Perspective Before Data

ABSTRACT

WP 1 showed that an organization must be able to review its own manner of measuring. This paper shows why: upstream of every data question lies a question of perspective, and in organizations it is almost never worked on.

Four AGI-26 contributions hit the same point. Hongju Pae introduces perspective as a structural condition that scaling does not replace, drawing on Husserl's distinction between intentional matter and intentional quality. Michael Timothy Bennett shows that "short" and "simple" depend entirely on the encoding. Antona, Ibias and Otero show in a concrete case how an encoding destroys the information about closeness. And Ibias and colleagues show the rare property that follows: a system that can output "I do not know" rather than inventing a label. From this follows Condition 2 and the Perspective Audit.

Keywords: Modes of Disclosure · perspective · encoding · world model · Perspective Audit · LEAD21 · Circle of Clarity

Suggested citation: Huesch, G. (2026). Perspective Before Data. Working Paper Series 2026, Series 1 "The Superintelligent Organization", WP 2. Global Leadership Consultancy.

PART 1 · THE FINDING

1 THE QUESTION

Twelve per cent market share.

One reads: we hold twelve per cent. The other reads: we are missing eighty-eight.

Same number, same precision, incompatible consequences — and neither of them is wrong. The usual shortcut: a matter of attitude, optimism versus pessimism. The more precise finding has nothing to do with temperament.

The question of this paper: how does the world model of an organization become explicit — and what happens for as long as it does not?

2 WHAT AGI-26 SHOWS

Perspective is a structural condition, not a quantity. On his poster Hongju Pae makes a claim that runs against the mainstream of the active-inference debate: scaling is not enough. A structural condition is missing — Perspective, which he defines precisely as "how a world is given, not what is inferred".

His warning hits every organization that believes more data solves a problem of perception:

"Showing that a system inferentially stabilizes itself does not show that there is something it is like to be that system. The mistake: treating scaling, hierarchy, or recursive self-modeling as if it closes a gap whose shape has not yet been drawn."

Pae draws on Husserl for this, and the example stands verbatim on the poster:

"The half-filled glass being half-full vs. half-empty: Same water volume. One encounters it as sufficient, the other as lacking. Not different beliefs about volume. Different modes of disclosure."

That is the distinction at stake. Intentional matter is what an act is directed toward — the volume of water, the market share, the figure. Intentional quality is the mode under which the content is given — as sufficient or as lacking. The organization argues about the matter and means the quality.

Four features make up perspective in Pae's sense: it is global (permeating perception, cognition and action at once), pre-reflective (shaping experience without becoming an object), consequential (weighting what counts as salient) and persistent (resisting transient evidence, modifiable over time). And decisively: it is coupled asymmetrically — it shapes inference but is not updated by it.

What counts as "simple" depends entirely on the encoding. In The Wrong Razor, Michael Timothy Bennett shows why Ockham's razor becomes unusable as a decision rule: "What counts as short depends on the encoding." On a finite class, any strictly positive prior can be made Ockham-like by a suitable code. And without a privileged abstraction layer, intrinsic semantic length collapses to uniformity.

For organizations: every firm has a codebook. It decides which explanation counts as simple, which situation as urgent, which option as unrealistic. It is documented nowhere. And it is more powerful than the strategy, because it determines which strategies appear thinkable at all.

An encoding can destroy the information about closeness. Hector Antona, Alfredo Ibias and Beatriz Otero show this in a case that translates one to one. In hyperdimensional encoding, numeric features are split into levels, each assigned a randomly generated vector. But random vectors are all equally dissimilar:

"two adjacent values look exactly as different as two values at opposite ends of the range. The core sub-symbolic property of numbers — their closeness — is destroyed."

The consequence: "the model can no longer sense magnitude — a near-miss looks no closer than a wild one." Their correction yields roughly six per cent more accuracy across five datasets, most strongly where magnitude matters most.

Translated: an organization that encodes deviations only in traffic-light colours has destroyed the information about closeness. A near miss then looks exactly like a wide miss — and the system can no longer sense how large something is.

The rare property is abstention. With Unsupervised Cognition, Alfredo Ibias and colleagues build a method that names a problem in current approaches:

"Every input is forced into some region, independently of whether it is actually similar to what lives there."

Their answer: inputs are classified only if they are similar enough to a learned archetype — otherwise the model abstains.

"I do not know." — "If no learned representation is similar enough, the Nodule returns no Projection rather than hallucinating a label — a rare property for an unsupervised learner."

That is the sentence every reporting system ought to be capable of and almost none is.

3 WHAT ORGANIZATIONAL RESEARCH HAS SAID FOR DECADES

Weick (1995) has the point in a formula: sense is made, not found. Organizations do not stand before a situation which they then interpret — they produce the situation by interpreting. What follows is uncomfortable: interpretation precedes information, not the other way round.

4 WHAT HAPPENS WHEN THE MODE IS NOT ON THE AGENDA

In every leadership meeting where the same figure lies on the table, sales sees an opportunity, controlling a risk, operations an imposition. The figure gets discussed. The mode is what is being fought over. And because the mode is not on the agenda, the meeting ends in compromises about measures while the modes of disclosure carry on unchanged.

Until the next meeting. A different figure lies there, and the same argument begins again — without anyone noticing that it is the same one.

And now we place a model on that material. It does not learn the organization's truth. It learns its manner of disclosure — and returns it with higher confidence. That is not a fault of the model. It is a precise description of what a model does.

5 THE PERSPECTIVE AUDIT

Five fields, in writing, before measure planning. No longer than one page.

Field 2 is the decisive and the most uncomfortable. It makes visible that the dispute was never about the figure.

6 ONE EXAMPLE, WORKED THROUGH

Illustrative example, not a case report.

A consumer goods manufacturer is losing share in one segment: from 14 to 12 per cent in two years. The figure is undisputed.

Sales discloses it as a gap: eighty-eight per cent lies outside, the segment is growing, we are too slow in retail. The conclusion: expand the field force.

Controlling discloses it as a holding under pressure: twelve per cent carries the fixed costs, the loss of share comes from a price war we cannot win. The conclusion: secure margin, give up volume.

Both are right — for their manner of disclosure. The meeting ends in a compromise: expand the field force slightly, hold prices slightly. Both too little to have an effect.

What the Perspective Audit would have changed: field 2 would have put the disagreement on the table before the discussion of measures. The question would then have been the one that actually has to be decided — are we defending a holding or attacking a gap? — and not the question of how many field staff.

7 HOLD ON

Objection 1: This is philosophy, not leadership.

The objection hits the origin, not the effect. Husserl formulated the distinction; what matters is that it makes visible a disagreement otherwise mistaken for a dispute about facts — and therefore not solvable.

The advantage: a disagreement about perspective is solvable, a dispute about a correct figure is not. Naming the mode moves the argument to the place where it can be decided.

Objection 2: If every perspective holds, none is wrong. That paralyses.

No. Perspectives are neither arbitrary nor equally good — they are differently fitting to the situation. The point is not to let them all stand, but to say them out loud so a choice can be made. An unchosen perspective operates anyway; it only operates unnoticed.

The advantage: after the audit, leadership decides deliberately which mode holds for the next twelve months. That is a decision one can take, justify and later correct.

8 THE CONDITION

Condition 2 · Perspective An organization's encoding decides what it can perceive at all as near or far, urgent or immaterial, simple or unrealistic. For as long as the mode is not on the agenda, the organization negotiates measures and leaves its perception unchanged.

PART 2 · WHAT LEAD21 ANSWERS

CRITICAL APPRAISAL

The Circle of Clarity achieves more than most goal-setting exercises, because it has the target state described rather than named. In the description, the manner of disclosure becomes visible involuntarily. But that is a side effect, not a step.

Two fields have no instrument in LEAD21. Resolution — whether the reporting system can see the difference between near and wide — does not appear. And abstention — an admissible "we don't know" with a defined consequence — does not appear either. Both are gaps that the finding from San Francisco made visible in the first place.

The mapping comes from the developer of the system and is not independently verified.

WHAT THE SERIES GIVES BACK TO LEAD21

The Perspective Audit as a step of its own ahead of the Circle of Clarity, with field 2 mandatory: everyone involved notes in writing, before discussion, whether they disclose the situation as a holding or as a gap.

A resolution check for the reporting system: for how many metrics does the presentation distinguish between a near miss and a wide one? Three-step traffic-light systems destroy this information systematically.

Reasoned abstention is worked out in WP 5 as a decision instrument; here it belongs in the reporting system as an admissible answer.

FALSIFIABLE PREDICTION

Prediction 2. Leadership groups that run a Perspective Audit before measure planning arrive at fewer compromise solutions and more unambiguous directional decisions on contested situations than comparable groups without an audit — measured by the share of resolutions containing two opposing measures at once.

PART 3 · ABOUT GERALD HUESCH

That the same situation carries two readings is not a theory from the seminar room — it is the experience of rooms in which both readings sat at the table at once. In his interview with Unternehmer Deutschland, Gerald Huesch speaks about leadership culture and emotional intelligence at exactly this point: what a leader perceives as the situation decides, ahead of any measure, which measures are eligible at all.

His piece Was unterscheidet den Leader vom Manager? in the ZEBRA interview treats the same point. The distinction lies there neither in the role nor in the authority, but in whether someone can make their own manner of disclosure an object — or takes it for the situation.

Systems shape behaviour. Through the selection of the right instruments, mechanics and leadership codes — derived from values and leadership guidelines — the behaviour of everyone involved is synchronized and aligned. Not through motivation, but through a system that produces the right behaviour when applied. Sales has Salesforce. Finance has SAP. Leadership has LEAD21.

About the author. Gerald Huesch is the creator of the leadership operating system category – "one of the most professional and creative minds in the domain of leadership and difficult change projects" (Boston Business School). With LEAD21 he built the first operating system for leadership – 21 evidence-based methods, the comprehensive solution and the missing link for complete transformation, extreme growth and high-volume post-merger integration: a unique leadership system for human coordination at scale, in which alignment and culture emerge as a by-product. He has trained more than 30,000 executives in over 30 countries (NASA, Rolls-Royce, Airbus, Omnicom and others) and delivers executive leadership training, leadership development and executive mentoring for executives in Germany, the Americas, Asia Pacific and the UAE. Participants from 300+ companies report +87% clarity, +58% efficiency, +50% performance, +46% alignment and −50% negative stress after LEAD21 (participant-reported); a one-year study with 567 leaders (Prof. Beate Hanson & Prof. Mohit Bhandari, McMaster University) showed improvement in 8 of 9 climate dimensions. Dr. Charles Pellerin (NASA) on LEAD21: "Unique in the world, and a capability that nobody else has." gerald-huesch.com/about-en

PART 4 · OUTLOOK

STATUS OF THE SERIES

WHAT COMES NEXT

WP 3 · No Side Door to Belief. Governed Uptake — the governed path by which a claim comes to hold durably, and the Epistemic Bypass, the side door around it. This paper showed how a situation is disclosed. The next shows how a claim about that situation attains durable status — and why in most organizations this happens ungoverned.

OPEN QUESTIONS FROM THIS PAPER

1. Can an organization's codebook be surveyed without the survey changing it?

2. Are there industries in which a single mode is durably right — or is the shift always only a matter of time?

3. What minimum resolution does a reporting system need for magnitude to be preserved?

NOTES

On method. Conception, argument and responsibility rest with the author. An AI system (Claude, Anthropic) was used as a working instrument for research, structuring, the transposition of the AGI-26 contributions and the bilingual drafting, with the explicit instruction to assess the relation of LEAD21 to AGI critically and with the greatest possible objectivity, including against the author's interest. Readers are expressly invited to evaluate the contribution of LEAD21 to AGI critically, with maximum objectivity of their own, and to draw their own conclusions.

Declaration of interest. The author is the developer and owner of LEAD21 and has a commercial interest in its dissemination. Every statement about LEAD21 is assigned a degree of coverage and an evidence status.

On sources. The author evaluated the posters he documented at AGI-26 as well as titles, abstracts and the proceedings tables of contents — not the full texts. Verbatim quotations come from the posters. The transfer to organizations is an analogy by the author.

REFERENCES

AGI-26

Antona, H., Ibias, A., & Otero, B. (2026). Generating semantic vectors for numeric values in hyperdimensional computing. In M. Iklé, A. Franz, A. Kemendo & H. Lowy (Eds.), Artificial General Intelligence. AGI 2026, Part I (LNCS 16854, pp. 17–30). Springer.

Bennett, M. T. (2026). The wrong razor. In Iklé et al. (Eds.), AGI 2026, Part I (LNCS 16854, pp. 48– 61). Springer.

Georgeon, O. L., Marrel, P.-E., & Cook, B. (2026). Emergence of comportment-grounded semantics. In Iklé et al. (Eds.), AGI 2026, Part I (LNCS 16854, pp. 253–259). Springer.

Ibias, A., Antona, H., Ramirez-Miranda, G., Guinovart, E., & Alarcón, E. (2026). Unsupervised Cognition. Conference poster, AGI-26, San Francisco State University, 27–30 July 2026; arXiv:2409.18624. — Poster documented first-hand; proceedings entry open.

Pae, H. (2026). Perspective as a missing structural condition for subjectivity in active inference systems. In Iklé et al. (Eds.), AGI 2026, Part II (LNCS 16855, pp. 137–143). Springer.

Rodionov, S. (2026). Executable world models for ARC-AGI-3 in the era of coding agents. In Iklé et al. (Eds.), AGI 2026, Part II (LNCS 16855, pp. 198–210). Springer.

Organizational and leadership research

Weick, K. E. (1995). Sensemaking in organizations. Sage.

Philosophy

Husserl, E. (1900/1901). Logische Untersuchungen. — cited after the account in Pae (2026); primary source not consulted.