物理 · Physics

Beyond the Standard Model

Day 27 · 2026 · Phase F Particles & the Standard Model
A theory accurate to the tenth decimal place, and the floor under its feet isn't even on its list. The clue isn't where it gets things wrong — it's where it has no entry at all.
The Standard Model computes the electron's magnetic moment to ten decimal places and the experiment agrees. Yet in several places it isn't a matter of "not precise enough" — there is simply no entry on the list: gravity isn't on it; ninety-five percent of the universe isn't on it; neutrinos plainly have mass and the theory cannot write one down. And in forty years, every proposal answering "where is the next new particle" has come back empty from the colliders.

The One Crack You Can Hold in Your Hand

Neutrino oscillation · 1998 · Nobel 2015
Intuition A beam of neutrinos leaves as a single "flavour," and a few hundred kilometres later part of it measures as a different one — then swings back. Periodic identity-swapping has exactly one prerequisite: the thing must carry a clock that ticks. And for anything travelling at the speed of light, its own clock is frozen. To change flavour is to travel slower than light, which is to have mass.
Mechanism A neutrino carries two sets of identities that don't line up. What shows at birth and at detection is the flavour — whether the particle born alongside it is an electron, a muon or a tau. What evolves on a fixed beat in flight is the mass state. One flavour is a superposition of several mass states; different masses run their phases at different rates, and the accumulated difference keeps re-mixing the superposition.
P = sin22θ · sin2(1.27 · Δm2L / E)
P is the probability of having turned into the other flavour; θ is read theta, the mixing angle, which fixes how wide the swing is; Δ is read delta and means "the difference between two things," so Δm2 is the difference of the squared masses; L is the distance flown and E the energy; 1.27 is only the factor that makes the units line up. Note that only the difference of masses appears — oscillation cannot tell you how heavy a neutrino actually is.
The beam changes identity as it flies 1 0 distance flown ÷ energy → still the flavour it set out as already turned into another A mass term must stitch left to right electron left-handed electron right-handed H it stitches → there is mass neutrino left-handed right-handed? never seen H no stitch → must be zero in the SM but experiment says it isn't zero
Left: the probability swings with distance, and the swing itself demands mass. Right: the recipe that gives fermions mass is missing half its ingredients for the neutrino.
The counterintuitive part The Standard Model didn't "forget" to give neutrinos mass; structurally it cannot. A fermion mass stitches a left-handed and a right-handed version together (that machine was taken apart in Symmetry and the Origin of Mass), and a right-handed neutrino has never been observed — so the mass can only be exactly 0 (the digit zero). In 1998 Super-Kamiokande saw atmospheric neutrinos thin out with the distance travelled through the Earth, and in 2001 SNO showed that solar neutrinos had only changed flavour. That zero is dead. It remains the single place where the Standard Model has been broken in a laboratory rather than inferred from the sky.
Cross-disciplinary reading · Engineering / AI
  • Engineering: neutrinos from a reactor pass straight through the building, and detecting them from outside reads off what fuel is in the core — a signal that cannot be faked. It is seriously studied as a safeguards tool for verifying nuclear material.
  • AI: the same object is "cleanly labelled" in one basis (flavour) and "simple to evolve" in another (mass), with the two rotated against each other. That is exactly the job a Fourier-type transform does.
In one line: neutrinos change identity, changing identity needs a ticking clock, and that clock is mass.
Think: if oscillation only measures a difference of masses, how do we know how heavy a neutrino is?
Two routes. Weigh it directly: look at the tiny bite the neutrino mass takes out of the very end of the electron spectrum in tritium decay — the limit is about 0.8 eV, six hundred thousand times lighter than the electron. Or look at the universe: heavy neutrinos would smooth out structure at the galaxy-cluster scale. That bound is tighter but leans on a cosmological model.

Our Existence Is the Counterexample

Baryon asymmetry · Sakharov 1967
Intuition In the first heat of the Big Bang, particles and antiparticles are born in pairs and die in pairs, with the rules showing no favour to either. Run that script to its end and the cooling universe should hold nothing but photons — not a single atom. Yet there are atoms, and there is you. So somewhere the script was biased: in every billion pairs, matter came out one ahead, and that leftover is every galaxy there is.
Mechanism The bias can be measured, and it is tiny:
η = nB − nB̄nγ ≈ 6 × 10−10
η is read eta; n is a number density, "how many per unit volume"; the subscript B means baryons (protons, neutrons and their kin), B with a bar means antibaryons, and γ is read gamma, the photons. The numerator is the excess of matter over antimatter; the denominator is essentially "the light the annihilation produced." Two unrelated measurements — the speckle of the microwave background and the abundances of the light elements — agree on this number.
Sakharov pointed out in 1967 that to get this number out of a symmetric start, three things must hold at once: baryon number must be changeable; both C and CP symmetry must be broken (otherwise every route that makes matter is matched by an equally fast one that makes antimatter); and the system must depart from thermal equilibrium (in equilibrium the forward and backward rates match and any net excess is erased — that machine was covered in Entropy and the Arrow of Time). The Standard Model grazes all three and satisfies none.
An almost perfect cancellation matter 10⁹+1 antimatter 10⁹ 1 left a billion pairs annihilate, one is left over stars, planets and you are that remainder Sakharov's three conditions ① Baryon number must be changeable There is a route, open only at extreme heat ② C and CP must both be broken Present, but ten orders of magnitude too small ③ It must leave thermal equilibrium Needs a violent transition; it was smooth All three required; the SM only grazes each
The two bars differ in height by one part in a billion — too little to draw. The small block on the right is the entire difference.
The counterintuitive part This is not a shortfall that finer calculation can close. A ten-order-of-magnitude gap means a new source of CP violation is required. The most-watched candidate hooks straight onto the neutrino thread: if very heavy right-handed neutrinos exist, the asymmetry in their decays first creates an imbalance in lepton number, which quantum effects at high temperature then convert into an imbalance in baryon number — the mechanism that gives neutrinos mass might also explain why there is anything at all. This is called leptogenesis, and it remains conjecture.
Cross-disciplinary reading · Chemistry / Cosmology
  • Chemistry: terrestrial life uses almost exclusively left-handed amino acids and right-handed sugars. Basic chemistry treats the two handednesses alike, yet the outcome is overwhelmingly one-sided — again "a minute bias amplified by feedback into a total majority," with a different amplifier.
  • Cosmology: if some region really held a slab of antimatter, annihilation at the boundary would emit characteristic gamma rays. Space telescopes have looked hard and found none — the asymmetry is global.
In one line: matter and antimatter very nearly cancel on the books, and everything visible is the one-in-a-billion rounding error.
Think: why not simply assume the universe started with a little more matter?
You can write it into the initial conditions, but inflation dilutes any pre-existing net baryon number to nothing. To regenerate the number after inflation you are back to Sakharov's three conditions — putting it in by hand is giving up on explaining it.

Grand Unification, and the Prediction That Killed It

Grand unification · 1974 · Proton decay
Intuition The "strength" of a force is not a constant; it shifts with the energy you use to look at it. Electromagnetism grows stronger at high energy because you punch through the screening that virtual pairs popping out of the vacuum provide; the strong force does the opposite and weakens up close. Extrapolate the three forces a dozen orders of magnitude up the energy axis and they converge on roughly the same place. If they truly met at a point, the three would be one force that split into three at low energy.
Mechanism
1/αi(μ) = 1/αi(MZ) − bi2π ln(μ / MZ)
α is read alpha, the coupling strength; written as 1/α, a larger value means a weaker force. The subscript i (the letter i) numbers the three forces; μ is read mu, the energy you observe at; MZ is the Z boson's mass, used as a reference point; bi is fixed by which particles the theory contains, and its sign decides whether the line slopes up or down; ln is the natural logarithm — so each decade of energy shifts the strength by the same small step.
How the three forces drift with energy Vertical: 1/coupling (higher = weaker force).　Horizontal: energy, one decade per gap 1/α₁　hypercharge 1/α₂　weak 1/α₃　strong they cross pairwise, not at one point 10² weak scale 10⁶ 10¹⁰ 10¹⁴ 10¹⁸ GeV Add supersymmetric partners and the three very nearly meet — its strongest indirect case gravity enters at 10¹⁹
The scale is logarithmic. Extrapolated with the particles the Standard Model actually has, the three lines cross at three different places and enclose a small triangle.
In 1974 Georgi and Glashow packed the three symmetry groups together into the larger SU(5). That step incidentally explained something that had never had a reason: why the electron's charge and the proton's match to the last digit. In the Standard Model those are separate numbers that happen to line up; in a grand unified theory quarks and leptons sit in the same multiplet, so the charges must be in whole-number ratios.
The counterintuitive part The same structure also lets a quark turn into a lepton — the proton decays. This is grand unification's cleanest, most falsifiable prediction. The simplest version gave a lifetime of roughly 1030 to 1031 years, so a few hundred tonnes of water watched for a few years should catch one. Super-Kamiokande watched fifty thousand tonnes for decades and caught nothing, pushing the limit past 1034 years. The most beautiful prediction is precisely the one the experiment killed. What survives are more elaborate versions, plus an awkward fact: the three lines don't actually meet at a point. Whether "so nearly unified" is a deep clue or a numerical coincidence is still unanswered.
Cross-disciplinary reading · Mathematics / Engineering
  • Mathematics: there aren't many ways to embed three groups inside one larger group, and group theory can nearly enumerate the candidates — SU(5), SO(10) and E₆ are the main ones. Once again, so many constraints that almost no freedom is left.
  • Engineering: the giant water tanks built to find proton decay never found it, and became the most important detectors in neutrino physics — the neutrinos from the 1987 supernova and the oscillation of atmospheric neutrinos both came out of them. An instrument's value often isn't in the job it was designed for.
In one line: the three forces look as though they are about to meet at enormous energy, and whether "about to" counts rests on a signal that decades of searching have not produced.
Think: if proton decay is never seen, is grand unification falsified?
Only its simplest version. Take a larger group, push the unification scale a little higher, and the predicted lifetime hides beyond current sensitivity. The cost is more free parameters and weaker predictions — which is exactly the setting for the usual argument about adding parameters to escape falsification.

Coming Back Empty-Handed Is Also a Result

Experimental frontiers · energy / precision / cosmos
Intuition There are only three ways to hunt for new physics: collide at higher energy and make the new particle directly; compute and measure known quantities so precisely that the tenth decimal place has to agree; or let the universe be the accelerator and wait for dark matter to hit a detector. All three have advanced, and none has brought back a new particle.
Mechanism The precision route carries a brutal arithmetic:
statistical error ∝ 1 / √N
N (capital letter N) is the number of events collected, √ is the square root, ∝ reads "is proportional to." To shrink the uncertainty tenfold you need a hundred times the data. That is why every small step takes years.
The actual score looks like this. The muon's anomalous magnetic moment once sat four to five standard deviations away from theory and was the most-hoped-for crack; as the hadronic part of the calculation moved to lattice methods, theory drifted toward the measurement and the tension narrowed sharply. The ratio of B mesons decaying to electrons versus muons once hinted at broken lepton universality, and reverted to the Standard Model once the data doubled.
Fifteen years of finding nothing bought this Everything below a line is excluded; horizontal is the hypothetical mass, lower = tighter 2012 2018 2024 neutrino fog: solar and atmospheric neutrinos begin faking the signal 10 GeV 100 1000 10⁴ GeV Nothing found ≠ nothing learned: three orders of magnitude of possibility removed
Three generations of detector, three upper limits. The space between them is what fifteen years ruled out for good.
The counterintuitive part Calling this "failure" is bad bookkeeping. Excluded parameter space is real knowledge, and the fact that many three-sigma hints eventually evaporate is a sign the process is working. What has genuinely been shaken is an expectation: the confidence that "new particles should show up near a TeV" came from naturalness, an aesthetic principle, not from any experiment. The possibilities that now have to be taken seriously are that new physics exists at an energy this generation's machines cannot reach — or that some numbers are simply what they are, with no deeper reason.
Cross-disciplinary reading · AI / Scientific method
  • AI: in a large search, a negative result is worth however much space it eliminates. Testing parameters one point at a time produces almost no knowledge; the experiment that rules out a whole region at once is the valuable one — the same logic as picking the sample that best discriminates between hypotheses in active learning.
  • Scientific method: blind analysis is standard in particle physics — the pipeline is fixed first and the real numbers unblinded last, precisely to prevent tuning selection cuts while staring at the data until three sigma appears. Other fields adopted such rules only after a replication crisis; this one paid the tuition early.
In one line: the most important experimental results of these years are a long list of "no" — which cut away most of the possibilities, and also the field's confidence in its own intuition.
Think: if new physics sits at an energy we can never build, is physics finished?
Not really. The work shifts to the precision and cosmic routes: new physics at high energy leaves faint traces low down — branching ratios of rare decays, electric dipole moments — and the early universe is an accelerator nobody could afford, with its residue written into neutrinos, gravitational waves and the microwave background.

Going Deeper

Will the Standard Model ever be "overthrown"?
Almost certainly not — it will be enclosed. It is the norm for an old theory to become the limit of a new one at some scale: relativity did not void Newtonian mechanics, it drew a boundary around it. The Standard Model is already used as an effective theory — correct below some energy, replaced by another description above it. The real question isn't whether it is right, but where it stops working.
Why does gravity count as something the Standard Model fails to explain, rather than another theory's business?
Because quantising general relativity the old way gives a theory that runs out of control at high energy: every deeper order throws up new parameters that only experiment can fill in, until infinitely many are needed and predictive power is gone. At everyday scales it remains an excellent effective theory; the trouble sits near the Planck scale, and that chasm is handled in The Open Problems of Physics.
Is "nineteen free parameters" a defect?
A dissatisfaction, not a contradiction. Parameters have to be filled in by measurement, which means the theory doesn't explain where they come from; but turning a finite set of inputs into a vast body of precise predictions is itself enormous compression. What is unsettling is how unevenly the values are spread — the top quark and the electron differ in their coupling to the Higgs field by more than three hundred thousand times (The Standard Model tabulated this) — as though there were structure nobody has read yet.

Further Reading