Kabuki is not background. In Pacific Overtures, it is the argument.

The western is a story about what happens when the map runs out.

Not guns. Not horses. Not Monument Valley at sunset. Those are the costume. The argument underneath is about men who can only exist in unmapped space, and what becomes of them when someone draws the line. Zane Grey knew this. Sam Peckinpah knew it in his bones. The western ends, always, at the same moment: the territory closes, and the men who belonged to it have nowhere left to go.

Stephen Sondheim spent thirty years writing and rewriting Road Show, and in all three versions, under all three titles, he was writing a western. He did not know this, or if he knew it, he never said so. The show is set in Florida swampland, in Alaskan gold rush camps, in the confidence-man economy of early twentieth century America. No horses. No sheriff. No saloon in the conventional sense, though there are plenty of rooms where men drink and make bad promises. What the show has, in every version, is the western’s actual subject: two brothers who can only be themselves at the edge of the known world, and the slow disaster of the world catching up to them.

Addison and Wilson Mizner are not outlaws in the legal sense, though Wilson comes close. They are outlaws in the western’s deeper sense. They do not fit inside structures. Addison dreams in buildings that cannot be built, baroque fantasies that require more land and more capital than any reasonable man would risk. Wilson runs cons the way a gunfighter runs card games, not for the money exactly but because the performance is the point. Both brothers require open territory to be what they are. Florida in the 1920s, before the bust, before the swamp became suburb, is their frontier. It is the last place.

The show’s central relationship carries the western’s central problem. The partnership of two men who need each other and cannot sustain each other is not a Broadway convention. It is Shane and the homesteader. It is Wyatt Earp and Doc Holliday. It is every western that knows the gunfighter cannot stay and the settler cannot follow. Addison wants to build. Wilson wants to move. The tension between those two impulses is what the western has always been about, and Sondheim puts it inside a single family, which makes it worse.

“Brotherly Love” is the show’s most western number and nobody calls it that. Two men, a fire, an argument that cannot be resolved because it is not really an argument. It is a statement of incompatible natures dressed as a quarrel. Wilson is not wrong that Addison needs him. Addison is not wrong that Wilson destroys everything he touches. Sondheim sets the number in a key that keeps threatening to resolve and doesn’t. The harmonic instability is not atmospheric. It is the argument: these two men occupy the same tonal space and cannot share it. The melody passes between them like a disputed deed, each brother completing the other’s phrases in ways that sound like agreement and mean something closer to annexation. In a western, this scene happens on a ridge at dusk with horses, and one of the men rides away. In Road Show it happens in a hotel room in Florida, and one of the men leaves anyway. The genre does not require the landscape. It requires the leaving.

Before the Florida sequence, the show gives us “That Was a Year,” and this is where Sondheim’s western grammar goes deepest. The number is formally a catalog song, a list of things that happened in a single year of the Klondike gold rush. But Sondheim does something strange with the list structure. Each item arrives faster than the last. The musical phrasing compresses. What should feel like accumulation feels instead like subtraction, like time speeding up not because there is more of it but because there is less. The melody never quite repeats; each return is harmonically shifted, slightly higher, slightly more exposed. By the final chorus the song is running. It is not celebrating the year. It is outrunning it.

This is the western’s relationship to time made musical. The frontier does not slow down to be appreciated. It moves, and the men who live in it move with it, and the men who try to hold it still are the settlers, the bankers, the men who arrive after the gold is gone and call the place a town. “That Was a Year” is Wilson’s number in the way that any Peckinpah chase sequence is the outlaws’ number: the form enacts what the character cannot say, which is that the best time is already behind him and getting farther behind him with every bar.

Peckinpah understood this formally in a way that most directors working in the genre did not. His westerns are not elegies, though critics call them that. An elegy mourns at a safe distance. What Peckinpah made was something more uncomfortable: films that put you inside the experience of obsolescence while it was happening, that made you feel the speed of the closing even as you were watching the men refuse to acknowledge it. The Wild Bunch does not mourn Pike Bishop and Dutch Engstrom from outside their story. It runs with them. The cross-cutting in the final sequence, the famous ballet of the ambush, is not choreographed violence. It is choreographed inevitability. Every cut accelerates the clock. By the time Pike turns and walks toward the General’s compound, the audience already knows the mathematics. There is no exit from the territory once the territory is gone. There is only the direction you choose to walk when the walking is the last thing left.

Sondheim’s Florida land boom sequence is where this grammar becomes literal. Addison sells plots of swampland to buyers who will never see them. The buyers are marks, yes, but they are also settlers, people who believe the territory is real and habitable and theirs for the taking. The Mizners are selling the frontier myth itself. They are not just con men. They are the men who always sold the West to people who had never been there, who took the money and moved on before the buyers arrived to find the soil wrong or the water bad or the deed fraudulent. Every land rush in American history was this transaction. Sondheim dramatizes it as musical comedy and never quite lets you laugh without swallowing something sharp.

The bust arrives the way the western’s ending always arrives. Not with violence but with arithmetic. The land is worth what people believe it is worth, and then they stop believing, and the number goes to zero, and the territory that existed in the gap between promise and delivery simply disappears. Addison’s buildings, half-built and grandiose, stand in the Florida scrub like the ruins of a civilization that never quite happened. This is Boot Hill. This is the abandoned mine. This is what the western always leaves behind when the men who made it possible have moved on or died or been broken by the closing of the space that made them possible.

Wilson ends the show dying in a Hollywood hospital, broke and unreformed, still running the con on the nurses. The western knows this ending too. Doc Holliday dying in a bed in Glenwood Springs, reportedly surprised he was going out with his boots off. The gunfighter who survived every shoot-out, outlived by the civilization he helped create, dying indoors, which is the wrong place for him, which was always going to be the place.

Peckinpah put this image at the center of Pat Garrett and Billy the Kid, a film in which the most damning thing that happens to Garrett is not that he kills Billy but that he survives Billy. He lives long enough to become the instrument of the civilization that made Billy necessary.

The killer of the last outlaw is the last outlaw’s executioner and his monument both. Addison Mizner is Pat Garrett. He gets to watch what he built outlast what he loved.

Addison builds Boca Raton. He gets his monument. He gets the thing the western’s builder figure always gets: permanence, respectability, a name on a building. He also gets to watch his brother die in a room that smells like disinfectant and regret. In the western, the man who stays always pays that price. He watches the man who couldn’t stay disappear, one way or another, and he keeps building, because that is what he does, and the buildings stand long after anyone remembers what they cost.

Road Show never found its audience. Critics called it minor Sondheim, a curiosity, a show that kept being rewritten because it never worked. This judgment mistakes the difficulty for failure. The show is hard to watch because the western is hard to watch when it is honest. Not the triumphalist version, not the version where the right man wins and the frontier is tamed and civilization arrives as reward. The version Peckinpah kept making until nobody would fund him anymore. The version where the territory closes and what lived in it was not noble exactly but was alive in a way that the civilization replacing it is not.

The last image Sondheim landed on, after thirty years, was Addison alone with what he built. Wilson is gone. The land boom is gone. Florida is becoming what it would become. The building stands.

In the western, this is called sundown.

In 1921, when Camille Saint-Saëns died at eighty-six, he was the most decorated composer in France and a member of the Institut. He had been famous since before the Franco-Prussian War. He simply refused to stop being famous, or to stop writing, for fifty years after the language of French music had moved past him. Within a decade his reputation had collapsed into a kind of genial footnote: the man who wrote the zoo music, the organ symphony with the big tune. He became a competent academic who lived too long into a century with no use for him. The composers who came after him built that collapse on purpose. Nobody has bothered to take it apart since.

Take the Cello Concerto №1, written in 1872 for Auguste Tolbecque, who had to fight his own conductor to get the premiere onto the program at all. It opens with no preamble: a single stab of A minor from the orchestra. The cello is already moving before the chord finishes ringing, a triplet figure that plunges rather than rises. Most cello concertos make you wait. This one assumes you are already listening. The middle section drops the key up a half-step into a minuet so antique, so thinned by muted strings, that it sounds like a buried memory of the eighteenth century. It surfaces mid-storm. It is the only moment in the piece where the cello stops declaiming and simply sings, briefly, before the opening triplets crash back and drag the music into the finale. When the opening theme returns there, transformed and sped up, Saint-Saëns doesn’t just let it recur. He uses it to license a closing idea that never appeared earlier in the piece at all. The whole concerto plays like one sustained argument whose final clause needed a term nobody saw coming. Shostakovich and Rachmaninoff both called this the greatest of all cello concertos, and orchestras still program it the way they’d program a contractual obligation.

That kind of structural ambition runs through the catalogue that orchestras have decided not to play. Saint-Saëns wrote twelve operas; only Samson et Dalila survives in regular rotation. Even that one gets staged mostly as a vehicle for a mezzo and a tenor rather than as the political and erotic argument it actually is. Henry VIII has a Tudor court scene as dramatically alert as anything in Meyerbeer, whom Saint-Saëns revered and whom history has also misplaced. Two cello sonatas and a clarinet sonata from his final year sit unplayed outside France, waiting on the same kind of attention the First Concerto has never gotten enough of.

Musicians who actually play this music treat its quality as settled. Ask any cellist about the First Concerto and you get something close to reverence. The puzzle is institutional rather than musical. Orchestras, opera houses, and record labels have spent a century treating a composer who wrote at this level of structural intelligence as the author of exactly four pieces.

The conventional answer is that he lost the argument to modernism, and this is true as far as it goes. Debussy’s hostility toward Saint-Saëns was structural to how Debussy wanted to be heard, not incidental to it. To define Pelléas and the late orchestral works as new, French music needed something old to push against. Saint-Saëns was the obvious candidate: still alive, still composing, still drawing audiences. He was prolific, fluent, and untouched by chromatic mystery, which made him a more useful target than Gounod or Massenet. Ravel was gentler in public and devastating in private. The two men had a real friendship early on, and Saint-Saëns’s praise mattered to the young Ravel. But as Ravel’s own harmonic language sharpened, the elder composer’s clarity started to read as a problem rather than a virtue. Add the Boulanger pedagogy that shaped two generations of French composition students. It taught them to think of Franck and the Schola Cantorum lineage as the serious one, the one with German depth grafted onto French elegance. Saint-Saëns ends up positioned as the composer of effects rather than the composer of structure. Underneath that positioning was a fight over who got to define seriousness in French composition after 1900. Saint-Saëns made an unusually convenient foil, surviving so long and writing tonal, formally lucid scores deep into the era of Le Sacre.

This is the real mechanism, and it accounts for most of the damage. But it is not the only thread, and critics writing about Saint-Saëns have mostly stepped around the other one rather than examining it.

Saint-Saëns’s private life generated speculation during his own lifetime and has generated more since, none of it conclusively resolved. He never remarried after his disastrous marriage to a teenager four decades his junior collapsed under the weight of two infant deaths. He lived for years with his mother and was reportedly suicidal after her death. He performed in drag at private parties, doing cruel, expert impersonations of divas. Biographers disagree sharply on what to make of any of it. Stephen Studd’s critical biography argues for an essentially platonic, sexless private life. Other scholars read the pattern differently: the close attachments to younger male pupils, the unhappiness around women, the documented relish for performing femininity. They take this as evidence of a suppressed homosexuality the composer had every legal reason to keep submerged, given an actual criminal statute against sodomy. No diary, no letter, no court record settles it. What is certain is only that the rumor existed in his own time and that it has never gone away.

Whether the rumor was true matters less than what it did to him on paper. A composer who reads, even ambiguously, as sexually unconventional in a culture actively criminalizing that unconventionality gets filed under decorative, not serious. Weight curdles into mere charm. An argument shrinks down to a surface. That filing system was already running on Tchaikovsky in Russia and would run on Britten in England, more violently, decades later. There is no letter proving French critics dismissed Saint-Saëns’s depth because they suspected something about his bed. But the dismissal followed the same pattern this kind of suspicion produces wherever it attaches to a composer. That resemblance is worth naming even where proof isn’t available.

This isn’t a handful of pleasant minor works going unheard. The Fourth Piano Concerto takes a single rising motif, first stated almost shyly in the piano’s left hand, and spends two movements finding out what it can become. It turns up as a chorale, then a scherzo’s punchline, then finally the triumphant material of the close. The same four or five notes do the work that a lesser composer would have needed three new themes to accomplish. Franck did this too, in the Symphony and the Quintet, but Franck’s cyclic returns arrive with incense and gravity. Saint-Saëns’s arrive like a joke landing, the motif popping up somewhere you didn’t expect it while the orchestra seems almost amused that you didn’t see it coming. Henry VIII has a third-act confrontation built on exactly this kind of structural confidence. It is a court scene that any company with a competent baritone and a sense of pacing could stage tomorrow and watch land.

A generation of French composers needed a foil to define itself against. Saint-Saëns, fluent and alive and still writing tonal music into the era of Le Sacre, was simply the one available. That argument doesn’t require the sexuality question to work, and it shouldn’t have to borrow strength from speculation. A culture already primed to file unconventional men under decorative instead of serious may have given that dismissal extra traction. The surviving evidence can’t settle whether it did, and a critic who pretends otherwise is writing fiction with footnotes. What the evidence does settle is right there in the notes. A concerto begins mid-thought because it trusts you to keep up. A finale earns the right to introduce something genuinely new only because everything before it has been so tightly accounted for. Eighty-six years of writing proved that clarity and structural ambition were never opposites. Nobody has gotten around to conducting like they believe it.

There is a sound near the start of Ikarie XB 1: the engine hum of a starship bound for Alpha Centauri, built from a chord slowed and smeared until pitch and noise stop being different categories. Czechoslovakia made this film in 1963, three years before Star Trek, and it sounds less dated than almost anything Hollywood was doing with space at the time. The reason has a name that almost nobody outside the Czech Republic knows: Zdeněk Liška, a composer who spent three decades treating the orchestra, the tape splice, the human choir, and the oscillator as a single undifferentiated toolkit, then used that toolkit to make Eastern Bloc cinema sound like it was arriving from several decades in the future.

Liška worked steadily from the early 1950s until his death in 1983, scoring well over a hundred films. The range is almost absurd to list in one sentence: the village tragicomedy of The Shop on Main Street (1965), the deadpan atrocity of The Cremator (1969), the medieval brutality of Marketa Lazarová (1967), the interstellar optimism of Ikarie XB 1, and later the puppet nightmares of Jan Švankmajer. Few composers anywhere have written convincingly across that range while also functioning as one of the more advanced electroacoustic experimenters of their generation, working quietly inside a national film industry with no particular ideological investment in the avant-garde and no real mechanism for exporting one. He built a body of work that anticipates ambient music, drone, and modern horror scoring by a decade or more, in relative obscurity. Rediscovering him now feels less like nostalgia and more like correcting the record.

Start with Marketa Lazarová. Everyone who has tried to rehabilitate Liška’s reputation starts there, for good reason. František Vláčil’s 1967 film is routinely named the greatest Czech film ever made, a brutal, hallucinatory account of two feuding medieval clans, shot in black-and-white that looks less like cinematography than weather. Liška’s score carries real structural weight in that reputation. He builds the music from plainchant-adjacent vocal writing, voices stacked in close, dissonant clusters that owe more to Penderecki than to anything liturgical, then pulled into slow glissandi that never settle on a pitch long enough to resolve. Against this he layers percussion that lands somewhere between tribal drumming and machine clatter, low drum strikes answered by metallic scraping, as though the thirteenth century and a synthesizer lab had collided in the same room. The effect is devotional music with the certainty bled out of it, sacred in shape and broken in execution. The score has had a second life as a concert and oratorio reconstruction, performed outside the context of the film entirely [Likely], a useful test of whether film music can stand as music, full stop. Liška’s passes that test in a way very little Golden Age scoring does.

Ikarie XB 1, four years earlier, solves the inverse problem with the same toolkit. The medieval sacred needed close vocal dissonance and clattering percussion. The technological sublime needed something colder, and Liška reached for electronic manipulation that Radio Prague has identified as integral to the film’s identity [Likely]. Tones are stretched past their natural decay, filtered until the attack disappears and only a held, breathing drone remains, then layered under or against orchestral writing rather than replacing it outright. There were no synthesizers in the modern sense available to him. This is tape work, filtered and layered orchestral sound, the studio itself treated as an instrument, built alongside the film rather than added on top of a finished picture. The result still sounds less like a prediction than a transmission, a signal caught from somewhere the rest of 1963 hadn’t reached yet.

A small set of operations carries across the comic, the historical, the cosmic, and eventually the surreal: voices stacked into mass rather than melody, rhythm built from non-orchestral or processed sound rather than a conventional rhythm section, harmony that clusters and holds instead of resolving toward a cadence. Liška applied those operations to whatever the image in front of him demanded. The later Švankmajer collaborations stand as a logical endpoint rather than a late-career swerve, because stop-motion animation, with its uncanny near-life movement, needed a sound that refuses to behave like conventional underscoring, something that could sit under a twitching puppet without explaining the twitch away. He had been building toward exactly that since the 1950s.

The obscurity is not a mystery. Czechoslovak cinema operated behind a political and linguistic barrier that limited circulation regardless of quality, and a composer’s international reputation in the twentieth century depended heavily on circulation: touring orchestras, published scores, critics who could actually see the films. Liška had none of the infrastructure that turned contemporaries like Ennio Morricone into global brands. He also worked, for most of his career, under a system that prized productivity over personal mythology. The result is a filmography instead of a legend, no tantrum stories, no genius performed for the press, just an enormous and strange catalogue of work that mostly exists only inside the films it was written for. Specialist labels have begun reissuing his soundtracks. Archivists and ensembles have staged reconstructions like the Marketa Lazarová oratorio. Critics have started revisiting Czechoslovak New Wave cinema as a major and under-translated chapter of Cold War art rather than a regional curiosity. The reassessment is doing, decades late, the work that geography and politics prevented from happening in real time.

Liška’s career complicates a comfortable story about where sonic experimentation happened during the Cold War. The canonical version runs through state institutions: Stockhausen at the electronic music studio in Cologne, Eduard Artemyev building Soviet film and synthesizer work inside official channels, composers with institutional backing, manifestos, and a clear claim on the avant-garde label. Liška had none of that cover. He ran his experiments inside commercial film contracts, under deadline, accountable to directors rather than to any movement, and the results stand up against work produced in research labs with better funding and a clearer ideological mandate. He was unfairly ignored. The deeper problem is that the place everyone agreed to look for radical Cold War sound was incomplete the entire time. The missing piece had been sitting in a Barrandov screening room since 1967, attached to a film about two warring medieval clans, waiting for someone to separate the music from the picture and listen to it on its own.

The Composer Who Made the Cold War Sound Like the Future was originally published in It's Only A Movie on Medium, where people are continuing the conversation by highlighting and responding to this story.

Kabuki is not background. In Pacific Overtures, it is the argument.

Stephen Sondheim and John Weidman built their 1976 musical around a specific formal conceit: Japan’s forced opening by Commodore Perry in 1853 would be told in the theatrical language Japan already possessed. The Kabuki conventions, the mie poses, the hanamichi runway, the stylized violence, the all-male casting, are not decorative. They are the show’s position on what it means to tell this story at all. But audiences, overwhelmingly Western, read those conventions as exotic surface. They take the style for atmosphere. This is the central misreading that has followed the show for fifty years, and it is not incidental to the experience. It is what the show is counting on.

Consider what Kabuki actually does. It is a theatrical form built on the assumption that the audience already knows the story. The pleasure is not revelation but execution, not plot but form. The white-faced actor playing Kayama or Manjiro is not hiding inside the role the way a Method actor hides. He is displaying it. The audience watches the display of a role, not the disappearance into one. That is the entire transaction.

Western theater has largely operated on the opposite premise since Stanislavski. The actor vanishes. The illusion holds. The audience forgets it is watching.

Pacific Overtures stages a collision between those two premises and calls it 1853.

When Perry’s black ships appear, they do so in a show constructed on Kabuki’s logic of display. The Japanese characters perform their crisis openly, aesthetically, in full awareness of the theatrical frame. Perry and his forces are rendered grotesquely, as Western interlopers in a Japanese story. The show is telling its audience, or rather showing them without telling them, that the Western gaze has always been the problem. Perry is not just a historical figure. He is a spectatorial position.

Here is where the third wall appears.

The Western audience watching Pacific Overtures occupies the same seat as Perry. They are watching Japan be watched. They are, in the logic of the show’s structure, the colonizing eye. The Kabuki form is not inviting them in. It is staging their exclusion while letting them believe they are included.

Most critics have not noticed this. The reviews that followed Harold Prince’s original production praised the show’s visual beauty, its unusual ambition, its willingness to engage with non-Western theatrical conventions. This is exactly the response the show is interrogating. The audience calls the Kabuki elements beautiful and unusual. Japan has been aestheticized again. The colonization has been performed again, in the theater, fifty feet from the stage, by the very people applauding the show’s political sophistication.

The number “Please Hello” makes this almost uncomfortably literal. British, Dutch, Russian, French, and American representatives arrive in succession to extract treaties from a Japan that cannot refuse. Each envoy sings in a pastiche of his national musical style: Gilbert and Sullivan for the British, French chanson, American march. The musical idioms are the point. The West does not come to Japan speaking Japanese. It comes speaking its own forms, loudly, insistently, and the show renders this through form itself. Sondheim does not describe cultural imperialism. He enacts it in the music.

The Japanese characters watch this succession with a horror the show gives them no theatrical language to refuse. Kabuki has no idiom for the Gilbert and Sullivan invasion. The form breaks against the content. This is intentional.

But the Western audience laughs. The pastiche is clever. Gilbert and Sullivan is funny. The show is good at this. And in laughing, the audience performs its own position: it recognizes its own musical inheritance and finds it charming, finds its power charming, even as the show is staging that power as catastrophe.

This is the third wall. The audience watches a show about people being watched and dominated by people who look like the audience. The show knows this. The Kabuki frame makes the Western eye visible as a frame. But the effect depends entirely on the Western audience not feeling implicated, not catching it, experiencing the show as a beautiful and somewhat alien theatrical experiment rather than as a mirror.

Sondheim’s one explicit acknowledgment comes in the final number. “Next” drives the story forward through a century of Westernization, the Japanese figures shedding their Kabuki costumes until they stand in business suits, until Japan has consumed the West’s methods and turned them back outward. The original production ended with a modern Tokyo projected overhead, bright and commercial, obliterating the stage. The circle closes. Japan wins a different argument than the one Perry was making.

The audience applauds. The house lights come up.

And here is what has just happened: a Western audience paid money to watch a Japanese theatrical form stage its own violation, laughed at the right moments, admired the craft, found the visual language striking and foreign, and left feeling they had witnessed something politically serious. They had. What they had not done was notice that the show was about them. That the Kabuki frame was not an invitation but an accusation. That the seat they were sitting in was Perry’s seat, and the black ships were their subscription.

Pacific Overtures is the only musical that requires its audience to fail in order to succeed. The beauty is the trap. The applause is the proof.

Opera has survived a lot: bad kings, bad patrons, bad acoustics, director’s-theater concepts involving folding chairs and fluorescent lights, and at least three centuries of people announcing that the art form is dead because some new technology has arrived to ruin civilization.

So I am not especially panicked by artificial intelligence.

Concerned? Yes. Suspicious? Naturally. Morbidly curious? Always. But panicked, no. Opera has always been a machine for combining the unnatural with the deeply human. It is an art form in which people sing their private thoughts at public volume while dying attractively under expensive lighting. It was never exactly realism.

That is why AI’s arrival feels less like an alien invasion than an awkward new comprimario wandering onstage in Act II: everyone knows it is there, nobody is sure what it wants, and it keeps standing in front of the principals. The question is not whether to acknowledge it. The question is whether to give it lines.

The first wave will probably be the least interesting: marketing copy, donor segmentation, ticketing predictions, automated supertitle drafts, rehearsal scheduling, production research, maybe a chatbot that tells a confused patron whether the tenor dies before or after intermission. Fine. If a machine can help an overworked opera company stop sending emails that begin “Dear Valued Patron,” I am prepared to applaud politely.

But giving the comprimario lines is a different matter.

Opera is not a painting, a poem, or even a symphony. A composer imagines sound. A librettist compresses psychology into singable language. A singer turns breath, muscle, diction, memory, fear, vanity, and technique into something that appears effortless only to people who have never tried it. A conductor imposes order on chaos. A director imposes chaos on order. Designers build worlds that must survive both metaphor and union rules. Somewhere in the building, a stage manager quietly prevents civilization from collapsing.

To insert AI into that is not a neutral act. It changes the terms of authorship, labor, taste, and responsibility.

Responsibility is the part nobody in the tech world ever seems eager to sing about.

When an AI system generates a melody, whose musical memory is it borrowing? When it produces a libretto in the style of a living writer, who gave permission? When it clones a voice, even imperfectly, what has been taken? Opera people understand better than most that a voice is not merely output. It is a body, a history, a livelihood, and sometimes a scar. Feeding a dead singer’s recordings into a machine and calling the result a tribute is not homage. It is grave-robbing with a better interface.

The danger is not that AI will suddenly become Maria Callas. It will not. The danger is that institutions under financial pressure will begin treating approximation as adequacy. A synthetic chorus for the demo. A machine-generated orchestration to save rehearsal time. A design concept produced in seconds, polished by humans later. Each step may seem harmless. Together they teach an industry to value the appearance of artistry over the conditions that make artistry possible.

Opera already has enough problems with that.

Dismissing the technology outright is its own trap. It does not make AI disappear. It only guarantees that the people with the least theatrical imagination define how it gets used. Opera should not hand that argument to consultants who use “content” as a noun without embarrassment.

There are things worth trying. AI could help visualize impossible stagings before anyone builds a set. It could give small companies real tools for research, translation, archival restoration. It could help composers test structures and explore unfamiliar materials without anyone pretending the machine deserves a program credit. These are comprimario uses: supporting the action without driving it.

The moment AI starts driving it, the evening is in trouble.

AI should be treated like stage machinery: powerful, useful, and not something you leave alone with the soprano. It should serve a human vision. It should be disclosed when it shapes the work. It should be built on consent. It should expand the room, not shrink the payroll. And it should never be mistaken for what happens when a living person stands in front of other living people and risks failure in real time.

That risk is opera’s soul.

A singer cracking on a high note is more interesting than a flawless synthetic imitation. A director’s bad idea is at least a human bad idea, with fingerprints and delusions and possibly a dramaturgical note nobody asked for. A composer struggling toward something new is more valuable than a machine averaging the past into a plausible present. Opera does not need plausibility. It needs danger.

The best use of AI in opera may be as subject rather than author. Let it appear as a character, a system, an oracle, a parasite, a chorus of false certainties. Opera has always loved the supernatural: gods, ghosts, curses, magic objects, poisoned letters, swans with complicated personal lives. AI fits perfectly into that lineage. It is our era’s enchanted mirror. It shows us ourselves, but smoother, faster, and with all the moral ambiguity airbrushed out.

An opera about AI could be thrilling. An opera by AI sounds, at least for now, like punishment.

The distinction matters. Wagner used stagecraft. Puccini used theatrical realism. Contemporary opera uses video, amplification, projection, electronics, and whatever else can be smuggled into a rehearsal room before someone from finance notices. Technology has always been in the building. The question is never whether it is allowed. The question is whether it knows its place.

Our comprimario is not going home. Fine. Cast it carefully, give it limited stage time, and keep it away from the title role.

Opera still requires a soul for that.

For nine years Petr Popelka heard opera from underneath. As deputy principal double bassist of the Staatskapelle Dresden from 2010 to 2019, he sat at the harmonic floor of the Semperoper pit, where Elektra, Salome, and Der Rosenkavalier first detonated. A bassist never gets the tune. He gets the foundation, the pulse, and a decade of watching conductors succeed or fail from the one chair where every miscalculation lands first.

On Wednesday the Bayerische Staatsoper announced that this bassist, now 40, will become its Generalmusikdirektor with the 2029/30 season, succeeding Vladimir Jurowski. Bavaria’s culture minister Markus Blume called him the orchestra’s wish candidate and described the chemistry between conductor and Staatsorchester as love at first sight. Intendant Serge Dorny praised a rare combination of musical substance and human quality. Strip the ceremony from those statements and one fact remains: the players asked for him.

The Munich post is the most scrutinized music directorship in opera. Kirill Petrenko held it from 2013 to 2020 with single-minded devotion and left for the Berlin Philharmonic as a legend. Jurowski, who arrived in 2021, has split his attention across several major positions. So the question hanging over the announcement is fair: what has this conductor actually done in an opera house?

Opera arrived in the first season of his conducting career. In December 2019, while still a member of the Staatskapelle, he led the Dresden premiere of Peter Eötvös’s Der goldene Drache at Semper Zwei. Staged productions have followed in Dresden, Oslo, Zurich, Vienna, Munich, and Berlin, and the reviews tell one story twice.

The first story is detail. His main-stage Semperoper debut came in July 2022 with Shostakovich’s The Nose in Peter Konwitschny’s staging. Concerti’s critic heard his former colleagues delivering brutally precise interjections, stupendous brass glissandi, and string runs of nearly irrational speed. In January 2023 he conducted a new Elektra in Oslo, with 97 players in a pit that bulged. NRK’s reviewer argued that the orchestra never turned massive or impenetrable. Strauss’s writing is a gigantic mosaic of detail, the critic wrote, and Popelka made every growling bass and trilling piccolo audible. The Scenekunst critic found the musical execution often breathtaking, the intensity never flagging for a second. In Tobias Kratzer’s Schwanda at the Theater an der Wien (November 2023), Die Deutsche Bühne heard shimmering orchestral color from the Wiener Symphoniker, then playing under their incoming chief conductor. And when Popelka took over Krzysztof Warlikowski’s Káťa Kabanová at the 2025 Munich Opera Festival, the French journal Première Loge called the Bayerisches Staatsorchester one of the undisputed stars of the evening. The critic filed the performance under nights one brags about having attended. Within a year, Munich had hired him.

The second story is weight. Concerti magazine’s Dresden notice added that the volume sometimes overshot. The Dagsavisen critic heard the Oslo orchestra, in places, showing more muscle than the soloists could answer. In a Zurich Don Giovanni (autumn 2023), Online Merker found the Philharmonia’s broad, romantic sound only partly reined in, with soloists at risk of being covered at the evening’s edges. In Stefan Herheim’s Fledermaus (October 2025), Bachtrack admired a muscular, swaggering account and found it, at times, “a hair too aggressive for Viennese lightness.” Four cities. Four orchestras. The same marginal note. This is a bassist’s vice. A musician formed at the bottom of the texture builds his sound from the floor up and trusts the singers to climb on top of it. The Káťa notices suggest the lesson is being learned. A Berlin Tosca run at the Staatsoper Unter den Linden closed this February to a long ovation. But a Generalmusikdirektor who buries his Isolde will hear about it by intermission.

The repertoire gaps are real. No Verdi. No staged Wagner production of his own, only revival performances of Tristan in Berlin. One Puccini run. One Mozart revival, and that one drew the balance complaint. Munich’s identity rests on Wagner, Strauss, and Mozart. Popelka has played that repertoire from the inside for a decade. Conducting it across a season, in a 2,101-seat theater, with casts who expect to be heard, is a different examination, and he has barely sat for it.

The institutional gap is larger than the musical one. Petrenko arrived in Munich having held music directorships at two opera houses and a staff conducting post at a third. Popelka has run three symphony orchestras and zero opera houses. A Generalmusikdirektor plans seasons, builds ensembles, arbitrates between stage and pit, and absorbs the blame for both. He also holds the Vienna Symphony post, and Munich has just spent five years learning what a part-time music director costs.

What Munich gets needs no embellishment. With Jakub Hrůša installed at Covent Garden since September, Popelka’s appointment puts Czech conductors atop two of opera’s commanding heights, at the exact moment Janáček has moved from specialty to center repertoire. Munich gets a conductor whom orchestral musicians choose. And it gets three years of runway, beginning with Rusalka at this summer’s festival, a return engagement that the announcement has turned into a victory lap.

The appointment is right, on one condition. Popelka must arrive the way Petrenko did, alone, with the Vienna post resigned and the singer’s half of the bargain learned. The talent is no longer the question. The reviews from Dresden, Oslo, Vienna, and Munich settled that. What remains unsettled is whether a man who rose this fast can hold still long enough for one house.

He spent nine years at the bottom of the orchestra, learning what conductors cost the people who play for them. In 2029, Munich finds out what that education is worth from the other side of the podium.

There is a specific problem with conducting Mozart, and it has nothing to do with the notes. The notes are not hard. The architecture is not obscure. The problem is structural and, for a certain kind of conductor, nearly insoluble. Mozart’s late scores, the da Ponte operas, the final symphonies, The Magic Flute, are written as transparencies. There is no padding, no rhetorical filling, no orchestral weight deployed to impress. By 1791 he had stopped trying to fill the room. He was writing in glass: thin, clear, load-bearing. Press your thumb against it and everyone in the house sees the smudge.

This is what punishes the wrong kind of greatness. A conductor who has built a career on the literature that rewards intervention, Wagner, Bruckner, Mahler, Verdi, arrives at Mozart carrying the wrong tools. The habit of shaping, of making the interpretive presence felt, of ensuring that the audience registers the conductor’s understanding of what this moment means: these are virtues in almost every other repertoire. They are liabilities here. Mozart’s transparency is not a stylistic preference. It is a structural fact. Any imposition shows.

Herbert von Karajan understood this late, if at all. His Wagner was among the most burnished of the postwar era, the Berlin Philharmonic achieving a smoothness that became, in that repertoire, an instrument in itself. In Mozart, the same smoothness is a problem. A late Romantic sound transferred wholesale onto scores that were not built to carry it, a finish applied where none was needed, a layer of gorgeous varnish over writing that required only air. The glass is too thick. You cannot see through it. Karajan’s Mozart, for all its surface beauty, is a story about a great conductor’s temperament encountering a composer whose music had no use for it.

Leonard Bernstein’s case is different and in some ways more interesting. His gift was not smoothness but abundance, an overflow of expressiveness so physical that his podium manner resembled choreography. In Mahler, that abundance was the point; those scores are built to absorb everything a conductor brings and still ask for more. In Mozart, the same instinct becomes crowding. You hear the conductor perpetually in the room, adjusting, commenting, ensuring his understanding is legible. Mozart does not want to be understood. He wants to be heard. The distinction sounds slight. It is the difference between a transparent score and a conducted one.

Daniel Barenboim carries both Karajan’s Romantic inheritance and something of Bernstein’s interpretive ambition, and Mozart catches him in the same trap. The scores get a broad symphonic sweep that flatters the conductor and obscures the writing. The intelligence is present everywhere and helpful nowhere.

Against these examples stands Karl Böhm, who solved the Mozart problem so completely that his solutions now read as obvious. Böhm came to Mozart through Bruno Walter and Richard Strauss, an unusual lineage, given that Strauss’s own orchestral world is as opulent as Mozart’s is spare. What Strauss gave Böhm, apparently, was a sense of the score’s internal logic: the understanding that Mozart’s effects are not effects at all but consequences, that the wind writing sits where it sits because the harmony requires it, that the clarity is not a surface quality but a structural one. Böhm’s Magic Flute recordings with the Berlin Philharmonic, made in 1964, are still benchmarks. The pacing is unerring, the orchestral textures clean without being cold, the drama unfolding as if the score were generating it rather than a conductor pressing it forward. Nothing is underlined. Nothing is explained. The music arrives without mediation, and the conductor is conspicuously, productively absent.

René Jacobs, the Belgian conductor who came to the podium through early music and through singing, has spent thirty years demonstrating what the historically informed movement actually offers Mozart beyond the timbre of gut strings. His da Ponte recordings with the Freiburger Barockorchester and Concerto Köln treat Mozart’s comic operas not as graceful confections but as precise emotional machines, and the job of the conductor is to keep the mechanism running at its own speed. He trusts the score’s own drama without announcing it, lets the recitative move because it was written to move, and does not pause over moments of beauty to ensure the audience has noticed them. The beauty is there. It does not need pointing.

Yannick Nézet-Séguin, one of the most capable conductors before the public today, runs into Mozart differently. His Mozart opera cycle for Deutsche Grammophon with the Chamber Orchestra of Europe is flowing, considered, clean. It is also not life-altering Mozart. His phrasing is intelligent, his tempos judicious, and the scores are being played rather than inhabited. The difference is audible in the way a phrase lands: in a conducted Mozart, you hear the conductor’s decision. In a transparent Mozart, you hear the phrase deciding for itself.

That absence is the hardest thing to achieve. Getting out of the way is not passive. It requires trust in the score at a level most conductors cannot sustain, trust that the writing will carry the evening without reinforcement, that the Masonic chords at the opening of The Magic Flute will land with solemnity because the harmony is solemn, not because a baton has instructed the orchestra to feel that way. The solo flute becomes a character because Mozart wrote it as one. The wind writing rises above the strings because that is where Mozart put it. A conductor who intervenes in any of this, however skillfully, has misunderstood the assignment.

There is a particular kind of conducting intelligence that Mozart requires and that almost nothing else in the standard repertoire develops. It is the intelligence of restraint, which is rarer than the intelligence of expression and harder to recognize from the house. The conductor who disappears behind Mozart has not failed. He has done the one thing this composer asks of the person standing at the podium: he has trusted that the transparencies will hold.

There is a particular sound a press release makes when it lands in a review unedited. It is the sound of a critic who went to a concert and came home with nothing. Los Angeles produces this sound at industrial volume.

You can hear it in a recent notice of Dudamel’s farewell Walküre (the critic’s name has been removed). Six hours of Wagner, two complete cycles, a Frank Gehry set conceived months before the architect died, and the orchestra, the entire reason the thing exists, gets two sentences. Dudamel “presided with commanding calm.” The woodwinds were “heartbreaking.” Calm. As if Wagner were a man being talked down off a ledge. Nothing about how the Act I prelude’s storm was paced, whether the brass chorales breathed or barked, what Dudamel’s famously elastic rubato did to those endless melodic paragraphs. The score, which is an argument the composer is making, gets treated as a fact the critic is confirming. Present, accounted for, lush.

This is not criticism. It is weather reporting. And in a city with one full-time classical critic of any real stature and a freelance bench that thins every season, the weather report has become the house style.

The problem is not a shortage of events. The problem is that the writing has stopped keeping pace with the playing. An orchestra rehearses a Bruckner symphony for a week. The critic gets it once, on deadline, and files eight hundred words that could have been written from the parking structure without going inside.

The adjectives are the tell. “Luminous” is what you write when you did not listen closely enough to say anything specific, and it has done more quiet damage to American music criticism than any modernist ever managed. It sounds like praise and means nothing. A soprano is not luminous. A soprano sang under the pitch in her first aria and found the center by her second, or she didn’t, and the critic who cannot tell you which one was not really listening.

Then there is the house-critic flinch, the LA inability to say a performance was bad. Everything is, at worst, “uneven.” An ambitious disaster gets called ambitious and the disaster goes unmentioned. A bored orchestra phoning in Dvorak becomes “a thoughtful, restrained reading.” The reader paid actual money. The reader can hear perfectly well that the brass blared and there was no Czech lilt anywhere in the strings. That reader is being gently lied to by the one person in the building whose entire job is to not do that. Criticism without the nerve for a verdict is publicity with better grammar.

What is missing underneath all of it is an idea. The strongest reviews are built around a thought that opens the piece and is still standing at the end. Pick an LA classical notice at random and go looking for the thought. It opens on the conductor’s biography. It closes on the curtain call and the roars of approval. Strip out the proper nouns and nothing remains, because nothing was holding the piece together except the chronological accident that the concert started and later it stopped.

This is fixable, which is the part that should embarrass everyone. It asks the critic to do the single thing the job is for: to have heard the concert, to hold a position about it, and to defend that position in sentences specific enough to be wrong. “The oboe was heartbreaking” has said nothing. Name the phrase. Name what the score wanted and whether the player gave it. That is the work, and it is not a mystery, and it is not even particularly hard.

Los Angeles has the orchestras and the halls and the once-in-a-generation conductors heading for the exits with their best work behind glass. What it lacks, most nights, is anyone in the press seats listening as hard as the musicians on stage are playing. One of the great Wagnerian evenings this city has staged produced, in print, “commanding calm.” The musicians earned a better witness. So did everyone who read it the next morning.

Daniel Harding holds a commercial pilot’s license and flies for Air France. On Tuesday the Los Angeles Philharmonic asked him to land something heavier. He becomes Music Director in the 2027/28 season, succeeding Gustavo Dudamel, and the detail about the cockpit is not a curiosity to set aside. It is the key to the appointment. Harding is a man who has spent a career splitting himself between the podium and the flight deck, and Los Angeles has just bet that the half it gets is enough.

Consider what Dudamel made of the job. Seventeen years, and the orchestra became a face. People who could not name a Mahler symphony knew the hair, the grin, the body that seemed to push the sound uphill. The LA Phil sold a personality. Harding sells nothing of the kind, and the search committee, which included the musicians, knew that when they made him their overwhelming choice. They did not pick a successor. They picked a correction.

What they get is one of the finest ears in the profession. Harding came up inside the Mahler Chamber Orchestra, which he helped found and shaped for two decades, and a chamber musician hears differently than a podium star. He hears the second oboe. His recorded Mahler proves it. In the Fifth with the Swedish Radio Symphony he scrapes away the big crescendos that a century of conductors layered onto the opening bars, crescendos Mahler never wrote, and what surfaces instead is the inner lines, suddenly load-bearing. The Scherzo turns funny in a darker key.

Listen to how he handles the same symphony’s Adagietto, the movement everyone slows toward sanctity. Harding keeps it moving. He treats the harp and the strings as a conversation rather than a shrine, and the music stops being a funeral and becomes what Mahler more likely meant, a love letter, private and a little embarrassed by its own warmth. That is the Harding signature. He distrusts the inherited gesture. He goes back to the page and asks what is actually printed there, and the answer is usually quieter and stranger than the tradition that grew over it. Walt Disney Concert Hall, a room that hides nothing, will carry every choice of it to the back wall.

That precise ear is also the problem. Harding controls. The discipline that clarifies a texture can also deaden it, and his Mahler Tenth drew the charge of being smooth where the music should bleed. The charge landed because it was fair. Dudamel gave this city heat for seventeen years. Harding offers light. A listener can love both and still feel the temperature drop, and not every subscriber will follow a conductor out of the warm room into the bright one.

Then there is the question the cockpit raises. Harding walked away from the Orchestre de Paris after two seasons. He told the orchestra to its face that it expressed a musical culture not his own. Whatever courtesy that line was meant to carry, it describes a man who leaves when the fit is wrong rather than staying to force it. He has divided recent years between the baton and the flight deck by open choice. The LA Phil contract asks for eight weeks in the first season and twelve after. That is a real commitment, but a fraction of Dudamel’s near-total fusion with the institution. The orchestra has hired an artist who guards his distances.

The repertoire instinct cuts the other way. Harding was the teenager who drafted his schoolmates into thorny modern scores rather than Haydn. His conducting next season folds Adès and Betsy Jolas in beside Brahms and Bernstein. On new music, at least, he intends to keep the LA Phil exactly what it already is. The orchestra that commissions more than it inherits has found a conductor who came up wanting the same thing.

So the city has traded charisma for craft, and the trade works only if the city lets the craft carry the whole evening. Dudamel filled the hall with his body. Harding will have to fill it with the sound, and the sound he makes is lucid and exacting and cool to the touch. He flew jets to feel the part of him that conducting could not reach. The orchestra believes the rest of him is enough. The answer will not be audible until the autumn of 2027, and it will arrive one quiet, unglamorous performance at a time.

Near the end of the first act of “Company,” a couple named Harry and Sarah perform a karate demonstration for their guest Bobby, each landing blows with a precision that slides almost imperceptibly from exhibition into aggression, until it becomes impossible to say whether they are showing off or fighting. The scene lasts only a few minutes, and Sondheim scores it lightly, but its implication runs through everything he wrote. The audience agrees, for the duration of an evening, to believe that the people in front of them are other people. What Sondheim understood, and what separates his work from most of the Broadway tradition he inherited, is that this agreement does not stop at the footlights. His characters are performing for one another. They are, in many cases, performing for themselves. The stage does not merely represent the social world; it enacts the mechanism by which that world operates. Everyone is always already in costume.

This perception gives his music its particular quality of unease. The scores move with a structural intelligence that invites formal admiration and then, almost without warning, use that very intelligence to expose something raw and unwilling. Conversations overlap. In “Company,” fragments of melody pass between singers like dialogue cues, most strikingly in “Side by Side by Side,” where the couples finish one another’s phrases with a togetherness so rehearsed it has the feel of a stage trick: the harmonies lock with a little too much confidence, the rhythms snap into place a beat before they should, and the cumulative effect is of people who have practiced being happy rather than achieved it. The music catches the seams of these presentations. It is scored like a crowded rehearsal room in which everyone is trying out a part and no one is quite sure of the lines.

Bobby himself occupies the position of the reluctant audience member who has wandered onstage. His refusal to commit to marriage can be read as a refusal to accept the role his friends have scripted for him, and yet that refusal is itself a kind of performance, a carefully maintained bachelorhood that holds genuine exposure at arm’s length. When “Being Alive” finally arrives, it is tempting to hear it as a conversion, as the moment the man behind the mask steps forward. But the lyric does not promise authenticity. It promises difficulty. What Bobby wishes for is an experience that will “hurt too much to stand,” a demand less for love than for a role demanding enough to be real. Identity, Sondheim suggests, is not what we discover beneath the performance. It is what the performance, done honestly, finally becomes.

Consider what happens to Ben in “Follies” when his carefully maintained persona begins to crack. He has been singing, or rather reciting in the show’s peculiar semi-melodic way, fragments of a self-narrative so well practiced that it has lost all internal pressure, until the second act strips him down to a ragtime breakdown in which the phrase “live, laugh, love” repeats itself against an increasingly frantic accompaniment, the words cycling past meaning into pure sound, like a record skipping. The sequence earns its horror because Sondheim has spent the first act showing us exactly how much psychic labor went into assembling the persona that is now disintegrating. The setting helps: a crumbling Broadway house awaiting demolition, where past and present coexist without resolving into each other. The characters carry their younger selves behind them like understudies waiting in the wings. Sally remembers herself as the romantic heroine whose love for Ben was the true story of her life. Phyllis maintains her sophistication the way a trained performer maintains composure under lights, the technique impeccable, which is exactly what makes it frightening.

The borrowed musical forms that structure the show’s second act, the torch song, the cakewalk, the soft-shoe, first appear as vessels of glamour and then reveal themselves as distorting mirrors. The point is not that these people have been living falsely but that the line between living and performing was never as clear as they believed. The showgirls of the past were literally theatrical; the supposedly ordinary people they became are no less so. Marriage, nostalgia, achievement each supply a script that promises coherence, and each script, under sufficient pressure, begins to show its stitching.

Listen to what Sondheim does with Charles Guiteau in “Assassins”: he gives the man who murdered James Garfield a jaunty cakewalk, “The Ballad of Guiteau,” in which the condemned prisoner two-steps his way to the gallows while singing of his inevitable elevation to heaven and historical fame. The tune is irresistible in the way that certain pop songs are irresistible, the kind of melody you find yourself humming in the car before you have registered what the words are actually saying. This is the show’s central compositional gambit: each assassin receives a historically appropriate musical idiom, vaudeville, folk, gospel, and those idioms are seductive in direct proportion to the horror of what they are packaging. The carnivalesque frame might suggest a satirist’s cool distance, but the music refuses that distance. It insists on a sympathy that is genuinely uncomfortable. These are people for whom the conventional scripts of American life have failed to provide a legible role, who have turned to public violence as a kind of audition for national significance. The impulse behind the most extreme acts in the show is not so different from the impulse behind the dinner-party performances in “Company.” Everyone wants to be seen. Sondheim’s particular darkness lies in his refusal to treat that want as trivial.

When the opening number of “Into the Woods” passes its melody from Cinderella to Jack to the Baker and his wife, each character announcing a wish in the same crisp rhythmic packet, the effect is of a very well-oiled machine, everyone slotting into position, the counterpoint locking with the self-satisfaction of a form that has been doing this for centuries and sees no reason to stop. The pleasure is real; so is, in retrospect, the unease. You are watching people accept their narrative assignments with a completeness that leaves no remainder, no self that exists outside the story, and the music is so confident in containing them that the confidence itself becomes a kind of dramatic irony. Fairy-tale characters are, by definition, people defined entirely by the roles assigned to them, and the first act honors that contract with genuine warmth. It is only when the wishes come true and the stories end that Sondheim and James Lapine begin their real inquiry: what are human beings when the cultural machinery that ordinarily shapes them gives out? The second act’s harmonic language answers the question in musical terms. Where the first act resolved its cadences with the satisfying click of a closing door, the later music wanders through suspensions that decline to settle, ensembles that layer competing voices without bringing them into agreement, a tonal restlessness that sounds like people searching for a form capable of holding lives that the inherited scripts no longer describe.

In “Sunday in the Park with George,” the opening number, “Sunday,” builds from a single sustained note in the strings, adding voices and instrumental colors the way a pointillist adds dots, each element discrete, the larger image emerging only as the parts accumulate. The compositional method mirrors Seurat’s, and the mirroring is the argument. Seurat is shown as a man who relates to other people primarily as compositional elements, studying posture and distance and color rather than engaging the emotional reality of the person before him. The accusation the show stages is one that any serious artist will recognize: that making things requires a kind of deliberate withdrawal from the ordinary circuits of feeling, and that the withdrawal has a cost. The cost appears in the lyric voice Sondheim gives Dot, a voice of radiant directness that keeps breaking through the painter’s careful formal arrangements, insisting on the primacy of what the compositional eye has converted into material. “George, I am not your painting,” she sings, and the line lands with such simplicity after so much musical elaboration that it feels like a window opening in a closed room.

By the second act, set a century later, the pressure has reversed. The modern George moves through an exhibition of his own work, a series of installations called Chromolume, and delivers a speech of such polished self-explanation to a gallery audience that the words begin to feel like a performance of artistic seriousness rather than an instance of it. At one point he simply runs out of language and stands there, the patter exhausted, the room waiting. The moment is a few seconds long and almost unbearably uncomfortable, not because George has failed but because his fluency had been covering for an absence that the silence makes audible. It is the anxiety that runs through all of Sondheim’s theater, the fear that the role has replaced the person, that the performance has become the only thing there is.

There is a moment late in “Follies” when Sally, having built her entire second act around the fantasy that Ben still loves her, finally hears him say that he does not, and her face does something the script cannot quite describe. She does not collapse. She reorganizes. Within a few lines she has assembled a new version of the story, one in which the rejection confirms rather than destroys the romance, because at least it proves the feeling was real. The gesture is painful to watch and also, in its way, admirable. The impulse that produces it is not deception. It is the same impulse that makes people fall in love, join causes, dedicate themselves to art: the need to live inside a narrative that makes the self feel legible. Sondheim never mocks that need. He is interested in what happens when the narrative calcifies, when a performance that was once a way of reaching toward something real becomes the only reality a person can afford to acknowledge.

His music enacts the alternative. The scores do not deliver single, fixed emotions. They wander through shifting registers, hold contradictions in suspension, allow a melody to begin in confidence and arrive at doubt. In the great ensemble passages, “Someone Is Waiting,” “No One Is Alone,” the finale of “Merrily We Roll Along,” voices layer competing perspectives without forcing resolution, the harmonies reaching toward consonance and then pulling back, less a refusal of feeling than an acknowledgment that feeling is more complicated than any single voice can carry. The stage lights come down, the roles disperse, and the self, like a song held in repertory, goes on being performed, never quite the same twice.