The Vacuum Has a History: Symmetry Breaking as Speciation

A resonance near 125 GeV was a measurement of the vacuum’s stiffness

On 4 July 2012, the ATLAS and CMS collaborations at the Large Hadron Collider each reported a new boson near 125 GeV, ATLAS quoting a mass of 126.0 ± 0.4 (stat) ± 0.4 (syst) GeV at 5.9 standard deviations and CMS a mass of 125.3 ± 0.4 (stat) ± 0.5 (syst) GeV at 5.0 standard deviations, combining searches across the diphoton, four-lepton, and other channels in proton-proton collisions at 7 and 8 TeV ^{[1, 2]}. The press coverage called it a particle discovery, which it was. It was also something less commonly said out loud: a direct measurement of a property of empty space.

The Standard Model does not merely contain a Higgs boson; it contains a Higgs field that is nonzero everywhere, including in what any experiment can prepare as vacuum. The particle that ATLAS and CMS detected is the observable excitation of that field around its resting value, and the excitation’s mass is fixed by the field’s self-coupling and its resting value together. Measuring the boson’s mass is, in that specific technical sense, measuring how strongly the vacuum resists being pushed away from the state it has already settled into — its stiffness, in the same sense that a spring’s stiffness is inferred from how it responds when disturbed. A universe with a different vacuum, or no condensed Higgs field at all, would have no such resonance to find, and no masses for the W and Z bosons, the quarks, or the charged leptons, because in the Standard Model those masses are not independent inputs; they are all read off the same condensed field.

This article is about what it means for a vacuum to be a condensed phase with a history rather than an inert backdrop, and about how far an analogy to biological evolution can honestly be pushed to describe that history. The analogy is not decoration. Spontaneous symmetry breaking really is a branching event: a system with several equally available outcomes commits to one, and the commitment is preserved afterward as structure. That much is exact, and it is why physicists have reached for words like “frozen” and “relic” for decades without needing biology’s permission. But a branching event is not a lineage, and freezing is not selection. The place where the analogy stops working is as informative as the place where it starts.

Spontaneous symmetry breaking is the vacuum selecting an angle, not a force pushing anything

The simplest version of the mechanism starts from a single complex scalar field $\phi$ with a potential energy density of the form

Plotted against the two real components of $\phi$, this is the shape popularly called the Mexican hat: a local maximum at $\phi = 0$, surrounded by a circular trough of degenerate minima at $|\phi| = \mu/\sqrt{2\lambda}$. The sign of the mass term is the entire mechanism. With $\mu^2 < 0$ the origin would be the unique minimum and there would be nothing to discuss; it is the negative mass-squared term, an assumption built into the Standard Model’s Higgs sector rather than derived from a deeper principle, that manufactures the trough in the first place. Nothing in the Lagrangian singles out any point on that trough. A system placed at the unstable summit will roll downhill in a direction fixed only by whatever infinitesimal fluctuation happens to nudge it first, and the direction it lands on becomes, from that moment, the vacuum every subsequent measurement will be made against.

This is the sense in which the term “spontaneous” earns its keep: the breaking is not driven by an external agent or a change in the system’s fundamental couplings, but by the system finding that its symmetric configuration is not its most stable one and settling into an asymmetric one that the underlying law permits without requiring. Goldstone’s 1961 theorem made the general consequence precise for any theory with a continuous global symmetry broken this way: for each broken generator of the symmetry, the spectrum must contain a massless scalar, later called a Goldstone boson, corresponding to fluctuations along the flat direction of the trough where no energy cost is incurred by sliding from one equivalent minimum to another ^[5]. A field that could slide freely around the whole circle of minima has, in exactly that direction, zero restoring force and therefore zero mass.

The Higgs mechanism eats the boson the ferromagnet’s spins never have to give up

The theorem is where the story becomes genuinely subtle, because the Standard Model’s electroweak sector does not contain any observed massless scalar of the kind Goldstone’s argument predicts, and the reason is itself a discovery rather than a loophole. Englert and Brout, and independently Higgs, showed in 1964 that when the broken symmetry is a local gauge symmetry rather than a global one, the story changes at the last step ^{[4, 3]}. The gauge bosons associated with the broken generators absorb the would-be Goldstone modes as their own longitudinal polarization states, and in doing so become massive. What would have been a massless, physically propagating Goldstone boson is not observed as a separate particle because it has been consumed as a component of a gauge boson that previously, being massless like the photon, had no longitudinal mode to have. Three of the electroweak sector’s four generators are broken this way, giving mass to the $W^+$, $W^-$, and $Z^0$; the fourth, corresponding to the unbroken electromagnetic $U(1)$, is not, which is why the photon remains massless. Weinberg’s 1967 paper assembled this mechanism into the specific gauge theory of leptons that, combined with the analogous quark sector, became the electroweak part of the Standard Model tested at the LHC decades later ^[6].

The popular explanation for all of this reaches almost automatically for a ferromagnet: above the Curie temperature the spins point every which way and the material has full rotational symmetry; below it, the spins align along some direction chosen by chance, and the rotational symmetry is broken in the resulting magnetization. The analogy is not wrong about the qualitative picture — a system with a symmetric high-energy configuration commits to a lower-energy asymmetric one — but it actively misleads on the one point that made 1964 a discovery rather than a restatement of Goldstone’s result. A real ferromagnet’s broken symmetry is global: spins point in a common direction, but nothing in the theory gauges that direction, and the theorem’s genuine massless Goldstone modes appear in the material as spin waves, or magnons, propagating at arbitrarily low energy. The electroweak vacuum has no analogous massless mode among its would-be Goldstone bosons for the simple reason that the symmetry being broken is local, not global; the modes are not there to be found because they have been absorbed into the W and Z as mass. A ferromagnet does not have a hidden mechanism by which its magnons quietly become anything else, because rotations of ordinary space are not a gauge symmetry of the magnet’s own dynamics. The analogy therefore illustrates the setup correctly and the payoff not at all — which is exactly backwards from what an explanatory device should do, and worth naming rather than passing over.

There is a second, more technical correction worth making honestly rather than skating past: a strict reading of Elitzur’s theorem holds that a local gauge symmetry cannot be spontaneously broken in the same sense a global one can, because any local gauge-variant quantity that could serve as an order parameter has vanishing expectation value in a properly gauge-fixed path integral. The physically meaningful, gauge-invariant statement is not “the gauge symmetry is broken” but rather a statement about the spectrum and interactions of gauge-invariant composite operators, which in the electroweak theory reproduces the same masses and couplings that the informal “broken symmetry” language predicts. Practicing physicists use the broken-symmetry language constantly and correctly, because in a fixed gauge such as unitary gauge it is calculationally identical to the gauge-invariant statement and vastly easier to compute with. But it is worth saying once, plainly, that the vacuum is not literally choosing a direction in some internal space the way a ferromagnet’s spins choose a direction in ordinary space; it is settling into a particular gauge-invariant configuration of fields, and the “chosen direction” language is a convenient fiction tied to a choice of gauge, not a physical pointer one could in principle go and look at.

Two more transitions gave the particle spectrum its sediment

The electroweak transition is not the only phase transition the early universe underwent, and treating it in isolation obscures how much of what looks like fine-tuned coincidence in particle masses is really historical residue from a specific sequence of events. As the universe cooled from a hot, symmetric phase, lattice calculations put the electroweak transition at an energy scale around 100 GeV, and non-perturbative lattice work by Kajantie, Laine, Rummukainen, and Shaposhnikov showed that for Higgs masses above roughly 70–95 GeV — with their best estimate for the endpoint of the first-order transition line near 80 GeV — there is no true electroweak phase transition at all, only a smooth analytic crossover ^[9]. Because the measured Higgs mass is about 125 GeV, comfortably above that endpoint, the real universe went through the electroweak transition as a crossover: a continuous change in the Higgs field’s typical value with falling temperature, not a sudden jump with latent heat, bubble nucleation, or the kind of sharp discontinuity that would have left detectable relics such as a stochastic gravitational-wave background from colliding bubbles. This matters beyond bookkeeping, because a strongly first-order electroweak transition was for decades the leading candidate mechanism for generating the universe’s matter-antimatter asymmetry through electroweak baryogenesis; a crossover forecloses that specific mechanism in the minimal Standard Model and is one of the standing reasons particle theorists are confident new physics beyond the Standard Model must exist somewhere, even without any single new particle yet found to require it.

A second transition follows close behind, at the vastly lower temperature scale of quantum chromodynamics. Aoki, Endrodi, Fodor, Katz, and Szabo showed with lattice calculations at physical quark masses, using finite-size scaling across simulation volumes differing by a factor of five, that the QCD confinement transition in the early universe is likewise an analytic crossover rather than a genuine phase transition, meaning quarks and gluons transition into confined hadrons through a rapid but continuous change rather than a discontinuous jump ^[10]. Between these two crossovers sits essentially the entire mass budget of ordinary matter as experienced day to day: the electroweak transition fixes the masses of the fundamental quarks and charged leptons through their couplings to the Higgs field, while the QCD transition, occurring at a temperature scale roughly a thousand times lower, generates most of the proton and neutron’s mass dynamically, from the energy of confined gluon fields and quark motion rather than from the quarks’ own comparatively tiny Higgs-generated masses. A proton’s mass is overwhelmingly QCD sediment, not electroweak sediment; the fine-grained content of visible matter is a layered deposit from two separate, sequential freezings, each of which locked in structure the next era had no choice but to inherit. Grand unified extensions of the Standard Model posit still earlier symmetry-breaking transitions at energy scales far beyond anything colliders can reach, potentially unifying the strong and electroweak forces into a single gauge group broken at some very high scale; this is a coherent and actively studied theoretical possibility, but it remains unconfirmed by any direct experimental signature such as the proton decay it would generically predict, and it belongs in this account only as a flagged extrapolation, not as an established rung on the ladder.

One branching event is not a lineage

It is at this point that the analogy to biological speciation becomes genuinely useful, and where its limits need to be stated as carefully as its reach. A branching event in evolution occurs when a population’s shared ancestry diverges into lineages that no longer interbreed, and which one to their genetic ground state each lineage settles into afterward, is not derivable from first principles; it is a frozen accident of which mutations arose and which environmental pressures happened to be in force, preserved afterward by reproduction and expressed as morphology. Symmetry breaking shares the structurally important part of that story: a system with several equally available ground states commits to one via a process that involves no further law beyond the initial instability, and the commitment persists afterward as observable structure — masses, couplings, the entire content of the Standard Model as measured today. In that specific, narrow sense, the particle spectrum is a fossil record exactly as a genome is a fossil record: both preserve information about a historical event that cannot be re-derived from present-day physics alone, only inferred from what it left behind.

The analogy fails, and fails precisely, at the mechanism that does the actual explanatory work in biology: selection acting on variation within a population. Speciation requires more than one lineage existing and reproducing differentially; it requires organisms competing, differential survival, and heritable variation accumulating generation over generation. Electroweak symmetry breaking, so far as any experiment has ever probed it, is a single trajectory taken by a single universe. There is no population of universes we can observe having tried the other angle on the Mexican hat’s trough and having done comparatively worse; there is no competition, no reproduction, and no differential fitness among alternatives, because within our own observable patch of spacetime there are no alternatives to compare against — only the one outcome that happened, and inference about what other outcomes were, in principle, available. Where evolutionary biology explains present structure by pointing at a process of trial, competition, and retention operating on a real population, particle physics can only point at contingency plus constraint: contingency in which vacuum got selected, constraint in what any resulting vacuum could look like given the symmetries built into the underlying gauge theory. That is a real and non-trivial similarity, but it is a much smaller claim than a full evolutionary analogy implies, and treating it as anything more risks manufacturing a mechanism that is not there.

The one place cosmology offers a mechanism that would restore something closer to a genuine population is explicitly speculative and belongs to string theory’s landscape of vacua, not to established physics. Susskind has argued, in a widely discussed and self-consciously provisional lecture, that string theory’s compactifications may admit an “unimaginably large and diverse” landscape of distinct vacuum states, each corresponding to a different local minimum with its own effective particle content and constants, and that this diversity lends credibility to anthropic reasoning about why our particular vacuum has the properties it does ^[18]. If eternal inflation populates different regions of a much larger multiverse with different vacua from this landscape, something resembling a real population under something resembling selection — in the weak, anthropic sense that only life-permitting vacua get observed by anyone capable of observing them — becomes conceivable. This is worth stating plainly as a live and contested research program with named advocates and named critics, not as an established extension of the confirmed physics described above; it has no confirmed observational signature, no consensus among theorists that string theory’s landscape is even finite in the relevant sense, and no way at present to test whether our own patch’s vacuum was selected from a real ensemble or is simply the only kind of vacuum there ever was to have. The honest statement is that the speciation analogy, absent the landscape, describes one accident preserved by physics rather than one outcome selected from a population, and that the landscape is the one place in the literature where someone is seriously proposing to close that gap — while conceding it remains a proposal.

Coleman’s semiclassical theory turns “is our vacuum final” into a calculation

If the electroweak vacuum is a local minimum rather than the global one, then in a quantum theory it is not perfectly stable; it can, in principle, decay by quantum tunneling to a lower-energy configuration, a possibility with no analogue in classical statistical mechanics. Coleman’s 1977 paper “Fate of the false vacuum” put the semiclassical machinery for this process on rigorous footing, treating decay as bubble nucleation: a small region of true vacuum appears through tunneling, and if the bubble is large enough that the volume energy released by the lower-energy interior outweighs the surface energy cost of the wall separating it from the surrounding false vacuum, the bubble expands, eventually at close to the speed of light, converting everything it touches ^[7]. The decay rate per unit four-volume in this picture takes the schematic form

where $S_E$ is the Euclidean action evaluated on the classical solution that mediates the tunneling — the “bounce” — and $A$ is a prefactor from fluctuations around that solution. The exponential dependence on the bounce action is the entire reason a metastable vacuum can be, for all practical purposes, permanent: a large action makes the exponent enormous and negative, suppressing the decay rate to a number that can be many, many orders of magnitude below one event per observable volume per age of the universe, without the vacuum being absolutely stable in any stronger sense. Coleman and De Luccia extended the calculation to include gravity in 1980, showing that gravitational corrections to the bounce action are not always negligible and can, in some regimes, dominate the outcome, particularly in the late stages of bubble growth and in cases where the vacuum energies involved are comparable to the Planck scale ^[8]. For the scales relevant to electroweak vacuum decay, later work established that as long as the bounce’s characteristic scale stays far below the Planck scale, gravitational corrections to the action remain a small, calculable correction rather than a qualitative change to the result, which is part of why modern calculations can treat the calculation in flat spacetime with reasonable confidence ^[12].

The measured Higgs and top masses put our vacuum near, not past, the metastability line

Coleman’s formalism turns “will our vacuum eventually decay” from a philosophical question into a number that depends on measured inputs, chiefly the Higgs self-coupling and the top-quark Yukawa coupling, both run up to very high energy scales using the renormalization group. Degrassi and collaborators performed the first complete next-to-next-to-leading-order analysis of this running in 2012, finding that the Higgs quartic coupling extrapolated to the Planck scale comes out remarkably close to zero, and that absolute stability of the electroweak vacuum all the way up to the Planck scale is excluded at 98 percent confidence for a Higgs mass below about 126 GeV — essentially the measured value ^[11]. A year later, with refined two-loop matching and three-loop running, Buttazzo and collaborators sharpened the verdict: using a Higgs mass of 125.1 ± 0.2 GeV and a pole top-quark mass of 173.3 ± 0.8 GeV, they concluded that metastability, not absolute stability, is now the preferred outcome, at 99.3 percent confidence, while stating candidly that “because of the present experimental uncertainties on the Standard Model parameters (mostly the top quark mass), we cannot conclusively establish the fate of the electroweak vacuum” ^[12]. Their own plotted result is worth describing precisely rather than rounding into a single headline figure: computed as a function of the pole top mass alone, holding the Higgs mass and strong coupling at their central values, the calculated lifetime of the electroweak vacuum swings from of order the current age of the universe near a top mass around 172 GeV to values climbing past $10^{500}$ years and beyond by a top mass of about 174 GeV — a shift of hundreds of orders of magnitude driven by roughly one GeV of uncertainty in a single input, which is precisely why the paper’s own conclusion declines to call the case closed. At the measured central values, they find the probability that vacuum decay has already occurred somewhere in our past light cone to be, in their own words, “spectacularly small,” and describe the vacuum as likely to survive “for times that are enormously longer than any significant astrophysical age (e.g. the sun will exhaust its fuel in about five billion years).”

A later, independent calculation by Andreassen, Frost, and Schwartz resolved technical infrared divergences that arise because the relevant part of the Standard Model potential is close to scale-invariant, an issue the earlier calculations had to sidestep, and arrived at what they describe as the first complete calculation of the lifetime of our universe: $10^{139}$ years ^[13]. That number should not be read as superseding the Buttazzo estimate so much as sitting alongside it as a different, more technically complete calculation under the same broad assumption — that no new physics intervenes between the electroweak scale and the Planck scale to change the running of the couplings — and every one of these papers is explicit that a discovery of new particles at intermediate energy scales could shift the extrapolated couplings and change the verdict in either direction. None of this licenses the periodic press cycle in which the Higgs boson is reported as poised to “destroy the universe.” As John Ellis put it plainly in a 2022 explainer for CERN Courier, “there is no immediate need to panic,” precisely because the calculated lifetime, whatever its exact exponent, so vastly exceeds every other timescale of practical or even astrophysical concern that metastability carries no consequence for anything alive to read about it ^[14]. What is genuinely striking, and stated as such in the technical papers rather than only in commentary, is the coincidence itself: the measured Higgs and top masses place the Standard Model’s high-energy parameters strikingly close to the boundary between stability and metastability, a proximity that has no accepted explanation and that several of the same authors flag as suggestive of some deeper organizing principle without endorsing any specific candidate for what that principle might be.

Testing whether the “constants” the vacuum froze in are actually varying

If the vacuum has a history of discrete jumps, a natural follow-up question is whether it also has a history of gradual drift — whether the constants fixed by electroweak symmetry breaking are quietly creeping away from their frozen values even now. This is a genuinely different physical question from phase transitions, and the Standard Model already contains a mild, well-understood version of scale dependence that should not be confused with time variation: coupling constants “run,” meaning their effective strength depends on the energy scale $Q$ at which they are probed, governed by renormalization-group equations of the schematic form

where the $\beta_i$ are calculable functions of all the couplings together. This is exactly the machinery Degrassi’s and Buttazzo’s teams used to run the Higgs and top couplings up to the Planck scale, and it is not evidence that anything is changing over cosmic time; a coupling’s value at 100 GeV and its value at $10^{19}$ GeV are both fixed, calculable numbers at any moment in cosmic history, connected by a known equation, not two different eras’ worth of physics.

Genuine time variation is a separate, empirical claim, and it has a genuinely contested history. Using large samples of quasar absorption spectra from the Keck and Very Large Telescope observatories, King, Webb, Murphy, Flambaum, and collaborators reported in 2012 that the fine-structure constant appears to vary spatially across the sky, well fit by a dipole model with an amplitude of roughly one part in $10^5$, discrepant from zero at the 4.1-sigma level ^[16]. The claim has not been widely accepted as established; subsequent reanalyses have raised concerns about systematic errors in the spectrograph wavelength calibration and about the statistical significance of the dipole once those systematics are modeled, and the community’s position, surveyed comprehensively in Uzan’s review of varying-constants research, remains that the observational case for a genuine spatial or temporal variation in the fine-structure constant is unconfirmed rather than refuted — a live dispute rather than a settled negative ^[15]. Meanwhile, laboratory tests using frequency comparisons between different species of optical atomic clocks have continued to tighten the constraints on any present-day drift, and recent work has extended this program beyond a simple steady drift to search for oscillating variations that would be the signature of coupling to hypothetical ultralight dark matter, deriving some of the tightest limits to date on that specific class of models even while finding no evidence of variation at all ^[17]. Taken together, the honest state of the field is this: if the vacuum’s constants are drifting today, no experiment has yet detected it convincingly, and the burden has shifted decisively from confirming a detected signal to bounding an undetected one ever more tightly. The vacuum’s history, on the evidence available, looks like a sequence of discrete jumps between phases, followed by constants that are genuinely constant on every timescale precise measurement has yet reached — not a slow evolution but a fossil record with no known process of ongoing revision.

Frozen accidents plus symmetry, not a lineage

Put the whole account together and the picture that emerges is neither the mechanistic determinism that a purely constant-obeying-eternal-law universe might suggest, nor the richly contingent, competitively selected picture that “evolution” imports from biology by default. It is a specific third thing: a small number of discrete transitions, each one a genuine accident of which point on a circle of equivalent possibilities got selected, each one thereafter constrained by symmetries that permit a narrow range of consistent outcomes and forbid the rest, with no observed mechanism for revising the outcome once it has been reached and no observed population of alternatives against which the outcome was ever in competition. The particle spectrum is a fossil in the sense that matters most: it preserves information about a historical event, is not re-derivable from present physics alone, and would very likely have come out differently under different initial conditions.

It is not a fossil in the sense biology needs the word to carry the rest of its explanatory weight. There was no reproduction, no differential survival, and — barring the landscape’s still-unconfirmed multiverse, honestly flagged here as exactly that — no population within which selection could have operated at all. Calling electroweak symmetry breaking a speciation event captures something true and worth saying about contingency layered on constraint; calling it evolution by natural selection claims a mechanism the evidence does not show. The vacuum has a history. Whether that history was chosen from a population or was simply the only outcome a single, non-repeating universe was ever going to produce remains, honestly, an open question — one that current physics can characterize with precision and cannot yet answer.