Unmasking pseudonymous public figuresCopy wiki link

Banksy is a pseudonymous street artist whose works are, in the jurisdiction where most are made, usually a criminal offence: applying paint to a building without the owner’s consent is criminal damage under English law, carrying a maximum sentence of ten years where the damage is valuable enough, with listed-building status able to lengthen a sentence further.¹ The artist has operated under that legal exposure for roughly three decades while remaining publicly unidentified, and anonymity has been treated by both the artist and much of the public as inseparable from the work: it lets pieces appear in cities worldwide without pre-emptive removal, prosecution, or the artist’s arrest, and it lets the art be read on its own terms rather than through a named author’s biography.²

In March 2026, Reuters published “In Search of Banksy,” a year-long investigation naming Banksy as Robin Gunningham, a Bristol-connected individual who had reportedly later changed his legal name to David Jones. The claimed evidence included immigration records placing Gunningham in Ukraine at the time murals attributed to Banksy appeared there in 2022, photographs from former associates, and a handwritten confession signed “Robin Gunningham” from a 2000 disorderly-conduct arrest in New York for defacing a billboard.³ Banksy’s lawyer, Mark Stephens, responded that Banksy “does not accept that many of the details contained within [the] enquiry are correct,” without specifying which details were disputed.⁴ A name matching this account had first been floated by the Mail on Sunday in 2008; Reuters presented its 2026 report as resolving the question “beyond dispute.”⁴

The stated public-interest justificationCopy wiki link

Reuters defended publication on the ground that Banksy is a public figure “subject to scrutiny, accountability, and, sometimes, unmasking,” and that “the public has a deep interest in understanding the identity and career of a figure with his profound and enduring influence on culture, the art industry and international political discourse.”⁵ This is a claim about why a reader should be told, not a claim that the underlying facts were hard to verify or that verification itself required identification — the investigation’s difficulty was evidentiary, not definitional. Reporter Simon Gardner made the strongest available version of this case directly: Banksy is “one of the world’s most famous artists, and he’s a public figure who helped shape political, social discourse,” so “the public has a legitimate interest in understanding how cultural figures operate” — “issues of power, access and accountability.”⁶ Gardner also stated that Reuters “deliberately withheld a wealth of information to prevent anybody from being able to actually find him,” naming neither the region of the UK where he lives nor further identifying detail beyond the legal name itself, specifically to let him “go out later today and go and paint just as he normally does.”⁶ This is a genuine attempt to draw a line between establishing a fact for the historical record and supplying an operational location, though it does not answer the objection below about what the disclosed fact itself accomplishes.

Critics of the decision to publish questioned what naming actually accomplishes. Art historian Peter Bengtsen, who had raised the identical objection against the 2016 geographic-profiling study a decade earlier,⁷ asked how identifying Banksy “somehow increases his transparency or accountability”⁵ when the entity actually subject to scrutiny — the work, its market, its authentication, and the commercial infrastructure around it — was already visible without a legal name attached to it. Bengtsen argued that Banksy’s cultural influence was “already well documented” before the unmasking, so naming him added no comparable public benefit while foreclosing a distinct one: anonymity had let the public project their own ideas onto both artist and artwork, including open speculation about whether Banksy might be a woman, which Bengtsen linked to broadening the appeal of street art itself beyond people who share Banksy’s presumed background.⁵ A related objection generalized the argument: if enduring cultural influence is enough to override a person’s chosen anonymity, the same logic reaches investigative sources and whistleblowers whose influence on public discourse can be just as large, and whose anonymity newsrooms otherwise treat as worth protecting rather than worth defeating.⁵

A legal framework for the objection: Article 8, Article 10, and the Berne ConventionCopy wiki link

Legal scholarship developed this critique into a formal argument rather than a normative preference. A Völkerrechtsblog analysis applied the European Court of Human Rights’s proportionality test for balancing press freedom against private life — developed in Von Hannover v Germany (No. 2) and Axel Springer AG v Germany — and concluded that the balance runs against Reuters.⁸ The argument holds two things at once. First, under Article 8, S and Marper v UK treats social identity, including a person’s name, as part of protected private life, and Banksy’s actual identity had never been treated as a matter of public concern by prosecutors, collectors, collaborators, or fans — only the artistic persona had.⁸ Second, under Article 10, case law including Müller v Switzerland and Mariya Alekhina v Russia (the Pussy Riot case) protects not only artistic output but the conditions that make its expression possible, and anonymity is not incidental to Banksy’s practice but constitutive of it: it determines who he can address, where, and how.⁸ Because Banksy’s interest draws on both articles at once while Reuters’s interest in publishing rests on Article 10 alone, the analysis concludes that the UK, as Banksy’s state of nationality and Reuters’s headquarters jurisdiction, bears a positive obligation to protect his identity that current law does not fulfil.⁸ The ECHR binds states rather than private publishers directly, so this is an argument about a regulatory gap, not a claim that Reuters committed an actionable wrong.

The analysis adds a legal dimension specific to timing. Banksy’s September 2025 Royal Courts of Justice mural responded to the arrest of 890 protesters after the UK government proscribed Palestine Action under the Terrorism Act 2000, whose section 12 makes supporting a proscribed organisation punishable by up to fourteen years’ imprisonment.⁸ At the time the mural appeared, whether the proscription itself was compatible with Articles 10 and 11 ECHR was still contested, and the English High Court later rejected it.⁸ An artwork widely read as endorsing Palestine Action’s cause, made without anonymity, would have constituted evidence usable in a prosecution carrying a much longer sentence than ordinary criminal damage — the specific chilling effect the ECtHR identified in Halet v Luxembourg as arising from compelled identification.⁸ This means the March 2026 unmasking did not simply resolve a decades-old celebrity mystery; it named a person against the backdrop of a live, contested counter-terrorism prosecution risk tied to a specific recent artwork.

A separate legal tradition, outside human-rights doctrine, reaches a similar starting point from a different direction. Article 7(3) of the Berne Convention for the Protection of Literary and Artistic Works protects the right of an author to publish anonymous or pseudonymous works, and the UK and EU are both bound by its substantive provisions.⁹ Banksy invoked this indirectly rather than directly: because pursuing copyright infringement claims would require him to prove ownership under his legal name, he instead built his intellectual-property defence around trademarks held through his company, Pest Control, which carries no equivalent identification requirement.⁹ Commentary on the resulting Full Colour Black v Pest Control dispute argued that courts should read Banksy’s trademark strategy as a response to a gap in the law — anonymous authors having no ordinary route to protect their work under their chosen name — rather than as evidence of bad faith.⁹ The same structural problem recurs across every dispute this note covers: legal and journalistic systems are generally built around the assumption that a subject can be named, and a pseudonymous actor’s options narrow sharply wherever that assumption is load-bearing.

The wiki’s existing framework for this question, developed for Swedish court-reporting and people-search disputes, separates journalistic purpose from journalistic form. Journalistic purposes under Article 85 GDPR, established in Legal Newsdesk Sweden C-199-24, holds that neither payment nor the sensitivity of the subject matter by itself excludes journalism, but that journalism requires a purpose of informing the public, an editorial policy, factual verification, and compliance with journalistic ethical practice — not merely that a story is true and a publisher calls it news. Naming convicted people in Swedish media applies the same separation to identification specifically: truth is a defence to defamation, but defensibility is a distinct question from whether the underlying facts are accurate, and it turns on what identification is for. The Banksy case sits outside Swedish law entirely, but the same distinction applies: establishing a name correctly is not the same operation as establishing that publishing the name serves the public rather than merely satisfying it.

What anonymity was doing before it was brokenCopy wiki link

Several practical functions of the anonymity were named in commentary on the unmasking, independent of whether any reader found them persuasive:

• It let the artist keep making work that is a prosecutable offence in most jurisdictions where it appears, without each new piece triggering an identifiable target for police, property owners, or civil claimants.¹
• It protected the artist personally — previous misidentifications had already exposed people wrongly linked to the Banksy name to public attention and harassment they had not sought.² § Collateral harm to people who are not Banksy develops this cost in its own right.
• It shaped how the work was received. Without a named author, audiences engaged the pieces directly rather than filtering them through a biography, and one commentator argued the mystery itself had become part of the art’s cultural function, broadening engagement beyond any single piece.²

A minority response inverted the harm framing entirely, arguing that anonymity had become a cage rather than a shield: frozen at a fixed public persona, Banksy could not evolve stylistically without inviting doubt about authenticity, and being named could let the artist reclaim ordinary authorship and change.² This is a claim about artistic consequence, not a rebuttal to the privacy or safety objections, and the two arguments answer different questions: whether unmasking was good for the art is separate from whether it was justified as a matter of the subject’s own interest in staying unidentified.

Collateral harm to people who are not BanksyCopy wiki link

An unmasking attempt does not only affect its actual target. It creates a pool of named suspects, each exposed before the investigation resolves, and there is no cost to the publisher or the public for naming someone incorrectly along the way. That risk does not end once a name is settled; the rush to illustrate a settled story can generate a fresh misidentification of its own. Banksy speculation has run for close to three decades and has repeatedly settled, wrongly, on real people who then had to answer for a claim they never made.

Robert Del Naja of the band Massive Attack was one recurring suspect, proposed because he had painted graffiti in Bristol under the name 3D before Banksy’s emergence and because Massive Attack tour dates were said to coincide with the appearance of new Banksy pieces in the same cities. Del Naja denied it publicly: “Rumors of my secret identity are greatly exaggerated… It would be a good story but sadly not true. Wishful thinking, I think.”¹⁰ The 2026 Reuters investigation later placed him not as Banksy but as an alleged collaborator, which does not undo the years in which he was the leading public guess and had to keep denying it.

Neil Buchanan, a television presenter best known for hosting the children’s art programme Art Attack, was pulled into the same speculation in 2020 after social-media users noted that Banksy pieces had appeared in cities where Buchanan’s former band had played. The rumour spread widely enough that Buchanan’s own website published a formal denial: “Neil Buchanan is NOT Banksy,” stating there was “no truth in the rumour whatsoever.”¹¹ Nothing about Buchanan’s actual conduct gave rise to the claim beyond a coincidence of geography and timing, and he had no means of pre-empting it — only of responding after it had already spread.

Robin Gunningham himself, before the 2026 Reuters report, had already been named as Banksy by the Mail on Sunday in 2008. Reuters reports that he subsequently changed his legal name to David Jones.⁴ A legal name change following unwanted identification is itself evidence of what exposure cost him, independent of whether the identification was accurate — and the 2026 investigation suggests that it was, which means the earlier 2008 exposure imposed most of the disclosure’s personal cost on him eighteen years before Reuters called the question settled.

A separate, independent 2016 attempt to identify Banksy through data analysis shows that the misidentification risk was recognized even by researchers with no journalistic stake in publishing a name. Queen Mary University of London researchers applied geographic profiling, a technique developed for locating serial offenders from the spatial pattern of their crimes, to 140 Banksy artwork locations in Bristol and London, and the resulting model pointed to the same person the Mail on Sunday had named in 2008.¹² Before publication, Banksy’s legal team intervened to delay the study over concerns that the analysis constituted a privacy breach against the named individual, whether or not the underlying identification was correct.¹² The study was published regardless, in the Journal of Spatial Science. Bengtsen’s contemporaneous academic critique went beyond the general privacy objection to a methodological one: the geographic-profiling model ignored the timing of when each artwork was created relative to the artist’s movements, concentrated on a single named candidate rather than testing alternatives, and, in Bengtsen’s assessment, appeared to name a specific individual primarily to attract media attention to an otherwise niche academic study rather than to serve any identifiable scholarly purpose.⁷ The delay and the critique together show that even a purely statistical exercise, disconnected from tabloid incentives, faced the same objection from two directions at once: naming a specific person carries a cost that accuracy alone does not settle, and in this case the accuracy itself was also contested.

The clearest case of collateral harm arrived only after the 2026 Reuters story broke, and it did not touch Gunningham at all. Multiple outlets, searching their photo archives for an image to run alongside the newly published name, used a photograph of a 69-year-old maintenance man, George Georgiou, who had been photographed years earlier supervising the installation of a protective sheet over a Banksy mural on a building his son owned. Georgiou bears a resemblance to the 2008 Jamaica photograph that tabloids had already circulated as a purported image of Banksy, and outlets ran his picture as though it depicted the unmasked artist.¹³ He began receiving as many as thirty phone calls a day and was accosted by strangers in public asking for autographs or demanding explanations for his art. “I’m not Banksy,” he told Reuters directly. “I’m a maintenance man. It’s an old joke now. I’d like people to get on with it now and leave me alone.”¹³ Reuters had to publish a separate fact-check correcting the other outlets’ claims about its own investigation. Georgiou’s case differs from Del Naja’s and Buchanan’s in one important respect: it was not produced by open-ended public speculation but by the publishing industry’s own demand for an accompanying image once a name existed to attach one to, which means the unmasking did not resolve the collateral-identification risk. It relocated it onto a new, unconnected person.

These cases separate two distinct harms that a purely evidentiary framing of “is the claim true” tends to collapse into one. A false identification harms an innocent person who did nothing to invite scrutiny and has no lawful claim tying them to the disputed conduct at all. A true identification, reached and published without the subject’s consent, can still harm the person correctly identified, particularly where the underlying conduct carries genuine legal exposure. Banksy’s lawyer invoked the second harm directly, writing that the artist had “been subjected to fixated, threatening and extremist behaviour” for years and warning that publication “would violate the artist’s privacy, interfere with his art and put him in danger.”¹⁴ An investigation’s confidence in its own accuracy answers only the first harm. It says nothing about the second, and a publisher weighing “is this true” against “should we publish it” is answering two different questions even when it presents the decision as one.

Legal jeopardy as a second unmasking routeCopy wiki link

Journalistic investigation was not the only pressure on Banksy’s anonymity. A 2025 mural depicting a judge, placed on London’s Royal Courts of Justice, led the Metropolitan Police to open a criminal-damage investigation, and commentary at the time noted plainly that a completed prosecution would force disclosure: Banksy could shield identity from journalists indefinitely, using copyright and defamation proceedings that do not require appearing under a legal name, but a criminal trial does not offer the same shelter.¹ Civil IP litigation had let the artist assert rights without becoming a named party in the ordinary sense; a criminal charge removes that option. This makes the artist’s anonymity conditional on prosecutorial choices made by an institution with its own reasons for restraint or action, not solely on the discretion of the press.

That two independent paths — investigative journalism claiming a public-interest warrant, and criminal process with no such warrant requirement at all — converge on the same outcome illustrates a general point about pseudonymity that does not depend on resolving the Banksy case specifically. A person doing something the state defines as unlawful, however broadly the public may sympathize with the conduct, holds anonymity that is contingent rather than secure, and the actor best positioned to force disclosure is often not the one making a public-interest argument at all.

A legal-blogging comparator: Scott Alexander and the New York TimesCopy wiki link

Banksy’s anonymity is inseparable from an underlying activity that is itself a crime, which makes legal jeopardy a second, independent unmasking route in a way that does not generalize to every pseudonymous case. Scott Alexander, the pseudonym of a Bay Area psychiatrist who wrote the blog Slate Star Codex, shows the same publisher-versus-subject structure with no illegal underlying conduct at all — only ordinary blogging about psychiatry, rationality, and policy.¹⁵ The comparison isolates what the Banksy dispute’s public-interest question looks like once legal exposure is no longer doing any of the work.

In 2020, a New York Times technology reporter told Alexander he intended to include Alexander’s full surname in an upcoming profile of the blog and its readership, and that this was Times policy for identifying subjects.¹⁵ Alexander objected on three specific grounds: patients could locate his personal writing and it would compromise the clinical relationship; he had already received death threats and seen another blogger swatted after being identified; and his employer might terminate him if he became a subject of national news coverage, disrupting care for the patients he already treated.¹⁵ The reporter offered no accommodation. Rather than let the paper reveal the name on its own timetable, Alexander deleted the blog outright in June 2020 and published a public statement explaining that the Times was “threatening my safety by revealing my real name.”¹⁵

Critics of the Times decision noted that the paper had accommodated pseudonymity before, in a 2020 profile of the podcast Chapo Trap House that referred to one host only by his pseudonym, Virgil Texas, and extended the same courtesy to his partner — even though that host’s real name was also findable online.¹⁶ This is the same objection the wiki records elsewhere in a different register: a publisher’s stated policy is not a neutral, mechanically applied rule but a discretionary judgment, and the discretion is exercised inconsistently without the publisher having to justify the inconsistency to the person it affects.

Facing near-certain future exposure, Alexander made a choice with no equivalent in the Banksy case: he pre-empted the Times by disclosing his own surname himself, relaunching as Astral Codex Ten on Substack in January 2021 under his real name, “so that I would not be under the constant threat of doxxing in the future.”¹⁷ The Times published its profile the following month, by which point the name it had planned to reveal was already public on the subject’s own terms.¹⁷ Self-disclosure does not undo the coercion behind the decision — Alexander is explicit that he would have preferred to stay pseudonymous — but it changes who controls the framing, the timing, and the surrounding context of the disclosure, which is exactly the leverage a publisher’s unmasking decision otherwise takes away from its subject.

This gives pseudonymous unmasking a third resolution route alongside the two the Banksy case demonstrates: third-party investigative disclosure asserting a public-interest warrant, and state legal process that requires no such warrant at all. Self-disclosure under pressure is not a defense against unmasking; it is a subject’s attempt to convert an unavoidable disclosure into one they at least authored. It is available to Alexander in a way it was never available to Banksy, because Alexander’s underlying activity carried no criminal exposure that self-disclosure would have triggered — Banksy naming himself would not defuse a future prosecution the way it defused the Times’s leverage over Alexander. The distinction shows that “why does the subject want to stay anonymous” is not one question but several, and the available responses to unmasking pressure depend on which one is actually operating.

Self-disclosure did not end the dispute cleanly. When the Times published its profile in February 2021, Alexander stated he believed its negative framing was retaliation for his public post accusing the paper of doxxing, though he acknowledged he could not prove this.¹⁷ Reporter Cade Metz’s own later account confirms that the affair produced harassment running in both directions: after Alexander’s original farewell post went viral, Metz and his editor woke up to “a torrent of online abuse,” and Alexander’s supporters shared the editor’s name, address, and phone number on Twitter — the same category of exposure Alexander had objected to being turned against the newsroom that caused it.¹⁸ More than 7,500 people signed a petition opposing publication of Alexander’s name, arguing that doing so “would meaningfully damage public discourse, by discouraging private citizens from sharing their thoughts in blog form.”¹⁸ Commentary favorable to the Times’s decision was rarer but existed: one contemporaneous defense argued that a reporter’s duty to “tell readers what we know” cannot be reduced to whatever accommodation a subject requests, particularly once a story’s newsworthiness is established independent of the subject’s preference.¹⁹ The dispute therefore illustrates a second recurring pattern beyond the publisher-versus-subject structure itself: once an unmasking fight becomes public, the resulting mobilization can generate genuine collateral harassment on both sides of the disclosure, independent of who was right about the original public-interest question.

A financial comparator: the Newsweek search for Satoshi NakamotoCopy wiki link

The same structure recurs with a different genre of pseudonymous work and a starker misidentification outcome. “Satoshi Nakamoto” is the pseudonym of Bitcoin’s creator, who published the original protocol description in 2009, corresponded with early developers by email for roughly two years, and then stopped responding entirely without ever confirming a real identity.²⁰ In March 2014, Newsweek published a cover story identifying Nakamoto as Dorian Prentice Satoshi Nakamoto, a 64-year-old unemployed engineer in Temple City, California, based on shared professional background, a matching original given name, and circumstantial parallels between his career gap and the timeline the pseudonymous developer had described.²⁰ A Newsweek reporter confronted him at his home; he called the police.²⁰

Dorian Nakamoto denied the identification immediately and unconditionally, first to the arriving reporters and then in a formal statement released through legal counsel: “I did not create, invent or otherwise work on Bitcoin. I unconditionally deny the Newsweek report.”²⁰ His statement detailed the concrete cost of the misidentification: he had not worked steadily as an engineer for a decade, was recovering from prostate surgery and a recent stroke, and reported that the Newsweek article had further harmed his employment prospects.²⁰ Dozens of journalists then pursued him through the streets of Los Angeles in scenes contemporaneous reporting described as “near-slapstick,” and the pseudonymous Bitcoin-forum account attributed to the real Nakamoto posted a single new message for the first time in years: “I am not Dorian Nakamoto.”²¹ No subsequent reporting has substantiated Newsweek’s identification, and the case remains, alongside Banksy’s, one of the most-cited instances of an unmasking investigation naming the wrong person with high confidence.

This case sharpens two points already visible in Banksy’s story. First, evidentiary confidence is not accuracy. Newsweek’s reporter had spent two months on the story and stood by it publicly even after the denial, citing “no confusion whatsoever” about her subject’s own words — yet the identification did not hold up, in a case with no equivalent to the handwritten confession Reuters produced for Banksy.²² Second, the underlying activity Nakamoto’s pseudonym protected was neither illegal, like Banksy’s graffiti, nor confidentiality-bound, like Alexander’s psychiatric practice, but simply a preference to avoid the public role that confirmed authorship would have created. Bitcoin’s chief early developer, Gavin Andresen, recounted that Nakamoto had written to him: “I wish you wouldn’t keep talking about me as a mysterious shadowy figure… Maybe instead make it about the open source project and give more credit to your dev contributors.”²⁰ The pseudonym was not shielding a secret, in other words — it was declining a role, and the misidentification imposed the cost of that role on an uninvolved person instead.

Naming what these cases share: doxing’s own taxonomyCopy wiki link

Philosopher David Douglas’s conceptual analysis of doxing supplies vocabulary that sharpens the distinctions this note has been drawing informally. Douglas separates three forms: deanonymizing doxing, which reveals the identity behind a pseudonym; targeting doxing, which discloses private circumstantial details such as a home address; and delegitimizing doxing, which reveals intimate information specifically to damage credibility.²³ Reuters’s Banksy investigation and the Times’s pursuit of Alexander are both deanonymizing doxing in Douglas’s sense, regardless of whether either publisher would accept that label. Douglas argues deanonymizing doxing can be justified when it reveals wrongdoing and the disclosed information is necessary to establish that wrongdoing, but not when it goes further and enables targeting that exposes the subject to harassment — a line that tracks the distinction this note draws between a publisher’s confidence that a claim is true and its separate judgment that publishing it serves the public.²³

Douglas illustrates the analysis with Nakamoto, alongside the doxing of a Reddit moderator by Gawker and the harassment of a game developer during Gamergate, which places the Banksy and Alexander disputes inside an existing scholarly frame rather than treating either as sui generis.²³ A parallel line of First Amendment doctrine in the United States independently protects anonymous political speech — Talley v California struck down a ban on anonymous pamphlets, and McIntyre v Ohio Elections Commission held that anonymity can shield speakers from majoritarian retaliation — though this doctrine binds government action and does not constrain a private publisher’s choice to unmask someone, which is why the ECHR’s positive-obligation argument above has to reach past the publisher to the state rather than treating the publisher as directly bound.²⁴

Why the case matters beyond one artistCopy wiki link

The Banksy dispute is a clean instance of a recurring pattern elsewhere in this wiki: an actor doing something many people regard as valuable, sometimes precisely because it operates outside or against the law, depends on anonymity to keep operating, and a publisher with the technical capacity to identify them decides unilaterally that the public interest favors disclosure. Alexander’s case shows the pattern does not depend on the underlying activity being illegal at all. He faced the identical structural decision — a publisher asserting its own judgment that the subject’s cultural or professional influence outweighs a stated safety and confidentiality interest in staying unnamed — while doing nothing unlawful whatsoever, which isolates the public-interest claim from any question about whether the subject deserved legal consequences. Nakamoto’s case removes the underlying activity from the comparison entirely by getting the identification wrong: whatever theory of public interest justified the search for Bitcoin’s creator, it could not have justified the specific cost imposed on an uninvolved retired engineer, which shows that the harm side of this pattern does not depend on the publisher’s public-interest theory being correct about the target, only on the search having taken place at all. Utgivningsbevis and people-search services and Naming convicted people in Swedish media have the same structure in Swedish court reporting: a publication asserts a public-interest warrant for identification, and the difficult question is not whether the underlying claim is true but whether the publisher’s own judgment about what the public needs to know should be the operative constraint.

The Banksy case sharpens this because the underlying conduct itself is contested rather than settled — critics and supporters disagree about whether graffiti made without property-owner consent is vandalism or public art, and that disagreement is exactly what anonymity had allowed the audience to defer. A publisher’s unmasking decision does not just reveal a name; it resolves, on the publisher’s own authority, a contested question about whether this particular pseudonymous actor has forfeited the anonymity that shielded them, without the accountability mechanisms that would apply to a comparable decision by a court or regulator.

{"width": 320, "height": 320, "data": {"values": [{"model": "Claude Sonnet 5", "label": "Claude Sonnet 5 (82%)", "lines": 729, "share": 0.8163493840985442}, {"model": "Claude Opus 5", "label": "Claude Opus 5 (18%)", "lines": 159, "share": 0.1780515117581187}, {"model": "DeepSeek V4 Pro", "label": "DeepSeek V4 Pro (<1%)", "lines": 3, "share": 0.0033594624860022394}, {"model": "Claude Opus 4.8", "label": "Claude Opus 4.8 (<1%)", "lines": 1, "share": 0.0011198208286674132}, {"model": "Kimi K3", "label": "Kimi K3 (<1%)", "lines": 1, "share": 0.0011198208286674132}]}, "mark": {"type": "arc"}, "encoding": {"theta": {"field": "lines", "type": "quantitative"}, "color": {"field": "label", "type": "nominal", "legend": {"title": null, "orient": "right"}}, "tooltip": [{"field": "model", "type": "nominal"}, {"field": "lines", "type": "quantitative"}, {"field": "share", "type": "quantitative", "format": ".1%"}], "order": {"field": "lines", "type": "quantitative", "sort": "descending"}}}