In the Room
The institute develops the instruments of Institutional Physics. A keynote is how the field first meets them.
A keynote shows an audience what is holding their institution up, what that costs, and what has to be built for a decision to stay answerable long after the people who made it have gone.
There are five talks. Each takes a single structural property, follows it down to the mechanism, and comes back with a condition the institution has to satisfy, whatever else it builds. The diagnosis is what earns the right to name that condition, and the talk ends there. An organization that hears all five can name five conditions, check for each one, and specify what is missing.
Choosing a talk
Each keynote asks one question. The question is the fastest way to tell which one fits the room you are filling.
| Talk | Slot | The question it asks | Room it lands hardest in |
|---|---|---|---|
| Dead Reckoning Flagship | 50 min | Are we still where we think we are? | Mixed rooms. Boards and executive teams. The one that needs no shared discipline. |
| The One Place | 20 min | Is the one condition everything else depends on actually present? | Governance. People who hold the authority to keep deciding. |
| The Instrument Stays Level | 30 min | What did we optimize away, and can anyone still act? | Engineers and operators. Anyone who has lived through a cost program. |
| Eight Tenths of a Millimeter | 30 min | Are we reasoning about the same object we think we are? | People who reason from models. Quantitative, research, and technical rooms. |
| Worse Than Empty | 25 min | Is our record still capable of warning us? | Anyone who owns a file. Compliance, safety, clinical, legal. |
The talks
Dead Reckoning
Are we still where we think we are?
In 1955 a competent crew flew a working aircraft into a mountain in Utah. Nobody panicked. No procedure was skipped. The distance between where they believed they were and where they actually were widened in silence, and there was no instrument aboard that could tell them. This is the talk about drift: how good institutions, run by capable people, lose contact with their own position, and what has to be built for them to notice while there is still time to correct. It works from cases an audience already half-knows, including a fire that eight cities could not fight because their hoses would not connect, and it does not end in warning. Some systems come home. The talk shows what was in place when they did.
The broadest of the five and the usual entry point. It is the only one that assumes no shared technical vocabulary, which makes it the safe choice for an audience you cannot fully predict.
The One Place
Is the one condition everything else depends on actually present?
Strengths do not add up the way we assume. A roof that is perfect everywhere except one botched seam does not leak a little; it leaks. An airplane with one control bolt torqued wrong is not mostly airworthy; it is lost. Some conditions do not average, and when one of them is absent the excellence everywhere else does not step in to cover for it. It simply stops mattering. This talk is about the condition an institution's authority actually depends on, the one that governs whether all the other strengths count at all, and about what it costs to treat a finding as feedback to be filed when it should have meant stop. On the coldest launch morning in the history of the American space program, an organization that had already put human beings on the moon had exactly one such condition in front of it, and flew anyway. The talk ends where the engineers ended: with a valve whose only job is to give up performance so the machine survives itself, and what running one costs an institution.
The talk that answers the objection the others raise. It has teeth only for people who hold the power to keep deciding, which makes it the sharpest choice for a governance audience and the wrong choice for a general one.
The Instrument Stays Level
What did we optimize away, and can anyone still act?
Aircraft wings are built with a deliberate twist that makes them less efficient on every flight they will ever make. The reason is that when the wing stops flying, the pilot still has control. Nobody funds that. It appears on every efficiency review as waste, and it is the only thing between the airplane and a spin. This talk separates four properties that are routinely confused with one another: whether a system holds its position, whether it recovers when shoved, whether anyone can still act when it doesn't, and whether anything is still telling the truth throughout. Each one is purchasable. Each one has a price. Most institutions have bought the instrument and sold the ailerons.
For organizations that have spent a decade getting more efficient and never asked what the efficiency was made of. Engineering-literate audiences respond to this one hardest, because they already know the trade is real.
Eight Tenths of a Millimeter
Are we reasoning about the same object we think we are?
A film of ice less than a millimeter thick, spread across the leading edge of a wing, can cost that wing a quarter of its lift. Nothing breaks. Every instrument in the cockpit reads correctly. The laws of physics operate perfectly, on a wing that is no longer the wing the airplane was designed around. This talk is about the failure that leaves no evidence: honest people, reasoning correctly, from accurate data, about an object that has changed underneath them and never sent word. It is the reason a clean audit can arrive three months before a catastrophe, and the reason integrity cannot fix it. What can is a design aviation already carries out every winter morning: somebody whose only job is to look at the surface, on a clock, with an answer that has to hold up when it turns out to be wrong.
The most intellectually demanding of the five and the one people argue about afterward. It is aimed squarely at anyone whose competence is a model, which is what makes it uncomfortable and why those rooms remember it.
Worse Than Empty
Is our record still capable of warning us?
A record has two jobs. It has to prove what happened, and it has to stay readable as a warning by whoever comes next. Most institutions only ever test the first, and discover the second was gone at the worst possible moment. This talk examines what happens when a record survives, and is complete, and is technically accurate, and can no longer warn anyone. An empty file admits that it doesn't know. A record that has lost its second job convinced everyone that it did. The most immediately actionable of the five: a room can hear this one and go audit something on Monday.
The most portable of the five, because every institution has files. It works from real cases with real casualties, including an assault that was reported in writing and never acted on. Nothing is told graphically. Organizers may still want to know before they book it.
Extending a keynote into a working session
Any of the five can be extended into a facilitated 90-minute session on the same day.
Participants bring one real artifact from their own environment, a dashboard, an operational record, a decision map, and apply the property from the talk to it. Realis brings the instrument. The organization brings the object. The session ends when they have found one instance of the property, or established that it cannot find one, which is also a finding. Nobody is grading the institution.
Sessions require pre-work: the artifact has to be identified in advance and physically present. A room working from memory produces a conversation about the institution. A room working from an artifact produces a reading of it. Realis runs the second kind. Full session outlines are provided to organizers during planning so that fit can be determined before anyone commits.
Who these are for
They fit organizations dealing with the specific problem the corpus studies: decisions that keep causing harm despite review, recurrence that audits do not stop, responsibility that cannot be located after the fact, and the moment those symptoms become urgent enough to look at directly.
Past that, the requirement is standing rather than job title. Every talk ends in a condition the institution has to satisfy, and satisfying one takes somebody with the standing to change a structure, so that is who needs to be in the seats. Bring the general counsel. Bring the engineer who thinks this is nonsense. The work is built to be checked, and a room that fact-checks from its seats is exactly who these talks are written for.
These talks sell nothing. Where they arrive at hope, the hope is earned the same way the warnings are, from the evidence, and the audience can check the work.
Where they come from
The talks are delivered by Christina Óvári, who developed the Realis corpus and maintains responsibility for the foundational work while the field is still forming.
She came to it from the building trades. She works in the Spenglermeister tradition, the German master craft of architectural sheet metal, where a roof laid down centuries ago can still be read by the next person who climbs up to it and judged: keep it, repair it, or take it off. It is a trade that lives in fractions of a millimeter, where the difference between a roof that lasts eight hundred years and one that fails in eleven is invisible from the street.
She is also embedded in the experimental aircraft community, among people who build these machines by hand and then fly them, where being wrong about the structure is not an abstract problem. She owns an experimental aircraft herself. The flying in these talks belongs to the builders and is quoted from them directly, in their own words. What she brings is the bench: the discipline of work that is judged by whether it lasts, and by whether someone who was not there can tell what was done and whether to trust it.
Booking
Engagements are arranged directly. For availability, format, session outlines, and fit, contact Realis Institute.
The keynotes are the field's first contact with the laboratory's work. The work itself, the corpus, the instruments, and what each one can and cannot yet do, is laid out across the rest of this site. See what is ready →