I found this article https://www.jonoalderson.com/conjecture/weirdly-human/ from Jono Alderson it is aone of the more interesting takes on where we are headed. Give it a read.

I like Jono Alderson’s piece, quite a lot actually, because beneath the shiny Chrome experiment and the inevitable developer excitement, there is something more human sitting there, something I think we have been quietly ignoring while pretending that pages, menus, hierarchies, scroll depth and all the rest of the digital furniture were somehow natural, rather than just the compromise we all agreed to live inside.[1]

The Core Idea

What lands for me is fairly simple. For decades the web has been shaped by the document metaphor, pages, structures, navigation, vertical scrolling, headings, menus and all the obedient little boxes we have trained ourselves to accept as 'the web'. Not because this is the best way to represent ideas, products, places, communities, knowledge or experiences, but because it was the most workable bargain between browsers, search engines, CMS platforms, analytics tools, accessibility requirements and the general machinery of the internet.[1]

We built a world of rectangles because the machines could understand rectangles.

That is the part worth sitting with.

HTML-in-Canvas, or whatever shape this ultimately takes, is interesting to me because it starts to loosen that old bargain. Chrome’s own explanation of the experiment describes the API as a way to draw DOM content into Canvas, WebGL or WebGPU environments while keeping the content interactive, accessible and connected to browser features.[2]

The machines get their clean DOM, the crawlable content, the accessibility hooks, the structured meaning, while the human being on the other side of the glass gets something that is allowed to breathe a little.

Yes, the demos are clever. Doom walls with interactive HTML, parallax environments, shaders running across real DOM elements, all very dogs bollocks in the right hands and absolute migraine fuel in the wrong ones.[1] But the demos are not the point. The deeper point is that the web may finally be able to stop pretending that every human experience should be poured into a rectangular document and then politely asked to behave itself.

Why 'Weirdly Human' Matters

This is the phrase that lands for me. 'Human' does not mean nostalgic, warm, fuzzy, skeuomorphic nonsense, like putting leather stitching on a calendar app and calling it design. It means something far more useful, more faithful representation.

A museum website should not have to feel like a brochure that lost its way in a CMS. It could feel like wandering through an exhibition space.

A product site should not always have to be a landing page with a hero image, three benefit blocks and a testimonial from someone called Sarah who may or may not exist. It could be an explorable environment.

A knowledge site should not always have to be a tree of pages, each one begging for internal links and schema like some little SEO hostage. It could become a landscape of connected ideas, something closer to how thought itself actually behaves.

That, to me, is the interesting bit. Because as machines get better at understanding meaning, through structured data, entities, embeddings, LLMs and all the other machinery currently being sold back to us as magic, the pressure on the human-facing layer may reduce. Jono makes a similar point when he argues that different audiences may increasingly consume different representations of the same underlying information.[1]

Machines can have their clean representations. Humans can have something richer, stranger, more textured.

For years we optimised human experience down to what machines could reliably parse. This flips the direction of travel. Or at least, it gives us the chance to.

My Take

I am cautiously bullish on this direction. The web has been creatively constrained for a long time, often by good reasons, but constraints are still constraints. Flash showed us that people wanted more than documents. It also showed us what happens when the web gets drunk on itself, forgets accessibility, forgets interoperability, forgets openness, and decides that spinning logos and loading screens are a substitute for substance.

Modern WebGL, Three.js and experimental spatial interfaces have kept that flame alive, but too often they require heroic effort and, in many cases, they trade away the very things that made the web useful in the first place. Crawlability disappears. Accessibility becomes an afterthought. Performance goes wandering off into the bushes. And then somebody in a meeting says 'immersive' and everyone nods like Deep Thought has finally coughed up the number 42.

If HTML-in-Canvas, or something like it, matures properly and is not trapped inside another 'works best in Chrome' cul-de-sac, it could unlock a genuinely new wave of experiential web design without repeating all the old sins. That is the promise. Presentation becomes freer, while meaning remains intact underneath.

That aligns very neatly with where AI is pushing things too. Not in the breathless LinkedIn sense where every intern with a prompt is apparently revolutionising civilisation, but in the more grounded sense that we are separating the representation of meaning from the surface treatment of it.

The Caveats

There are caveats, because of course there are. The internet has never met a good idea it could not over-commercialise, standardise, ruin, resell and then write a thought leadership piece about.

  • Adoption matters. Origin trials are one thing. Cross-browser support, developer tooling, performance on lower-end devices and actual production reliability are quite another. I do not think we need another era where the web becomes a polite hostage to one browser’s implementation choices.
  • Accessibility and crawlability must hold up in practice. Chrome’s article says HTML-in-Canvas content can remain exposed to accessibility systems and indexable by crawlers and AI agents, but the real test is what happens at scale, inside complex production sites, with imperfect teams and moving deadlines.[2]
  • Freedom often creates excess before it creates culture. We may get beautiful, idiosyncratic, memorable things. We will also get a great deal of motion-sickening garbage, because people will confuse movement with meaning, as they usually do.
  • The 'weirdly' part matters. Truly human things are often inefficient, personal, slightly awkward, oddly shaped and not immediately measurable. That is their beauty. But platforms, SEO incentives, analytics dashboards and business pressures tend to sand those edges off.

The human bits are usually the first things dragged outside and quietly shot.

Overall

Overall, I think this feels like a healthy counterweight to the dreary 'AI will eat the web' narrative. Maybe the web does not have to become a landfill of bland generated sludge, optimised landing pages and machine-readable mush. Maybe, if we are careful, the machine layer can become cleaner while the human layer becomes more varied, more expressive, more spatial, more playful and, yes, more weirdly human.

Jono has a good eye for these shifts, particularly where SEO, semantics and the broader evolution of the web overlap. This one is worth paying attention to.

Because a web that becomes less 'page-shaped' and more 'idea-shaped' or 'experience-shaped' is not just a technical upgrade. It might be a small act of cultural correction.

And God knows, the web could do with one.

References

[1]: [https://www.jonoalderson.com/conjecture/weirdly-human/]("Jono Alderson, The future of the web is weirdly human")
[2]: [https://developer.chrome.com/blog/html-in-canvas-origin-trial]("Chrome for Developers, Introducing the HTML-in-Canvas API origin trial")