Impossible yesterday. Probable today. Possible tomorrow.
Engineering physics is the forcing function of this Institute: the discipline by which discovery becomes innovation and innovation becomes something you can build. Read that sentence as a method rather than a slogan, because it has a procedure, and the procedure is checkable.
Report a position, not a verdict.
Every hard thing was impossible until the constraint that made it impossible was named, and then it was merely difficult. So the useful output of a study is never whether something works today. It is where on the trajectory the work sits, and which constraint is currently binding.
A verdict is a claim about the world that stops the conversation: this is refuted, that is a dead end, this will never scale. A position is a claim about a trajectory that continues it: here is the number today, here is what limits it, here is the measurement that would move it. Both can be honest. Only one of them is useful to the person who reads it next, and only one of them can be wrong in a way that teaches anybody anything.
The distinction is not softness. This Institute publishes results that went against it, at the same size as the ones that did not, and retracts its own claims in public. A negative result with the constraint named is a contribution. A negative result presented as a verdict is a forecast, and forecasts about engineering have a poor record precisely because they assume the binding constraint will not move.
Only one of them is physics.
When something does not work yet, one of these is usually binding. Saying which is the whole content of a serious assessment, because they have completely different timescales and completely different responses to effort. The first five sit inside the artifact. The last three sit in the world it has to land in, and they bind more often than an engineering institute finds comfortable.
A conservation law, a thermodynamic floor, a diffraction limit. The only constraint that does not move.
How you know: It can be written as an equation with no free parameters. Landauer's kT·ln2 is one; the speed of light is one. Almost nothing is.
No substance yet has the property the design needs, at the temperature or the price it needs it.
How you know: Progress arrives in steps, from discovery, and is hard to schedule, but the target is a number, and you can say what number.
The joules per task are wrong by a factor you can state, on a budget you can measure.
How you know: It is the constraint most often mistaken for physics. A factor of a thousand in efficiency has happened repeatedly; a violated conservation law has not.
Every part is understood and the integration is not done: tolerances, interfaces, manufacturing, reliability.
How you know: Nobody disputes it can work. The question is only how many person-years and how many revisions.
The representation or the method does not exist yet, though nothing forbids it.
How you know: The most volatile of the five. It is where a single idea can move a field by orders of magnitude overnight, which is why forecasts that assume today's methods age worst.
The three that follow were added because the corpus outgrew the five. Counted from the constraint column of the three most recent survey papers, twenty-four of thirty-five resolved hypotheses bind outside the artifact: seven of eleven on the agricultural estate, six of eleven on the transportation estate, and eleven of thirteen in the sim-to-real record, where what binds is most often that no published protocol reports the quantity at all. Four topic pages here carried Engineering while their own text said the constraint was the human pipeline, or governance and consent, or a curriculum. They had the finding right and no word for it, and a vocabulary that cannot say what the work concluded will quietly rewrite the conclusion into the nearest word it has.
The device works and an instrument in force will not admit it, or admits it only at a threshold nobody has set.
How you know: Say which rule, and say what number it would have to carry. "Regulation is slow" is a complaint. "23 CFR 490.409 grades a bridge by the minimum of three integers, so a fourth measurement changes nothing until the rule reads it" is a constraint, and it tells you what to build.
The people who would operate, maintain or certify the thing are not there, or are not trained, and the pipeline that would produce them has a rate.
How you know: It is countable, which is what separates it from a mood: training throughput, instructor supply, certification latency, hours to competence. If you cannot put a rate on the pipeline you have named a worry, not a bound.
Nothing forbids the result and no instrument or protocol reports the quantity, so the question cannot be settled either way.
How you know: The tell is that both sides of the argument cite different numbers for the same word. The fix is an instrument, a denominator or a published protocol, and it is usually cheaper than the research it is blocking.
Naming one of those three is not a way out of the physics. It carries a condition: say what quantity the rule would have to read, what rate the pipeline runs at, or what the missing instrument would measure. A constraint you cannot put a number against is not a constraint, it is a reason for not having one.
The common error is to report an energy or engineering constraint in the language reserved for physics. "This cannot be done" is a strong claim, and it is almost always false: what is meant is that it costs a thousand times too much energy, or that nobody has done the integration. Those are the two constraints that have historically moved the most.
From our own corpus, including our own retractions.
Each of these is a result that would read as a dead end if the constraint were not named, and reads as a direction once it is. The verdict column is what a careless reading gives.
An earlier stage of a generated world left 0.000% trace in the finished one, in two independent receivers, at every settling budget.
The verdict reading: "Staged generation does not work."
The trajectory reading: The constraint was algorithmic and one line deep: a mark stopped being held when its stage ended, so the cell relaxed back to whatever the physics preferred. Carrying the constraint forward made the earlier stage survive at full strength, in both receivers, at every budget. The measurement that produced the null is the same one that verified the fix. the measurement →
Zero of eleven headline efficiency claims in the 2026 AI-energy literature survived audit; the real gains are 1.4–3×, not 7–96×.
The verdict reading: "Edge efficiency is hype."
The trajectory reading: The audit also found which term binds: on-device inference loses on joules and wins on interconnection, and mixture-of-experts costs energy at the edge rather than saving it. Those are engineering and architecture constraints with numbers attached, and they say where the next factor comes from. the Charlot Lab →
A covert channel of 5.0 bits per second survives our own detector, about 18 kbit an hour, ample for a key.
The verdict reading: "The certificate architecture is broken."
The trajectory reading: The floor is a property of that detector, not of the approach, and the report says so and says what would move it. Publishing the number about our own architecture is what makes the next detector measurable against something. the report →
The gap is the work.
A recurring anxiety in this field is that the important problems will be solved elsewhere, soon, and that studying them now is therefore wasted. That gets the shape of engineering backwards.
Nothing arrives solved. It arrives as a constraint that somebody moved, and the moving is done by people who understood the constraint well enough to attack the binding one instead of a comfortable one. A student who can identify which constraint is stopping a system, and design the measurement that would settle it, is doing the work that closes the gap, not preparing to do it later.
This is why the courses here are built on instruments rather than on slides. You do not learn that a sampler has a mixing time by being told; you run two chains from opposite extremes and watch where they meet, and afterwards you know what the number depends on and what would change it. The same instrument that teaches the concept is the one that measures the frontier, which is the point: there is no separate simplified version for students, because a simplified version would not be able to surprise anybody.
It is also why the negative results are published. A field where only the successes appear is a field where every newcomer re-runs the same failed experiment, and the binding constraint stays hidden behind a decade of unpublished disappointment. The fastest way to make something possible tomorrow is to say precisely, today, what makes it merely probable.
A standard, so it can be failed.
Where a piece of our own work does not yet meet this, the gap is a task rather than a secret. The corpus is audited against this standard rather than assumed to meet it, because a standard nobody checks is a preference.