The slides and transcript from a great talk by Maggie Appleton, including this perfect description of the vibes we get from large language models:
It feels like they’re either geniuses playing dumb or dumb machines playing genius, but we don’t know which.
A plea to let users do web things on websites. In other words, stop over-complicating everything with buckets of JavaScript.
Honestly, this isn’t wishlist isn’t asking for much, and it’s a damning indictment of “modern” frontend development that we’ve come to this:
- Let me copy text so I can paste it.
- If something navigates like a link, let me do link things.
- …
Chat is rarely a suitable interface for most tasks. Here, Maggie Appleton explores and prototypes some alternatives.
In some ways, the fervor around AI is reminiscent of blockchain hype, which has steadily cooled since its 2021 peak. In almost all cases, blockchain technology serves no purpose but to make software slower, more difficult to fix, and a bigger target for scammers. AI isn’t nearly as frivolous—it has several novel use cases—but many are rightly wary of the resemblance. And there are concerns to be had; AI bears the deceptive appearance of a free lunch and, predictably, has non-obvious downsides that some founders and VCs will insist on learning the hard way.
This is a good level-headed overview of how generative language model tools work.
If something can be reduced to patterns, however elaborate they may be, AI can probably mimic it. That’s what AI does. That’s the whole story.
There’s very practical advice on deciding where and when these tools make sense:
The sweet spot for AI is a context where its choices are limited, transparent, and safe. We should be giving it an API, not an output box.
Of course, users can learn over time what prompts work well and which don’t, but the burden to learn what works still lies with every single user. When it could instead be baked into the interface.
I like how Luke is using a large language model to make a chat interface for his own content.
This is the exact opposite of how grifters are selling the benefits of machine learning (“Generate copious amounts of new content instantly!”) and instead builds on over twenty years of thoughtful human-made writing.
Here’s an aggregator of components from multiple design systems.
So then the question becomes: how do you most effectively communicate designs, to facilitate the best discussions about those designs? My answer is: lots of little prototypes built with HTML, CSS, and JavaScript.
It doesn’t bother me much that bleeding-edge ML technology sometimes gets things wrong. It bothers me a lot when it gives no warnings, cites no sources, and provides no confidence interval.
Yes! Like I said:
Expose the wires. Show the workings-out.
At Clarity last week, I had the great pleasure of introducing and interviewing Linda Dong who spoke about Apple’s Human Interface Guidelines. I loved the way she looked at the history of the HIG from 1977 onwards. This collection of videos is just what I need to keep spelunking into the interfaces of the past:
A curated collection of HCI demo videos produced during the golden age from 1983-2002.
A good ol’ rant by Vasilis on our design tools for the web.
A terrific piece by Maggie Appleton that starts with a comparison of graphical user interfaces and command line tools—which reminds me of the trade-offs between seamless and seamful design—and then moves into a proposed paradigm for declarative design tools:
Small, scoped areas within a graphical interface that allow users to read and write simple programmes
Folk creations fill a gap. They solve problems for individuals and small communities in a way that that centralised, top-down, industrial creations never can. They are informal, distributed practices that emerge from real world contexts. Contexts where individuals have little or no control over the “official” means of production – of furniture, urban architecture, crockery, artwork, media stories, or taxonomies. In response people develop their own unpolished, unofficial, and deeply practical creations.
Now apply that to software:
Only professional programmers and designers get to decide what buttons go on the interface, what features get prioritised, and what affordances users have access to. Subverting that dynamic is the only way people can get their needs met with the computational tools they have at hand.
A stylesheet for when you’re nostalgic for the old Mac OS.
We noticed a trend: students who pick a UI framework like Bootstrap or Material UI get off the ground quickly and make rapid progress in the first few days. But as time goes on, they get bogged down. The daylight grows between what they need, and what the component library provides. And they wind up spending so much time trying to bend the components into the right shape.
I remember one student spent a whole afternoon trying to modify the masthead from a CSS framework to support their navigation. In the end, they decided to scrap the third-party component, and they built an alternative themselves in 10 minutes.
This tracks with my experience. These kinds of frameworks don’t save time; they defer it.
The one situation where that works well, as Josh also points out, is prototyping.
If the goal is to quickly get something up and running, and you don’t need the UI to be 100% professional, I do think it can be a bit of a time-saver to quickly drop in a bunch of third-party components.
The only person who wants a carousel on your site is you. That’s it. It’s a self-serving vanity project so that you can showcase all of your babies at the same time without telling the world which one is your favorite.
Running up against a paper cut bug feels a little bit like getting a physical one: not the end of the world, but certainly unpleasant. These types of tiny annoyances accrete over time, especially when no one is paying attention to them. In a single day of using my phone, I encounter dozens of these minor bugs that each annoy me just a little bit, making the task I’m trying to accomplish just a little bit more complicated.
If you haven’t seen it yet, the new redesign of WebPageTest is lovely!
A wonderful bit of spelunking into the annals of software interfaces by Elise Blanchard.
What about a scarf or collar so the back of your neck prickles when somebody is talking about you on Twitter.
Or a ghost detector for homes, restaurants, etc that glows when someone is “visiting” in Google Maps/Facebook Pages or looking through a webcam? Maybe it would be better to control the air conditioning to produce a chill, or play barely audible infrasound, indications that there is a haunting in progress and the veil here is thin.