{"slug": "solo-browser-passes-acid3-now-show-the-code", "title": "Solo Browser Passes Acid3 — Now Show the Code", "summary": "A solo developer, HN user coolwulf, claims his from-scratch Zig browser engine, cwbrowser, passes the Acid3 test with a perfect 100/100 score, but the project has no source code, binary, or license, only a landing page and a claim of being '2× faster than Chrome.' Acid3, released in March 2008 by Ian Hickson, is now deprecated and no longer reflects modern web standards, with current Chrome and Firefox scoring 97/100, making the achievement a rite of passage rather than proof of real-world capability. The developer acknowledged heavy use of Claude and agentic tooling, sparking debate about the meaning of 'from scratch,' and the project lacks Web Platform Tests results, the current benchmark for standards compliance.", "body_md": "[Dev Tools](https://sourcefeed.dev/c/dev-tools)Article\n\n# Solo Browser Passes Acid3 — Now Show the Code\n\ncwbrowser claims a from-scratch Zig engine scoring 100/100, but in 2026 Acid3 is the easy part.\n\n[Mariana Souza](https://sourcefeed.dev/u/mariana_souza)\n\nA Show HN post hit the front page this week with a claim that would have been front-page news in 2008: a browser engine written from scratch by one person, over two years, now passes the Acid3 test with a perfect 100/100. The project, [cwbrowser](https://code.intellios.ai/cwbrowser/), pairs a hand-written rendering engine in [Zig](https://ziglang.org/) — tokenizer, DOM, CSS cascade, layout, paint — with Google's [V8](https://v8.dev/) for JavaScript. No Chromium, no WebKit, no Gecko.\n\nIf true, that's a genuinely rare feat. The problem is the \"if.\" There's no source code, no downloadable binary, and no license — just a landing page, a screenshot-grade claim of being \"2× faster than Chrome,\" and a macOS Apple Silicon build that's \"coming soon.\" The Hacker News thread split accordingly: real admiration for the ambition, and pointed questions about why an engineering milestone arrived with nothing you can run or read.\n\n## Acid3 is a rite of passage, not a bar\n\nSome history matters here, because Acid3's meaning has inverted since it launched. Ian Hickson released the test in March 2008 as a gauntlet for the browser wars: 100 subtests covering DOM traversal, JavaScript edge cases, SVG, and the standards that powered \"Web 2.0.\" WebKit and Opera raced to 100 within weeks; Firefox took years; IE8 scored 20. In September 2011 the test itself was amended — the SVG fonts subtests were dropped because the specs had moved on — and that's what finally let Firefox and IE9 claim perfect scores.\n\nThen the platform kept moving and the test didn't. Today's Chrome and Firefox actually score 97/100, failing three subtests that no longer match current specs, and Hickson himself has said the test no longer reflects web-standards consensus. Acid3 is deprecated. Passing it in 2026 tells you an engine correctly implements a snapshot of the web circa 2008 — no flexbox, no grid, no ES modules, none of the ten-thousand-odd behaviors a modern site actually exercises.\n\nThat doesn't make it worthless. It makes it a rite of passage. [Ladybird](https://ladybird.org/) hit 100/100 in April 2022 back when it was still the SerenityOS browser, and treated it as a morale milestone on the way to the real work. Ekioh's Flow did it earlier with the same architecture cwbrowser uses — proprietary from-scratch layout engine, borrowed JavaScript VM (SpiderMonkey in Flow's case, V8 here). The pattern is established: Acid3 is where a new engine announces it's serious. It is not where an engine proves it can browse the web.\n\nThe actual scoreboard now is the [Web Platform Tests](https://wpt.fyi/) suite — tens of thousands of test files, millions of subtests, continuously updated, publicly dashboarded. Ladybird and Servo publish their WPT numbers and let anyone watch the line go up. That's what \"standards compliance\" means in 2026, and it's conspicuously absent from cwbrowser's page.\n\n## Extraordinary claims, zero artifacts\n\nThe \"2× faster than Chrome\" line deserves the most skepticism. Faster at what? Commenters asked for Speedometer 3.1 numbers — the industry's shared yardstick, where Chrome and Safari cluster in the 50–60 range on Apple Silicon — and got none. A young engine can easily beat Chrome on cold start or memory floor, because it isn't carrying site isolation, sandboxing, accessibility trees, GPU process architecture, or twenty years of compatibility shims. Those aren't bloat; they're the product. Any benchmark that ignores them is measuring a different category of software.\n\nThe author, HN user coolwulf, was also candid that the project leaned heavily on Claude and personal agentic tooling during development — which triggered a second round of debate about what \"from scratch\" means now. I'd separate the two questions. AI-assisted authorship doesn't diminish a working engine; if anything, a solo dev standing up a layout engine in two years with LLM leverage is the most interesting fact in the whole story, because it suggests the fixed cost of engine development — historically hundreds of person-years — is genuinely deflating. Zig is quietly becoming the language of this movement, too: Lightpanda's headless browser is Zig, and now this.\n\nBut AI leverage makes verification more important, not less. It's never been easier to produce something that demos well and collapses on the long tail. The long tail is the entire job of a browser.\n\n## What would change my mind\n\nFor working developers, the immediate answer is: nothing to adopt, nothing to test, nothing to fork. This isn't a tool decision yet; it's a credibility decision. And the credibility playbook for new engines is well established, because Ladybird, Servo, and Flow already wrote it:\n\n**Ship a binary.** Flow is proprietary too, but Ekioh ships downloadable builds anyone can point at Acid3 — or at their own site.**Publish WPT numbers.** Even a 30% pass rate, honestly reported, is more persuasive than a perfect score on a deprecated 2008 test.**Benchmark on shared yardsticks.** Speedometer 3.1 against Chrome and Safari on the same M-series hardware, methodology included.\n\nUntil at least one of those lands, the right posture is friendly suspension of judgment. I genuinely hope it's real — the web needs engine diversity, and the two credible independent efforts (Ladybird, targeting alpha releases, and Servo, rebuilding momentum under Linux Foundation Europe) are still years from daily-driver status. A world where a motivated individual can field a fourth engine is a better web than the Blink-WebKit-Gecko oligopoly we've had since Opera folded Presto in 2013.\n\nBut the burden of proof runs in one direction. Acid3 was designed for a world where every claimant shipped a browser you could run the test in yourself. cwbrowser has, so far, inverted that: the score arrived before the software. Two years of work earns respect. A URL and a number don't yet earn belief.\n\n## Sources & further reading\n\n-\n[cwbrowser — a browser engine written from scratch in Zig](https://code.intellios.ai/cwbrowser/)— code.intellios.ai -\n[Show HN: I worked on a new browser for 2 years, today it passed Acid 3](https://news.ycombinator.com/item?id=49128826)— news.ycombinator.com -\n[Acid3](https://en.wikipedia.org/wiki/Acid3)— en.wikipedia.org -\n[SerenityOS Browser now passes the Acid3 test](https://news.ycombinator.com/item?id=30853392)— news.ycombinator.com\n\n[Mariana Souza](https://sourcefeed.dev/u/mariana_souza)· Senior Editor\n\nMariana covers the fast-moving world of machine learning and generative AI, with a particular focus on how these technologies are reshaping development workflows. When she isn't stress-testing the latest foundation models, she's usually at a local hackathon.\n\n## Discussion 0\n\nNo comments yet\n\nBe the first to weigh in.", "url": "https://wpnews.pro/news/solo-browser-passes-acid3-now-show-the-code", "canonical_source": "https://sourcefeed.dev/a/solo-browser-passes-acid3-now-show-the-code", "published_at": "2026-08-01 09:08:47+00:00", "updated_at": "2026-08-01 09:26:25.413035+00:00", "lang": "en", "topics": ["developer-tools", "artificial-intelligence"], "entities": ["cwbrowser", "Zig", "V8", "Acid3", "Ian Hickson", "Ladybird", "SerenityOS", "Web Platform Tests"], "alternates": {"html": "https://wpnews.pro/news/solo-browser-passes-acid3-now-show-the-code", "markdown": "https://wpnews.pro/news/solo-browser-passes-acid3-now-show-the-code.md", "text": "https://wpnews.pro/news/solo-browser-passes-acid3-now-show-the-code.txt", "jsonld": "https://wpnews.pro/news/solo-browser-passes-acid3-now-show-the-code.jsonld"}}