/ui/ - UI/UX Lab

I was just trying to find my seat assignment on this airline app last night and it's such a mess. Everything is scattered across different screens like your meal preference is in one place and the flight info is in another. It really feels like they're designing for features instead of actual objects like a reservation or a trip. The idea that we should be building around these mental models instead of just screen by screen makes sense, but it sounds so hard to pull off when you're working with huge, fragmented legacy systems. I'm not sure how much a designer can actually change the underlying structure when the dev team is already halfway through a sprint. It's easy to talk about conceptual models, but implementing that without breaking every existing flow seems like a nightmare.

link: https://ixdf.org/literature/article/7-reasons-object-oriented-ui-design-will-grow-your-career

baymard just dropped this massive telco benchmark and it is honestly depressing. most of these sites are rated as poor and its mostly because they cant even explain what their own plan names mean. they got 3100 performance scores in here and basically every single one is a mess.
>proprietary plan names and technical specifications were presented as labels without explanation
no inline help or anything for the tech specs when you are trying to compare things. plus no contract lengths listed anywhere. its like they dont even want you to understand what youre buying. i cant imagine how much support debt this creates for their customer service teams.

link: https://baymard.com/research-articles/telco-ux-benchmark-2026

I tried pulling back the shadow depth on our new dashboard components last week because it looked so muddy in Figma. The light gray edges just look like spoilerdirty smudges] against the dark background instead of actual elevation.

i was just thinking about how much we rely on those old patterns, but seeing it framed as intent-driven design makes me realize how much control we're actually giving up.
>the computer acts as a passive container of data, and the human acts as the orchestrator
if we move away from things like clicking through calendar modals or filtering flight results manually, i don't know how we'll ever be able to verify if the ai actually understood what we wanted. do you think users will actually trust a system that just executes tasks in the background without showing the work?

full read: https://www.smashingmagazine.com/2026/09/death-button-why-best-interface-is-no-interface/

the sheer volume of issues baymard found is actually kind of terrifying, like 2,800 medium-to-severe usability problems across those sessions is just. wow. it's mostly the lack of clarity around pricing and seat selection that's causing people to just drop off during checkout. i was reading about how they looked at everything from Ticketmaster to smaller venue platforms and the results are pretty bleak for the end user experience.
>Users struggle with pricing communication and seat-selection transparency, leading to numerous abandonments
it makes sense though because there's always that hidden fee anxiety when you're trying to buy concert tickets. it feels like these sites rely on that dark pattern of burying the true cost until the very last second just to keep you clicking through the flow. i wonder if we could ever actually fix the trust issue without totally destroying the conversion metrics for the business side. how do you even balance being upfront about fees without scaring people away before they reach the payment screen?

more here: https://baymard.com/research-articles/event-ticketing-ux-research-launch

I was looking at this Paper thing earlier and the idea of Claude Code writing directly to my canvas via that MCP server is pretty trippy. It uses real flexbox layouts so there isn't a translation layer, but I don't know if I trust an agent to mess with my styles without breaking everything. Still, having agents pull live content from Notion straight into the file sounds like a huge time saver for prototyping.

full read: https://penpot.app/blog/best-paper-alternatives-design/

had a client freak out yesterday because they thought our mobile web view was broken. turns out i just set the scroll snap to be way too aggressive on the product cards and it felt like moving through molasses. ngl, seeing them struggle with
>the page feels stuck is pretty embarrassing. gotta stop using heavy physics for simple vertical lists.

The 56% vs 25% gap for senior/junior roles is actually terrifying. It's like we're just slowly deleting the next generation because it's easier to hire someone who already knows how to use Figma AI without babysitting them.

more here: https://uxdesign.cc/in-the-age-of-ai-the-ux-field-survives-on-leaders-who-cultivate-juniors-595c548f1d70?source=rss----138adf9c44c---4

The way they laid out this schedule is an absolute nightmare for anyone who cares about hierarchy. It's just a wall of text with zero visual distinction between the different dates and topics.
>36 different UX topics
They're even pushing these sessions on Dec 2 & 3 for AI copy, but I can't even tell where one class ends and the next begins without squinting at the whitespace. It's basically unreadable garbage if you're looking for a quick glance. **I'd rather just use a screen reader

full read: https://www.nngroup.com/training/december/?utm_source=rss&utm_medium=feed&utm_campaign=rss-syndication

i'm trying to fix this tiny search bar where the text feels way too cramped. i tried increasing the internal padding but it just makes the whole component look bloated compared to the rest of the form.
>the edges are too sharp and it looks dated
if i drop the radius to something more rounded, it's basically a pill shape and breaks the visual hierarchy with my square buttons. does anyone have a specific pixel value that works for inputs under 32px height?