping	Examples of full MBTI types included
NTJ	Intuitive–Thinking–Judging	INTJ, ENTJ
NTP	Intuitive–Thinking–Perceiving	INTP, ENTP
NFJ	Intuitive–Feeling–Judging	INFJ, ENFJ
NFP	Intuitive–Feeling–Perceiving	INFP, ENFP
STJ	Sensing–Thinking–Judging	ISTJ, ESTJ
SFJ	Sensing–Feeling–Judging	ISFJ, ESFJ
STP	Sensing–Thinking–Perceiving	ISTP, ESTP
SFP	Sensing–Feeling–Perceiving	ISFP, ESFP

Method

MB3I evaluates written text and produces family-level scores. These scores are relative signal strengths rather than absolute truths. A text sample may produce a strongest family, a ranked list of families, or a probability-like distribution across all eight families.

In moderation, analytics, or research settings, MB3I can be applied to individual comments, posts, batches of text, or running user aggregates. The system can then aggregate results over time, producing statistics such as family wins, total score distribution, entropy, score rate, family win rate, and daily family distributions.

A simplified aggregate result may appear as:

Family	Wins	Win rate
NTJ	787	30.7%
NTP	543	21.2%
STP	355	13.9%

In this form, MB3I is not claiming that the users themselves are those types. It is reporting which family patterns most often won under the classifier for the processed text.

Core model detections

Some MB3I deployments use multiple internal models. In those cases, one processed item may generate more than one family detection because multiple core models evaluate the same text. A report may therefore distinguish between item count and core model detections.

Term	Meaning
Item count	Number of posts, comments, or text samples processed.
Core model detections	Number of times the internal MB3I models selected a family as their top result.
Score rate	Average family score or probability-like share across processed items.
Family wins	Count of top-family selections under the configured scoring method.
Entropy	Measure of how concentrated or dispersed the family distribution is.

This distinction matters because a family table may show more detections than processed comments when several core models contribute to the same aggregate record.

Use in Reputation Flair

In Reputation Flair, MB3I is an optional experimental feature. When enabled, it can add family-distribution analysis to the Reputation Portal, contributor lookup views, subreddit transparency statistics, analysis charts, mod notes, and optional user flair.

MB3I does not replace the main scoring system. Reputation Flair's moderation logic can still rely on reputation balance, contribution counts, trigger totals, PTT scoring, bot-like behavior signals, verification status, and configured review or removal thresholds. MB3I functions as an additional interpretive layer for pattern analysis.

Common uses include:

Use	Description
Contributor lookup	Shows a user's strongest MB3I family signals when enabled.
Subreddit transparency	Adds aggregate family-distribution tables or charts.
Flair display	Optionally displays the top MB3I family or top two families in user flair.
Trend analysis	Tracks family-level distributions over time.
Mod notes	May include top MB3I families when enabled and available.
Personal statistics	May expose a dedicated MB3I stats page when the feature is active.

Display behavior

When MB3I flair display is enabled, a flair may show a single family marker such as:

NTJ

If top-two family display is enabled, it may show a combined marker such as:

NTJ/NTP

MB3I flair output does not require an emoji. In Reputation Flair, MB3I display is controlled by moderator configuration. If preserve-flair mode is enabled, MB3I flair display should be disabled so the application does not overwrite the preserved user flair slot.

Enabled and disabled states

When MB3I is disabled, MB3I-related views and outputs should be hidden from user-facing and moderator-facing surfaces. This includes MB3I personal stats, subreddit aggregate MB3I tables, contributor lookup MB3I comparisons, analysis charts, flair output, and MB3I entries in mod notes.

When MB3I is enabled, it may appear in:

Surface	Possible MB3I output
Contributor Stats Lookup	Family rows, score rates, and comparison tables.
My MB3I Stats	User-level family distribution and top-family information.
Subreddit Transparency	Aggregate and daily MB3I family statistics.
Analysis charts	Family trends by day, month, or all-time range.
Mod notes	Top family summaries when enabled.
Flair	Top family or top-two family marker when configured.

Interpretation

MB3I results should be interpreted comparatively. A high NTJ score, for example, does not prove that a person is an INTJ or ENTJ. It indicates that the processed writing sample matched the classifier's NTJ-family pattern more strongly than the other seven families.

The most accurate interpretation is:

Within this model, this text most closely matched the NTJ family.

The least accurate interpretation is:

This person is definitely an NTJ.

MB3I is therefore more suitable for aggregate pattern research than for individual labeling.

Privacy considerations

When used in Reputation Flair, MB3I is designed to work from scoring outputs and aggregate statistics rather than storing raw post or comment bodies in long-term application storage. The practical goal is to support analysis without creating a permanent archive of user text.

A privacy-conscious deployment should state whether MB3I is enabled, what it affects, whether it appears in flair, whether it appears in mod notes, whether it contributes to moderation routing, and what form of aggregate output is retained.

Relationship to PTT

MB3I is related to Parametric Text Tessellation in that both are text-pattern systems, but they serve different purposes. PTT is used as a structural text-classification layer for scoring and routing in Reputation Flair. MB3I is a family-distribution classifier that estimates broad typological writing-pattern signals.

PTT can affect good, bad, bot, or trigger scoring when enabled. MB3I is better suited for aggregate analysis, comparison, and optional display. A deployment may use one without the other.

Scope and constraints

MB3I depends on training data, feature design, scoring assumptions, and sample quality. Text length, subject matter, platform norms, moderation pressure, quotation, imitation, sarcasm, roleplay, emotional state, and audience can all change the output. A single short comment may contain too little signal for meaningful interpretation.

Because of this, MB3I should not be used as a hidden identity profiler, clinical instrument, employment filter, disciplinary basis, or proof of personality. Its practical use is comparative, aggregate, and experimental.

References