Design: Weekly Summary (January 05-11, 2026)

This week’s reading felt like wandering through a busy market of small, sharp ideas about design. Little stalls everywhere. Some shouted about AI. Some whispered about icons. A few had practical tricks you can actually use tonight, like a friend nudging you over a pint. I would describe them as close-up looks at tiny decisions that end up meaning a lot. To me, it feels like the conversation is no longer just about pretty pixels. It’s about tools, stories, rules, and mistakes — and about whether we keep the doors open for change or bolt them shut.

A quick quiz that turns into a conversation about what’s next

Jakob Nielsen put out a long quiz called “The Big Tough UX-AI Quiz” and followed up with the answers a few days later. The quiz itself is a fun, slightly smug nudge: 70 questions that make you squint at the last year and ask whether you actually noticed the shifts. It’s a neat trick. Ask people to test themselves and suddenly they care about the landscape.

The answers post expands on some phrases you’ll keep hearing: ‘‘Slow AI’’ and ‘‘Generative Engine Optimization’’ (GEO). Those two sounded like jargon at first. But in the answers they feel useful. Slow AI isn’t an argument to avoid speed. It’s a reminder that some interfaces want deliberate, human-paced interactions. GEO is a reminder that the way we query and shape models becomes a craft, not a button press. I’d say both ideas put people back into the center, in different ways — one asks for patience, the other asks for craft.

Also in that thread: the change to job roles. It’s not just “AI replaces X.” It’s more like, ‘‘AI changes what counts as skill.’’ Some jobs will become about framing problems and spotting edge cases. Others will be about validating AI output and keeping it honest. The quiz-and-answers combo is an invitation. It says: if you want to keep designing for people, learn these moves.

If you like testing yourself, the quiz is a cheeky start. If you like reading answers, the follow-up is a tidy map of where UX met AI in 2025. Check both for the details and the little surprises.

Icons, menus, and why one tiny change makes people mutter

This week’s small-but-loud debate: Apple’s menu icon choices. Nick Heer and Lucio Bragagnolo were not happy. Nick Heer called MacOS Tahoe’s menu icons inconsistent, confusing, and basically illegible. Lucio Bragagnolo wrote in Italian that adding icons to menus harms the Mac interface, leaning on Nikita Prokopov’s analysis. Same gripe, different flavors. The two posts are like neighbors who both complain about the new lamp post on the street. You notice the lamp post every time you walk by.

I would describe the core complaint as this: icons are being sprinkled everywhere without clear metaphors. That’s not decoration. That’s a language. If the language is messy, the whole menu feels noisy. It’s like someone decided to put emojis on every room label in a house. Funny one day. Confusing the next.

This links to another small annoyance from Michael J. Tsai about CarPlay and the Go button. He reminds us how tiny visual cues — shape, behavior, context — carry a lot of weight. The Go button looks like a passive label, not an action, and that mismatch trips people up. It’s like putting a light switch where you expect a thermostat. You fiddle with it because your brain expects one thing and the interface gives another.

These posts together make the same point: visual consistency and clear metaphors still matter. They’re not old-fashioned. They’re the plumbing. Mess with them and people will notice. They’ll notice slowly, then loudly.

Layers, colors, and keeping your options open

If icons are the language, then layers and colors are the alphabet. Alice Packard wrote a small manifesto about not flattening SVG icon layers. The post is short and practical but quietly defiant. Flattening layers is tempting. It looks tidy. But then color overrides stop working. Animations become harder. The author suggests keeping separate layers and using shape-building operations so you can still ship clean assets while retaining flexibility.

This is the kind of post that feels like advice from someone who’s ruined a Friday night because of a messed-up export. Practical, with a little sharpness. To me it feels like the difference between living out of a messy drawer that you actually can find things in, versus a neatly folded wardrobe where everything feels stuck under a sweater. Keep the layers separate if you might need to recolor later.

It pairs naturally with Dave Rupert writing about the contrast-color() function and how CSS only returns black or white today. Dave walks through limitations and a clever workaround (Lea Verou’s method). That’s fond-of-hacks stuff. It’s also a reminder: browsers and standards are slow. Designers invent workarounds. Developers and designers trade those tricks like recipes.

And then there’s Jan Maarten in their 2025 review who mentions annotating design systems. It’s a small thread: design systems need to be usable, editable, and forgiving. Layers, colors, contrast — they all feed that same need. Keep options open. Keep the system readable. Don’t bake in choices you’ll regret.

Don’t fall in love with the product — love the problem instead

Greg Morris wrote a straight-to-the-point piece: don’t fall in love with the product. I’d describe his voice as blunt and kind in one breath. He divides features into ‘‘painkillers’’—urgent, must-have fixes—and ‘‘vitamins’’—nice-to-have luxuries. The warning is practical: if you’re emotionally hooked on an idea, you’ll ignore evidence that it’s not solving anything real.

This is a theme that appears again in other pieces. Paul Jun asks for new stories in design. He wants a shared narrative that goes beyond nostalgia. That’s aligned with Greg’s thought. If your product or movement is more in love with its own rhetoric than the problems it solves, you end up recycling images from the 20th century rather than solving 21st-century pain.

The language in Greg’s piece is useful because it’s not academic. It’s the kind of talk you have at a café when a friend convinces you to stop chasing a pet project and start talking to actual users. Less vanity. More questions.

When an LLM stops your microcontroller from dying

This was a neat one. Bert Wagner explains how a large language model saved an ATtiny85 microcontroller from certain death. The author was buried in datasheets and choices about pull-ups and tolerances. The LLM pointed out a likely fatal flaw. The chip survived.

This feels like a small miracle, but not a flashy one. It’s the kind of help that saves you an evening of head-scratching and a handful of burned components. To me it feels like having a really competent friend on the phone at 2 a.m. who says, ‘‘Try swapping that resistor. Trust me.’’

Bert’s story connects back to the quiz about AI changes. The models aren’t just for writing copy or generating images. They’re becoming useful as practical checkers in engineering workflows. Use them to double-check specs, to spot edge-case failures. That doesn’t remove responsibility. It just changes who you ask for a sanity check.

A call for new stories, Bauhaus echoes, and the itch to remake things

Paul Jun pushed for new aesthetics and new stories in design. He’s looking for a collective narrative that is more useful than nostalgia. He pulls Bauhaus into the conversation as an example of how a shared vision can reshape environments and everyday life.

This is the romantic corner of the week’s discussions. It asks big questions. Not everyone will be into the manifestos. But even small pieces, like Chris building a new site that combines blog, diary, and photos into a static site, show the same urge: how can we reframe old content? How can we make things faster and more intimate at the same time?

That eagerness sits next to a cautionary tale from Political Calculations about the Pug anti-bandit briefcase. It’s a charming story: an invention that ejects your bag’s contents when grabbed, to deter thieves. Creative, but wildly impractical. You see the point. Outside-the-box thinking matters. But execution and context matter more. Like building a wild treehouse in a city with building inspectors. Great idea, wrong regulations.

So there’s tension. Dream big. But check whether your dream would be banned by the HOA. The point is not to dampen creativity. It’s to remember constraints. Paul’s call for stories isn’t naïve. It’s a nudge to combine ambition with real-world sense.

Small UI confusions that feel like big problems

A few posts remind us that small UI choices are more important than you think. The MacOS icons, CarPlay’s Go button, even the anti-bandit bag idea — these are small things that expose deeper thinking about metaphors and trust.

This is where Nick Heer and Michael J. Tsai overlap without quoting each other. Both say: if the interface lies to you, you stop trusting it. When a button looks like a label, you hesitate. When a menu icon resembles nothing, you stop reading menus the way you used to. That hesitation piles up. Suddenly a simple task takes longer, or you click something you wished you hadn’t. It’s a small erosion of patience.

And that erosion matters. UX is often invisible until it breaks your rhythm. Then it becomes noisy. It’s like a car that starts rattling once a week. You notice until you fix it — or until you trade the car in.

Recurring patterns I kept spotting

• Small visuals matter. Tiny icon choices. Button shapes. A misplaced label. They compound.

• Tools and standards lag. People ship clever workarounds — CSS tricks, layered SVGs, LLM checks — because the platform isn’t giving them what they need yet.

• There’s a split mood between nostalgia and invention. Some people want to reframe the story of design (Paul Jun). Others are caught up in small, practical fixes (Alice, Dave, Bert).

• AI is both craft and safety net. It asks for new skills (GEO, prompt framing) and it also helps with old problems (hardware failure, spec checking).

• Product thinking matters. Don’t fall for features that are cute but useless. The ‘‘painkiller vs vitamin’’ line keeps showing up in different clothes.

• Practicality wins most battles. The anti-bandit bag is cute, but the one that ships is the one that people are willing to carry.

You’ll notice these patterns across the posts. They repeat like a chorus because these are the questions designers actually have to answer on Monday mornings.

A few pieces of practical advice worth stealing

• Don’t flatten icon layers if you need color overrides later. Keep structure. It’s future-proof.

• Try the Lea Verou method (or similar) if you want more flexible color contrast with CSS. Browsers are slow. Hacks are OK.

• Ask whether a feature is a painkiller or a vitamin before you bury months of work into it.

• Use LLMs as a sanity-check tool for engineering. Don’t treat them as infallible, but a second pair of eyes helps.

• Watch for UI metaphors that don’t behave like they look. If a control looks passive, people will treat it like a label.

If you want the how-to steps, the posts have them. These pointers are more like bookmarks to the parts of each post that actually make late-night work easier.

Where the week felt a bit mixed or noisy

Sometimes the week slipped into repetition. Multiple voices were saying the same thing about icons and menus, and that’s okay — it’s a real problem. But after two or three takes, you want to move from critique to alternatives. A few posts did offer fixes (Alice and Dave), but others mainly raised alarms.

Also, the call for new narratives is inspiring. Yet it’s easy to say ‘‘we need a new aesthetic’’ without showing how anyone can start building it. Paul’s essay points to history and big ideas, and you leave wanting a workshop or a recipe. Which is fine. Sometimes you need the appetite before the meal. Still, I’d have liked a few more concrete steps.

Little tangents that are worth thinking about

• The anti-bandit bag makes me think of how design sometimes prioritizes statement over practicality. It’s like buying a designer jacket with no pockets. Nice to look at, annoying to wear.

• The LLM story feels like one of those late-night garage fixes where someone says ‘‘that resistor looks wrong’’ and saves you a trip to the electronics shop. It’s small, but it saves time and money. That kind of help is under-appreciated.

• The whole icons debate sounds a bit like an argument over pasta shapes when you really mean sauce. But, well, sometimes pasta shape matters in how sauce clings. Little choices do matter.

Why you might want to read the originals

Each post does one thing well. The quiz and answers are a tidy way to map the AI-UX shifts. The icon critiques show how a tiny UI change can crack the user’s trust. The SVG and CSS posts give real fixes you can use today. Bert’s hardware story is the kind of reassurance you want when you’re soldering at 11 p.m. Greg’s product piece is moral support for ruthless problem-checking. Paul’s essay gives a mood and a challenge.

If you’re the type who likes to tinker, flip through Alice’s and Dave’s how-tos. If you like context and history, Paul’s piece will sit well with your morning coffee. If you want a quick health check of your knowledge, take Jakob’s quiz and then read the answers.

Much of this week’s conversation feels like a neighborhood meeting. People are annoyed about the same things. People are sharing hacks. Some want a manifesto. Others just want usable tools. There’s friction. There’s help. There’s curiosity.

If you want the meat, go read the original posts. They have the screenshots, the code samples, the datasheet references, and the small rants that make the whole thing feel alive. These little posts feed into a larger picture: design is both craft and conversation. It’s not solved. It’s argued over, and then someone ships a fix and life goes on.

So go on, poke the original pieces. There’s a lot to graze on. And if one of the posts makes you nod and mutter, well, that’s the point. Design is messy, and that’s not bad. It keeps us busy.