Paper VII — Engineering: References
90 SOURCES · 30 QUESTIONS · ALL FREELY AVAILABLE
Each source carries a technical description and a plain one; authors’ own pages and encyclopaedia entries are linked where they exist. Every address on this page was verified live at publication. The paper itself: Paper VII.
1. Is engineering applied science, or a distinct way of knowing? Argue from cases in which the artefact preceded the theory.
The question's scholarly frame: engineering knowledge as design knowledge, distinct in aim and validation from science.
In plain terms: The free survey of whether engineering merely applies science or knows in its own right.
The Art of Approximation in Science and Engineering (6.055J, open course)
The distinct idiom exhibited: estimation, lumping, and design reasoning that runs ahead of exact theory.
In plain terms: A free MIT course in the engineer's actual mode of thought — good numbers before complete science.
Reflections on the Motive Power of Heat
The paradigm case in primary text: working steam engines preceding — and provoking — thermodynamics itself.
In plain terms: The free original written to explain machines that already ran — the artefact ahead of the theory, in one book.
Background: Engineering · History of thermodynamics
2. Why do bridges stand? Give an answer that a physicist would accept and an engineer would find sufficient.
Mechanics and Materials I (2.001, open course)
The engineer's sufficient answer: equilibrium, load paths, stress below allowables, stiffness and stability margins.
In plain terms: The free MIT course in which 'why it stands' becomes a checked calculation rather than a hope.
The first condition for equilibrium
The physicist's acceptable core: forces and torques summing to zero, stated cleanly.
In plain terms: The free physics page giving the honest minimum — nothing accelerates because everything balances.
The load-carrying mechanism itself: bending stresses, second moments, and why sections are shaped as they are.
In plain terms: A free Cambridge teaching package on how beams actually refuse to fall — geometry doing quiet work.
3. The original Tacoma Narrows Bridge is popularly said to have failed by resonance with a periodic wind. State the correct modern explanation, and say precisely why the resonance account is wrong.
Resonance, Tacoma Narrows bridge failure, and undergraduate physics textbooks
The definitive correction: self-excited aeroelastic flutter — negative aerodynamic damping — not forced resonance with periodic vortices.
In plain terms: The famous free paper that dismantled the textbook story: the wind did not beat time; the bridge amplified itself.
Forced oscillations and resonance
What resonance actually requires — external periodic forcing near a natural frequency — the precise claim the failure does not satisfy.
In plain terms: The free textbook definition of resonance, so the popular account can be tested against it and found wanting.
The primary record: the 1940 collapse documented, with the torsional oscillation the theory must explain.
In plain terms: The custodian state's free archive of the event itself — film, findings, and aftermath.
Background: Tacoma Narrows Bridge (1940) · Aeroelasticity
4. The de Havilland Comet fuselage failed in service after a number of pressurisation cycles far below the number implied by static strength. Reconstruct the reasoning error, and identify the roles of stress concentration and metal fatigue.
Fracture and Fatigue (3.35, open course)
The mechanism the static analysis missed: crack initiation and cyclic growth at stresses far below yield, taught in full.
In plain terms: A free MIT course on metal's memory — damage accumulating cycle by cycle while every single load seems safe.
Mechanics and Materials I (2.001, open course)
The static-strength frame whose sufficiency the Comet disproved — and the stress-concentration analysis that completes it.
In plain terms: The free course containing both the reasoning that was used and the correction it needed.
Stress concentration made quantitative: geometric factors at windows and rivet holes multiplying nominal stress.
In plain terms: A free specialist site on how sharp corners betray sound structure — the Comet's square windows in principle.
Background: de Havilland Comet · Fatigue (material)
5. Fracture mechanics gives the Paris law for fatigue-crack growth per cycle. State its regime of validity, and identify at least two regimes in which it fails to describe crack growth.
Fracture and Fatigue (3.35, open course)
The law in its setting: da/dN against ΔK, valid in the intermediate power-law regime between threshold and fast fracture.
In plain terms: A free MIT course placing the famous straight line on its log-log plot — and marking where it bends away.
The bounding regimes' physics: threshold behaviour below and toughness-controlled fast fracture above the power-law range.
In plain terms: The free primer on the quantity that ends the straight line — the toughness at which growth becomes rupture.
The stress-intensity machinery — K, ΔK, R-ratio effects, closure — on which the law and its exceptions run.
In plain terms: A free working site for the quantities in the law, and the complications that push cracks off its line.
Background: Paris' law · Fracture mechanics
6. Is a safety factor a quantified confession of ignorance? Distinguish the components of a safety factor that could in principle be eliminated by better knowledge from those that could not.
NASA Systems Engineering Handbook
Margins as managed ignorance: allocation, verification, and the residual uncertainty no analysis retires.
In plain terms: The free agency handbook where safety factors are administered honestly — as insurance against what is not known.
Mechanics and Materials I (2.001, open course)
The eliminable components in view: load and property scatter, model error — each shrinkable by better data and analysis.
In plain terms: The free course showing which parts of the factor are ignorance on a schedule for removal.
The Art of Approximation in Science and Engineering (6.055J, open course)
The ineliminable remainder framed: variability, unknown unknowns, and the cost of certainty itself.
In plain terms: The free course on bounding what cannot be pinned down — the confession the factor will always contain.
Background: Factor of safety · Uncertainty quantification
7. Design a machine that reports its own remaining life. What must it measure, and what must it assume about its own failure modes?
Feedback Systems: An Introduction for Scientists and Engineers (open edition)
The estimation core: observers reconstructing internal state from measurements through a model — the machine's self-knowledge formalised.
In plain terms: The free standard text on inferring what cannot be sensed directly — exactly what a self-reporting machine must do.
NIST/SEMATECH e-Handbook of Statistical Methods
The failure-model half: life distributions, degradation models, and the assumptions any remaining-life estimate inherits.
In plain terms: The free national handbook of reliability statistics — the fine print behind every 'hours remaining' display.
Identification, Estimation, and Learning (2.160, open course)
The measurement-to-model bridge: Kalman filtering and system identification for tracking hidden damage states.
In plain terms: A free MIT course in the mathematics of watching a machine age through noisy instruments.
Background: Prognostics · Structural health monitoring
8. Is failure the engineer's primary datum? Contrast the epistemic value of a structure that failed with that of a structure that has merely not yet failed.
Report of the Presidential Commission on the Space Shuttle Challenger Accident
Failure as the richest dataset: a full causal reconstruction impossible for any structure that has merely survived.
In plain terms: The free canonical inquiry — what one failure taught that a thousand successful flights had concealed.
Personal observations on the reliability of the Shuttle (Appendix F)
The epistemic asymmetry stated exactly: success cannot certify small failure probabilities; only failure calibrates.
In plain terms: Feynman's free appendix — 'nature cannot be fooled' — on why not-yet-failed proves so little.
The general logic underneath: severe tests, and why surviving weak tests confirms weakly.
In plain terms: The free philosophical account of why what fails informs more than what merely holds.
Background: Space Shuttle Challenger disaster · Failure analysis
9. Reynolds-averaged and large-eddy simulation persist despite growth in computing power. Explain why direct numerical simulation of turbulence at high Reynolds number remains infeasible, using an estimate of the number of grid points required.
The scale separation at the heart of the estimate: energy cascade from integral to Kolmogorov scales as Re^{3/4} in each direction.
In plain terms: The free expert account of turbulence's nested whirls — the reason resolving them all defeats any computer.
Advanced Fluid Mechanics (2.25, open course)
The governing equations and scaling from which the Re^{9/4} grid-point count follows.
In plain terms: A free MIT course supplying the arithmetic — double the Reynolds number, and the required grid explodes.
Chaos: Classical and Quantum (open book)
The dynamical-systems view: high-dimensional deterministic chaos, and why averaged descriptions remain the working recourse.
In plain terms: The free book explaining why turbulence must be summarised rather than solved — and how averaging earns its keep.
Background: Direct numerical simulation · Kolmogorov microscales
10. State the Betz limit for a wind turbine in an open flow, derive the value 16/27, and explain physically why extracting all the wind's kinetic energy is impossible.
Sustainable Energy — Without the Hot Air (open book)
Wind power quantified honestly, with the Betz factor in the technical chapters' per-area estimates.
In plain terms: The celebrated free book that does the wind arithmetic — including the fraction no turbine can exceed.
Fundamentals of Advanced Energy Conversion (2.60J, open course)
The control-volume method behind the derivation: momentum and energy accounting through an actuator disc.
In plain terms: A free MIT course with the bookkeeping that yields 16/27 — and shows why stopping the wind stops the harvest.
Introduction to Sustainable Energy (22.081J, open course)
The physical explanation completed: extracted energy requires mass flow, so the wake must keep kinetic energy to leave.
In plain terms: The free course making the impossibility vivid — dead air behind the blades would dam the very flow that feeds them.
Background: Betz's law · Wind turbine
11. A real heat engine falls short of the Carnot bound. Separate the loss attributable to finite-time operation from that attributable to irreversibility internal to the working fluid.
Thermodynamics and Kinetics (5.60, open course)
The Carnot bound and entropy accounting — the ledger in which each loss term is booked.
In plain terms: The free course establishing the ceiling, so the shortfall can be itemised against it.
Statistical mechanics: entropy, order parameters, and complexity (open textbook)
Internal irreversibility at its source: entropy production from gradients and friction inside the working fluid.
In plain terms: The free text on disorder manufactured within — the loss no slower operation can recover.
Fundamentals of Advanced Energy Conversion (2.60J, open course)
The finite-time term isolated: finite-rate heat transfer across real temperature differences, power bought with availability.
In plain terms: The free engineering course on the toll of hurry — heat moved quickly is heat moved wastefully.
Background: Carnot heat engine · Endoreversible thermodynamics
12. State the Lawson criterion (triple product) for a fusion reactor. Given that ignition has been achieved at the National Ignition Facility and long high-performance plasmas sustained in magnetic-confinement devices, identify the principal engineering obstacles that still stand between these results and net electrical power.
Progress toward fusion energy breakeven and gain as measured against the Lawson criterion
The criterion stated and every experiment placed on it: triple-product nτT against ignition and gain thresholds.
In plain terms: The free definitive scorecard — the fusion race plotted on the very axis Lawson defined.
The engineering front line: superconducting magnets, plasma-facing materials, tritium breeding, and heat exhaust at reactor scale.
In plain terms: The project's own free account of what still stands between hot plasma and a power plant.
Introduction to Plasma Physics I (22.611J, open course)
The confinement physics beneath the obstacles: transport, instabilities, and why τ is the hard-won factor.
In plain terms: A free MIT course on the unruly fourth state of matter that every obstacle traces back to.
Background: Lawson criterion · National Ignition Facility
13. The rocket equation makes staged chemical launch a tyranny of exponentials. Derive the equation, and quantify the penalty paid for a delta-v requirement twice the effective exhaust velocity.
The momentum derivation in the agency's own teaching pages: thrust, mass flow, and the logarithmic mass ratio.
In plain terms: NASA's free classroom for the equation — why every kilogram of speed costs exponentially in fuel.
Variable-mass dynamics done properly — the derivation, and the Δv = 2v_e case giving mass ratio e² ≈ 7.4.
In plain terms: The free MIT course where the tyranny becomes a number: twice the exhaust speed demands seven parts fuel in eight.
Rocket Propulsion (16.512, open course)
The engineering consequences: staging, propellant fractions, and exhaust-velocity limits of chemistry.
In plain terms: The free course on living under the exponential — why rockets are shells of fuel and stages of surrender.
Background: Tsiolkovsky rocket equation · Specific impulse
14. The square-cube law limits the scaling of structures and organisms. Give one structural and one thermal consequence, and explain why a scaled-up model of a working machine may fail where the original stood.
The Art of Approximation in Science and Engineering (6.055J, open course)
Scaling reasoning as method: stress rising with size, surface-to-volume falling — the machine-model failure predicted on the back of an envelope.
In plain terms: The free course whose central skill is exactly this — knowing what a change of scale will break.
Sizing up allometric scaling theory
The organismal side audited: metabolic and structural scaling, where the thermal consequence is a law of life.
In plain terms: A free scientific examination of how living things negotiate the same arithmetic machines must.
The classic treatment: magnitude as constraint, from insect legs to elephant bones.
In plain terms: The free masterpiece on why nothing can simply be enlarged — size itself is a design decision.
Background: Square–cube law · Allometry
15. Weibull statistics describe the strength of brittle materials. Explain why the strength of a ceramic component is a property of the population of flaws rather than of the material, and why larger specimens are weaker.
The flaw-controlled mechanism: Griffith cracks, with strength set by the worst defect present.
In plain terms: A free Cambridge package on why ceramics break from their weakest secret — not their average self.
NIST/SEMATECH e-Handbook of Statistical Methods
Weibull statistics operational: extreme-value reasoning, shape and scale parameters, size effects from flaw sampling.
In plain terms: The free handbook where 'bigger is weaker' becomes a calculable consequence of sampling more flaws.
Mechanical Behavior of Materials (3.032, open course)
The materials context: why brittle solids cannot blunt their flaws, making the population, not the lattice, decisive.
In plain terms: The free MIT course explaining the deep reason strength here is a lottery over defects.
Background: Weibull distribution · Brittleness
16. Reliability is statistical. Explain why "this component will not fail" is not an admissible engineering claim, and reformulate it as a claim that can be tested.
NIST/SEMATECH e-Handbook of Statistical Methods
The admissible reformulation: hazard rates, confidence bounds, demonstrated reliability at stated levels.
In plain terms: The free handbook that converts 'it will not fail' into a claim with numbers, tests, and error bars.
NASA Systems Engineering Handbook
The institutional practice: reliability requirements written, allocated, and verified as probabilistic statements.
In plain terms: The free agency handbook where absolute promises are forbidden by format.
Probabilistic Systems Analysis and Applied Probability (6.041, open course)
The underlying grammar: probability as the calculus in which testable engineering claims must be cast.
In plain terms: A free MIT course in the language reliability speaks — and certainty cannot.
Background: Reliability engineering · Failure rate
17. Can a system be proved safe, or only not yet unsafe? Relate your answer to the distinction between verification against a specification and the adequacy of the specification itself.
The philosophy of computer science
The distinction analysed: verification against specification, and the irreducibly empirical adequacy of the specification itself.
In plain terms: The free scholarly account of why a proof can be perfect and the system still wrong — the proof answers only the question asked.
The strongest existing case: machine-checked functional correctness — and its explicit statement of what remains assumed.
In plain terms: The free home of the world's most-verified kernel, candid about the line where proof ends.
Personal observations on the reliability of the Shuttle (Appendix F)
Specification inadequacy in the field: hazards outside the analysed envelope, discovered by reality.
In plain terms: The free classic on systems certified against the wrong question — and nature grading the real one.
Background: Formal verification · Safety engineering
18. Shannon's capacity theorem sets a bound on reliable communication over a noisy channel. Explain how low-density parity-check and turbo codes approach that bound in practice, and what "approach" costs in block length and decoding effort.
A mathematical theory of communication
The bound itself: capacity as supremum of reliable rates, proved by random coding without constructive codes.
In plain terms: The free founding paper that promised near-perfect communication — and left finding the codes to half a century.
Information Theory, Inference, and Learning Algorithms (open book)
The practical approach: sparse-graph codes and iterative message-passing decoding, with the block-length and effort costs explicit.
In plain terms: The free classic — by the man who resurrected LDPC codes — on buying Shannon's promise with long blocks and patient decoding.
Low-Density Parity-Check Codes (doctoral monograph)
The original construction, decades early: the codes and iterative decoding that would later approach capacity.
In plain terms: The free 1963 thesis that invented the winning codes before the hardware existed to run them.
Background: Noisy-channel coding theorem · Low-density parity-check code
19. Control an underactuated system: a system with fewer actuators than degrees of freedom. Explain, using an example, why underactuation makes some trajectories dynamically inaccessible.
Underactuated Robotics (open course book)
The subject's home text: dynamics with fewer actuators than freedoms, acrobot and cart-pole showing inaccessible directions.
In plain terms: The free MIT book on machines that cannot push every way they can move — and must swing instead of steer.
Feedback Systems: An Introduction for Scientists and Engineers (open edition)
The controllability framework: reachable sets and the conditions under which trajectories exist to be commanded.
In plain terms: The free standard text on what control can and cannot reach — the general law behind the example.
Engineering Dynamics (2.003SC, open course)
The mechanics substrate: degrees of freedom, constraints, and generalised forces — the counting the question turns on.
In plain terms: The free course where 'fewer actuators than freedoms' becomes a precise, countable statement.
Background: Underactuation · Nonholonomic system
20. Chaotic systems can be controlled by small, timely perturbations. Explain the principle by which sensitivity to initial conditions becomes a control resource rather than an obstacle.
The principle from its co-inventor: stabilising unstable periodic orbits with small parameter perturbations — OGY at source.
In plain terms: The free expert page on chaos tamed — tiny nudges, timed by the system's own sensitivity.
Chaos: Classical and Quantum (open book)
The skeleton exploited: dense unstable periodic orbits threading the attractor — the targets small control selects among.
In plain terms: The free book revealing chaos's hidden order — the infinity of orbits a clever controller can choose between.
Nonlinear Dynamics: Chaos (12.006J, open course)
The sensitivity itself: Lyapunov exponents and stretching — the amplifier that turns small inputs into large steering.
In plain terms: A free MIT course on the exponential lever — the very instability that wrecks prediction powers control.
Background: Control of chaos · Butterfly effect
21. Landauer's principle sets a thermodynamic floor on the cost of erasing information. State the bound, and explain why logically reversible computation escapes it while logically irreversible computation does not.
Stochastic thermodynamics of computation
The modern statement: kT ln 2 per erased bit as a special case of general dissipation bounds on logical operations.
In plain terms: The free treatise pricing computation in heat — with erasure as the operation that must pay.
The mechanism made physical: logical irreversibility as phase-space compression, demanding entropy export.
In plain terms: The free review connecting forgetting to heating — why losing a bit means warming the room.
Back to the future: the case for reversible computing
The escape route engineered: logically reversible architectures that sidestep the bound by never erasing.
In plain terms: The free case for computers that remember everything — and thereby owe the thermodynamic toll on nothing.
Background: Landauer's principle · Reversible computing
22. Dennard scaling has ended. Explain what Dennard scaling asserted, why its end decoupled transistor count from usable performance, and what "dark silicon" names.
Design of ion-implanted MOSFETs with very small physical dimensions
The original scaling laws: dimensions, voltage, and current shrinking together at constant power density.
In plain terms: The free 1974 paper whose promise — smaller, faster, no hotter — powered forty years of computing.
Microprocessor trend data (open dataset)
The decoupling in data: transistor counts climbing while frequency and power flatten — the end, plotted.
In plain terms: The free canonical charts showing exactly where the free lunch stopped.
The chips are down for Moore's law
The aftermath surveyed: power-limited utilisation — dark silicon — and the turn to specialisation.
In plain terms: The free Nature feature on the era after scaling — chips rich in transistors they cannot afford to light.
Background: Dennard scaling · Dark silicon
23. A power grid loses synchronous rotating generation. Explain the role of mechanical inertia in frequency stability, and why grids dominated by inverter-based sources require synthetic inertia.
Introduction to Electric Power Systems (6.061, open course)
The swing dynamics: rotating kinetic energy buffering load imbalance, frequency as the grid's balance gauge.
In plain terms: The free MIT course on the spinning mass that gives the grid its seconds of grace.
Electric Machines (6.685, open course)
The electromechanics of the synchronous machine — the coupling that makes inertia a grid service at all.
In plain terms: The free course on the great rotating machines whose physics inverters must now imitate.
Sustainable Energy — Without the Hot Air (open book)
The system context: renewable-dominated supply, fluctuation, and the balancing problem synthetic inertia serves.
In plain terms: The free classic on running a grid from sources with no flywheel — the reason imitation inertia is needed.
Background: Utility frequency · Electrical grid
24. Design a self-healing material. Specify what "healing" must mean physically, and identify the trade-off between healing capacity and load-bearing capacity.
Mechanical Behavior of Materials (3.032, open course)
The trade-off's terms: strong fixed cross-links versus mobile reversible ones — stiffness bought against reparability.
In plain terms: The free course supplying the tension the design must resolve — what holds hard cannot easily re-knit.
What healing must physically undo: crack surfaces, broken bonds, and the energy of new area.
In plain terms: A free Cambridge package defining the wound — the crack — any healing must genuinely close.
Supramolecular chemistry — scope and perspectives (Nobel lecture)
The healing chemistry's foundation: reversible, self-assembling bonds — the mechanism by which broken can rejoin.
In plain terms: The free lecture on bonds designed to re-form — the chemical principle every self-healing scheme borrows.
Background: Self-healing material · Supramolecular chemistry
25. Biomechanics: bone remodels in response to load. Treating bone as a control system with a set-point, identify the sensed variable, the actuator, and the failure mode when the loop is opened by disuse.
Exercise, nutrition, hormones, and bone tissue
The loop's phenomenology: load driving deposition, disuse driving resorption — the open-loop failure in orbit and bed rest.
In plain terms: The free textbook page on bone's bargain with gravity — and what happens when the loading signal is unplugged.
Feedback Systems: An Introduction for Scientists and Engineers (open edition)
The control vocabulary applied: sensed variable, set-point, actuation, and the consequences of opening the loop.
In plain terms: The free control text supplying the exact grammar the question asks the biology to be written in.
Bone formation and development
The actuators identified: osteoblasts and osteoclasts — the paired effectors the controller commands.
In plain terms: The free companion page naming the demolition and construction crews the loop directs.
Background: Wolff's law · Bone remodeling
26. Finite-element modelling discretises a continuum. Identify two distinct ways in which a converged finite-element solution can nonetheless be wrong about the physical structure.
Finite Element Analysis of Solids and Fluids I (2.092, open course)
The method proper: weak forms, elements, convergence — what a converged solution is converged to.
In plain terms: The free MIT course on the machinery, making clear the answer converges to the model, not to the world.
NASA Systems Engineering Handbook
Verification against validation institutionalised: solving the equations right versus solving the right equations.
In plain terms: The free handbook enforcing the two questions the question is about — and refusing to let one answer the other.
The epistemology of experiment in physics
The validation side's epistemology: how measurement earns the authority to correct a converged model.
In plain terms: The free scholarly account of why the test rig outranks the mesh.
Background: Finite element method · Discretization
27. A pressure vessel must be designed against both yielding and fracture. Explain why a leak-before-break criterion can be safer than a design that merely maximises burst pressure.
The governing property: critical crack size from toughness and stress — the quantity leak-before-break engineers.
In plain terms: The free primer on how big a flaw a wall can carry — the number that decides leak or burst.
The catastrophic alternative characterised: unstable crack propagation once criticality is reached.
In plain terms: A free Cambridge package on the failure mode the criterion is designed to forbid.
Fracture and Fatigue (3.35, open course)
The design logic: wall thickness and toughness chosen so a through-crack leaks detectably before it runs.
In plain terms: The free course on the safer bargain — a warning drip preferred to a maximised, silent bomb.
Background: Pressure vessel · Fracture toughness
28. Communications and control share the concept of feedback but use it differently. Distinguish the role of feedback in a phase-locked loop from its role in a feedback amplifier.
Circuits and Electronics (6.002, open course)
The amplifier's use: feedback trading gain for linearity, bandwidth, and insensitivity — regulation of a value.
In plain terms: The free MIT course on feedback as discipline — taming a wild amplifier into an exact one.
Introduction to Communication, Control, and Signal Processing (6.011, open course)
The loop's use: feedback as synchronisation — tracking phase and frequency of a reference, acquisition and lock.
In plain terms: The free course where feedback becomes a follower — locking one rhythm to another.
The origin of the amplifier half: Black's negative-feedback insight, in the historical record.
In plain terms: The free history of the ferry-ride idea — give away gain, receive fidelity — that made modern electronics.
Background: Phase-locked loop · Negative-feedback amplifier
29. Process engineering: a continuous chemical reactor must be stable at its operating point. Explain how an exothermic reaction can produce multiple steady states, and how runaway is prevented by design rather than by operation.
Chemical and Biological Reaction Engineering (10.37, open course)
The multiplicity analysis: exponential heat generation against linear removal — intersections giving multiple steady states.
In plain terms: The free MIT course with the famous crossing curves — three possible operating fates for one reactor.
T2 Laboratories Inc. reactive chemical explosion (investigation report)
Runaway realised: cooling capacity outrun by reaction heat — the design inadequacy no operator could retrieve.
In plain terms: The free federal investigation of a reactor that crossed the line — the case for designing out what vigilance cannot catch.
Chemical Engineering (open bookshelf)
The design remedies catalogued: inherently safer sizing, relief, dilution, and stable-branch operation.
In plain terms: A free reference shelf on prevention by architecture — reactors built so runaway has nowhere to start.
Background: Continuous stirred-tank reactor · Thermal runaway
30. Aerospace structures are certified partly by test and partly by analysis. Argue for the proposition that certification is an epistemic activity: what does a certified structure entitle its designer to believe, and on what grounds?
Certification as institutional epistemology: entitlement to belief grounded in specified evidence and procedure.
In plain terms: The free scholarly frame for the claim — a certificate as a warranted, bounded reason to believe.
The epistemology of experiment in physics
The grounds analysed: how test evidence, with calibration and error control, licenses inference beyond the article tested.
In plain terms: The free account of what a passed test really entitles — and how far past the test rig the entitlement travels.
Aircraft Systems Engineering (16.885J, open course)
The practice itself: building-block test pyramids and analysis credit — the composite warrant certification assembles.
In plain terms: The free MIT course — taught with shuttle and airliner cases — on how belief in a structure is actually earned.
Background: Type certificate · Verification and validation