Arguing With a Talk

There is a sentence in the middle of Frank Chimero's talk The Web's Grain that I have quoted to other designers for years without ever checking whether it was true. It runs: "But, this contradiction is unavoidable and unsolvable, so the only choice is to recognize it as implicit to the medium, and devise strategies to manage it." It was correct when he said it. It stopped being correct about eight years later, and the way it stopped being correct is the most interesting thing about the talk.

The occasion is precisely documented, which is why I picked it. Webstock '15 ran from 16 to 20 February 2015 at the St James Theatre in Wellington, New Zealand. Chimero spoke on a bill that also included Genevieve Bell, Cory Doctorow, Brad Frost, Harper Reed and the typeface designer Kris Sowersby. The conference's own talk page carries the abstract. The full text was published on Chimero's site on 18 February 2015, and the recording sits on the Internet Archive under the title "Webstock '15: Frank Chimero - The Web's Grain". I have read the published text. I was not in the theatre, and every quotation below comes from the text as Chimero published it.

The argument, as he made it

The abstract states the frame plainly: "We've been talking about what it means to design natively for the web for ages, but the conversation has been rekindled with the prevalence of responsive design paired with the unpredictability of devices, environments, and connectivity. If we look at the web as a material and not as a canvas, how do its affordances guide a designer's hand?"

That is the whole move, and everything else follows from it. The talk asks what happens if a designer treats the web the way a cabinetmaker treats oak.

Grain

Chimero's central proposition is short enough to fit on a card:

I believe every material has a grain, including the web.

He sets it against the opposite belief, which he says is the default: "Too often, the internet is cast as a wide-open, infinitely malleable material." The corrective question he poses is "What would happen if we stopped treating the web like a blank canvas to paint on, and instead like a material to build with?"

The worked example that makes the idea concrete is the behaviour of an image and a paragraph in a resizing window. In his words: "When I change the window width, the image gets taller as it gets wider, because its proportions are fixed. The text, on the other hand, gets shorter as it gets wider." Anyone who has built a responsive layout knows this in their hands. As the window widens, a picture occupies more vertical space and a block of copy occupies less. Put them beside each other and the relationship between them will not hold at two different widths. This is what makes layouts "look weird", and it is the reason designers reach for a breakpoint.

His conclusion from the example is the sentence I opened with. The contradiction is "unavoidable and unsolvable"; it is "implicit to the medium"; the designer's job is to devise strategies to manage it, not to defeat it.

Edgelessness

The second half of the talk names the property Chimero thinks is most characteristic of the web. "A lack of edges permeates the web at all levels." He then gives four levels explicitly:

• "Edgelessness is in the web's structure: it's comprised of individual pages linked together, so its structure can branch out forever."

• "Edgelessness applies to the screens that show the web, because they offer an infinite canvas that can scroll in any direction for however long."

• "Edgelessness speaks to the diffusion of device and viewport sizes."

• "And, most interesting to me, edgelessness means blurred lines between the disciplines that work together to make things for the web."

Assemblage

The prescription comes via David Hockney's photographic joiners, the collaged panoramas Hockney made from dozens of overlapping prints. "So just like Hockney's joiners, we're creating assemblages of elements, then associating them with the appropriate space." The unit of design is not the page. It is the element plus the rule about where it goes.

The talk ends somewhere other than technique. "My wish is simple: I desire a technology of grace, one that lives well within its role." And the last line: "How will we know that we're there? I suppose we'll look at what we've built, notice how the edges have dropped away, and actually be pleased it looks like it could go on forever."

That is a fair summary. It is a good talk, better written than most essays, and I am about to disagree with it in one specific place.

A conference audience. Photograph by Lf Asia, CC BY 3.0, via Wikimedia Commons.

The grain was a backlog

Here is the disagreement. A material's grain is a physical fact. Oak has a grain because of how a tree deposits cellulose over seasons, and no committee can revise it. Chimero's argument borrows the authority of that kind of fact. When he says the image-and-text contradiction is "implicit to the medium", he is claiming that the behaviour belongs to the web the way fibre direction belongs to timber.

It did not. It belonged to the CSS specification as it stood in February 2015.

The specific thing that made his example unsolvable is that in 2015 an element could only respond to the viewport. A media query asks how wide the window is. It cannot ask how wide the box the component happens to be sitting in is. So a component that needed to change its internal proportions when it got narrow had no way of knowing it had got narrow, unless the whole window had got narrow at the same time. That is why designers added breakpoints, and it is why layouts "look weird".

Container queries removed that limitation. They let a component interrogate its own container rather than the window, which is exactly the missing question. The MDN reference describes them as "an alternative to media queries". According to web.dev's Baseline 2023 summary, support landed in Chrome 105, Edge 105, Firefox 110 and Safari 16, and that combination is what makes a feature Baseline Newly available: the same article defines it as the point when a feature "becomes interoperable in the following browsers: Safari (macOS and iOS), Firefox (desktop and Android), Chrome (desktop and Android), Edge (desktop)".

None of this is an argument that Chimero was careless. It is an argument that the metaphor was doing more work than a metaphor can survive. "Grain" carries a claim of permanence. What he had identified was a temporary gap in a specification maintained by a working group with a queue. Calling that gap the grain of the material told a generation of designers that a solvable problem was a law of nature, and the practical consequence was real: breakpoint-driven layout stayed the default approach for the best part of a decade, defended on the grounds that the medium demanded it.

There is a name for the move. Treating the current state of a designed artefact as though it were the nature of the world is how conventions become invisible. The web is not a material. It is a stack of documents written by people who take comments and ship revisions. Its grain is a changelog.

Four things in one word

The second place I would push is "edgelessness", and here my objection is about precision rather than truth.

The four senses Chimero lists are not the same kind of claim. The first, that a hyperlinked structure branches indefinitely, is a genuine property of the web's design: it follows from what a link is. The second, that screens offer an infinite scrolling canvas, is a property of a scroll container in a browser engine, and is contingent in the same way container queries were contingent. The third, "the diffusion of device and viewport sizes", is not a property of the web at all. It is a fact about a hardware market in 2015. The fourth, the blurring of disciplinary boundaries, is a fact about how studios and employers organise labour.

Bundling these gives the talk its force, and it is the bundling that makes the prescription slippery. If edgelessness is one thing, then designing for it is one skill. It is not one thing. A designer who accepts that hyperlinks branch forever has learned something about information architecture. A designer who accepts that viewport sizes are unpredictable has learned something about a specific market, and the correct response to a market is to look at the numbers for your own audience, not to infer a design philosophy from the word "diffusion".

The strongest of the four is the one Chimero flags as most interesting to him, and I think he was right about it in a way that was not obvious in 2015. Disciplinary boundaries did keep blurring, and the institutional answer that emerged, the design system with its shared component vocabulary and its joint ownership by design and engineering, is precisely an attempt to make an edgeless collaboration tractable. His fourth sense of edgelessness predicted the dominant working method of the following decade. His third sense described a phase of the smartphone market.

What I had wrong

For several years I used "the web has a grain" to end arguments in studio. Someone would want a fixed layout, or a specific relationship between a photograph and a caption that only held at one width, and I would say the medium did not work that way, and the conversation would stop. That is the wrong use of a good idea. Chimero offered it as a way to start looking at the thing in front of you. I was using it as a rule to avoid looking.

The test of whether you are doing this is simple: if your reason for not doing something is a property of the medium, go and find the specification that makes it a property. If there is no specification, it is a habit. If there is a specification, check the date on it.

What I am not claiming, and what I could not check

I am not claiming Chimero would defend the sentence today. He published the talk in 2015 and I have not found a later statement from him revising it, but I also have not looked exhaustively, and the absence of a retraction is not evidence of a position.

I am not claiming container queries fixed responsive design. They removed one specific impossibility. Whether designers took the opportunity is a separate question, and one I cannot answer with the sources I have; I could not retrieve usable figures on the adoption of container queries in production sites, and I would not want to guess.

I am also aware that a talk is not a paper. A metaphor delivered from a stage in Wellington to a room of a few hundred people has a different job from a claim in a journal, and holding it to the standard of a claim is slightly unfair. My defence is that the text was published, stayed online for more than a decade, and became one of the most quoted pieces of design writing about the web. Once something is a citation, it has to answer like one.

The last line of the talk asks how we will know a technology of grace has arrived: we will look at what we have built and be pleased it looks like it could go on forever. The web did go on forever. The part that turned out to have edges was the specification, and the specification moved.

Sources

• Frank Chimero, "The Web's Grain", published text of a talk given at Webstock '15, published 18 February 2015. frankchimero.com/blog/2015/the-webs-grain/

• Webstock, "The Web's Grain", talk page with abstract, Webstock '15. www.webstock.org.nz/talks/the-webs-grain/

• Webstock, "Webstock, 16 - 20 February 2015", conference site listing dates, venue (St James Theatre, Wellington) and speakers. www.webstock.org.nz/15/

• Webstock, "Webstock '15: Frank Chimero - The Web's Grain", video recording, Internet Archive. archive.org/details/vimeo-122880890

• "Baseline 2023", web.dev blog, Google; definition of Baseline Newly available and browser versions for container queries, subgrid, CSS nesting and other 2023 features. web.dev/blog/baseline2023

• MDN Web Docs, "CSS container queries", Mozilla; describes container queries as an alternative to media queries. developer.mozilla.org/en-US/docs/Web/CSS/CSS_containment/Container_queries

Related reading

• The Helvetica and Arial Question, Settled Properly

• Notes From an Exhibition

• What a Portfolio Is Actually For

• How Studios Lose Their Own Work

More from the Moc Mallen Studio journal