Scott Jenson

Scott Jenson

@jenson.org

UX Strategy: Apple System 7, Newton, & Apple HI guidelines. UX Director Symbian, managed Mobile UX at Google, creative director frog design SF. Head of Product for two startups. Google Physical Web + Android UX research. Semi-retired

3.9K followers414 following650 posts

Brad Myers' Pick, Click, Flick is a historical reference of UX interaction, cataloging the evolution of UX over the decades. As a historical record of what has been built, the book is reasonably thorough. While reading through it though, I found myself wanting more reflection from the author on...

View full thread

I was hoping Brad's historical perspective might impart more insight. Ultimately, it is unfair to critique a book for what it never set out to be.

Pick, Click, Flick is a historical index of what has gone before, and it succeeds at that goal. It's just whet my appetite for a follow up book that discusses what we've learned from that history (and what we've forgotten!) helping frame what we should be doing next.

Brad Myers' Pick, Click, Flick is a historical reference of UX interaction, cataloging the evolution of UX over the decades. As a historical record of what has been built, the book is reasonably thorough. While reading through it though, I found myself wanting more reflection from the author on...

why these changes occurred. (e.g. listing every scrollbar is helpful, but a deeper discussion of how they evolved, which ones succeeded, how mobile has turned scrolling on its head, all would have been fascinating) A raw catalog just begs so many "why?" questions.

I worked on the Android team for YEARS and I just learned that you can turn on "Clipboard History" on GBoard. So many useful features are just too hard to find/setup.

I'm now using a local on-device voice to text transcriber on my computer and it is revolutionizing how I input text. It allows me to write much longer passages than I would normally want to by hand. Downside: I have to lock myself in my office as using it in the living room frustrates my wife.

I'm at a bar and this happens. Apparently it's called CNC snowfill engraving. I had to buy another drink just to capture it in action

Now THIS is an elevator panel. Sure it's a bit over the top but it is exceptionally clear and #a11y friendly. Car rental area Minneapolis #ux

Do PWAs have an image problem? I was talking with someone today about using a PWA instead of two native apps and there was some resistance. As you can put PWAs in app stores (either directly or through a wrapper), I'm not seeing the reason to avoid them. Please let me know what I'm missing...

I've now sorted my cables: 1. Legacy 2.0 cables 2. 3.1 cables with higher data and modest power 3. 3.2/4.0 cables that did more data, power and video 4. Thunderbolt cables. (there are even variations within each camp! What I'd like to see is some color coding on the plugs...

Scott Jenson@jenson.org

I have a drawer with a ziplock bag for each type of USB cable. And for years I've had one bag for USB-C to C. I'm finally just giving in, buying a cable tester, and breaking them up into USB-C power only, USB-C power+data, and USB-C video, because this is just getting out of control.

I'm watching the first season* of the British tv show The Avengers. It is shockingly good! I'm only two shows in and I'm hooked. The writing, the characters, and the overall vibe are so much more than I expected * Season 1 was lost, I'm actually starting with season 2

I have a drawer with a ziplock bag for each type of USB cable. And for years I've had one bag for USB-C to C. I'm finally just giving in, buying a cable tester, and breaking them up into USB-C power only, USB-C power+data, and USB-C video, because this is just getting out of control.

I know, I know, maybe it wasn't built for me. But some the choices (like ridiculously small output token limits) just boggle my mind. There needs to be a "JustMakeItGo" extension that adjusts these silly defaults so I can actually, you know, ask it a question and it replies with out crashing.

Is anybody using DFOS? I learned about it a LocalFirst a few weeks ago and I wanted to check it out. It appears to be opaque by design, you need to know a bit about the group you want to join first. Are there any DFOS public spaces that anyone would recommend? (Assuming they are open of course)

I've found LLM tech to be useful in limited, controlled conditions. Its current models are hard to support for ethical/ environmental reasons. However, I want to keep exploring how best to use it, and of course, when NOT to use it. I'm not a freaking fanboi here. This tech has serious limits!

As open and local models improve, I'll transition over. IMHO, it's inevitable that I'll eventually be 100% local, but it might take a few years. This setup allows me to keep working with and learning the pros and cons of the tech while steadily reducing my reliance on frontier models.

Scott Jenson reposted

Nothing makes you less of an expert than "believing it's possible" because that clearly tells me you don't know how AI works or how consciousness works or how philosophy works or how ethics works.

As this is getting some traction, I want to be clear that I still believe in PMs, programmers, and most importantly, the UX process. I'm not trying to erase the lines between roles or avoid UX research. I'm just pleasantly shocked by this ability to iterate so quickly,

Scott Jenson@jenson.org

There are LOTS of reasons to be cautious about using foundational AI models to code. But for a UX designer, having the ability to generate frontend code is mind-blowing. I get to explore EXACTLY what I want in seconds, not days. No debating feasibility, just pure, instant iteration.

There are LOTS of reasons to be cautious about using foundational AI models to code. But for a UX designer, having the ability to generate frontend code is mind-blowing. I get to explore EXACTLY what I want in seconds, not days. No debating feasibility, just pure, instant iteration.

View full thread

This is literally 100x faster. Of course, I still need my engineering partners for actual production code. It's possible to write bad code this way! But having the power to explore in my own little corner with real, functioning code is a shift I didn't expect to be this powerful.

Alas, the data privacy and IP issues with the foundational models are a huge roadblock I can't ignore. We desperately need viable open and local versions to make this a viable alternative. I may have to wait a bit for this to happen, but it feels like the only way forward.

There are LOTS of reasons to be cautious about using foundational AI models to code. But for a UX designer, having the ability to generate frontend code is mind-blowing. I get to explore EXACTLY what I want in seconds, not days. No debating feasibility, just pure, instant iteration.

As a UX designer who has worked with programmers for 40+ years, I can't overstate how liberating this is. The speed is head-snapping. Because the learnings come so much faster, the quality of the final UX is more refined. I’m not being anti-programmer here! This is strictly about the speed.

I've been using Fable this past week on the "promo tier" at just $20/month. It's very powerful, with clear value. I just used ccusage to get my token count this past week and at standard rates, my little side project would have cost roughly $100/day! Powerful but a BIT too expensive

The biggest shock was my input/output tokens were a reasonable 0.1M vs 2.4M. What was massive was my cached token usage: 170M! Clearly I can bring that down but: a) It was SO easy to do! b) Claude encourages you to not /compact too often (caveat: rough calculations, could be wrong)

I'm REALLY trying to move off of foundational LLM models for coding. I've heard that Qwen2.5-Coder or CodeGemma are SLMs that are targeted on just coding so are much smaller, and in theory anyone can run on a 16GB Mac. Has anyone tried this? Are there better alternatives?

Hiking on the Curonian Spit in Lithuania near the Russian border. Apparently the Russians jam GPS and my hiking log appears to prove that!

I've been in UX for 40+ years (I started at Apple). Recently, I've been working on bringing UX to FOSS projects, and I see this attitude far too often (but not always!). It's terribly exciting to see this thread get so much support and love. There is hope yet.

if your response to “users aren’t dumb, they have their own motivations and incentives and it’s good to try to understand those when you make software you want people to use” is “well actually, users are dumb” I’m sorry, but you are part of the problem

I'm leaving the #localfirst conference intellectually stuffed. Such a great conference with excellent conversations. I have so many things I want to try when I get home. Thank you @bmann.ca and the entire team for putting on such a great event.

Image from a talk at #localfirst conference. We've all seen cartoons like this. It's clearly AI generated but WHY is it so obvious? You'd think the models could produce something with a little variation so they aren't so stylistically identical?

As I speaker, I would never use an image like this, especially at a tech conference, as it seems likely the audience would be distracted by it and wouldn't focus on my talk

Scott Jenson reposted

Look, I’m an open source maintainer, and if you want people to use that FOSS software as an alternative to that proprietary software it’s good to understand why people wanted to use that proprietary software in the first place and why they might be hesitant to switch It’s not because users are dumb

Arrived at 08:30 Saturday in Berlin for the local first conference (starting tomorrow) I've got the speakers lunch in 3 hours but otherwise my day is wide open. Any suggestions on what I should be checking out here in sunny Berlin?

Excited to be flying into Berlin tomorrow morning to speak at the LocalFirst conference. I'll be doing part 2 of my "Future of desktop UX" talk.

I just composed a two part reply to a post, wording things VERY carefully only to realize that after posting, I needed to edit the second post. Argh, I need to delete and retype, OK... I go to the second post and delete it but there is a bug and the FIRST post is deleted instead!!

I'd only saved the second in the clipboard so.... i just deleted the remaining one and let the whole thing go. Classic example of how a simple #UX bug can completely derail a user.

I have to say, using the 'free period" of Claude's Fable model over the weekend and it really is nailing nearly everything I'm asking it to do, even on the first try. I'm NOT a fan of the ethical/environmental issues surrounding LLMs but you *can't* say they are useless.

Of course, after this trial period is over, it will be very expensive and would make no sense to use it. But I'm hopeful the open models, over time, will catchup and, maybe a bit further out, I could run it locally.

"Oh no! We forgot Timmy!" I know I'm on a tear right now with Automobile #UX but messages like this just have me scratching my head? Yes, I know, baby's have been left in cars but two extremely obvious points: 1. It is a statistically insignificant number 2. This message wouldn't help anyway.