yosito

Location: Nomadic. US and EU (Hungarian) citizen, no sponsorship needed in either. Remote: Yes, 10+ years with async teams across time zones Willing to relocate: Yes, with visa sponsorship to Australia. Also open to on-site in the EU. Technologies: TypeScript, React, React Native, Next.js, Vue, Node.js, Python, modern CSS, d3.js, design systems, a11y, i18n, CI/CD, automated testing Résumé/CV: https://josprague.com Email: [email protected]

Senior/staff front-end engineer, 21 years of experience. Recently led front end on the new Firefox.com for Mozilla: 70+ component design system, 19 locales, Lighthouse accessibility 100. Before that, built climate data visualizations for Climate Central used 10M+ times by US media, and cut NASA Worldview's release cycle from weeks to minutes.

Now focused on AI-ready design systems that both developers and AI agents can use. Looking for senior/staff front-end or design systems roles, contract (via my US LLC) or full-time, ideally in climate, civic tech, open source, education or other mission-driven work.

My approach is more "ship fast, iterate later". Your criticism of AI slop is valid and I share it. I've manually re-written the post today, no more AI slop. Of course, that came with the opportunity cost of not spending time on 100 other projects competing for my attention. AI has enabled me to do a lot more with my limited resource of attention.

You're partially correct, in the sense that agentic software development means that exact tech stack choices no longer restrict what's possible or easy to quickly achieve. But having well designed technologies and design systems is actually more important than ever, because AI agents can use those systems to build UIs on the fly more efficiently, with less tokens, and achieving more consistency and reliability. If you want to build an AI system that creates UI for you on the fly, an essential component is a good design system. Take a look at how Claude Design works: if you want to create a design there, one of the first things it encourages you to do is to build a design system, and underneath, that feature has been programmed to use specific frameworks that are also carefully designed and built to make agentic software development more efficient.

Thanks for your kind words. Luckily, I thrive on criticism of my work, and don't take negativity personally. Working with the team at Mozilla on this truly felt really great!

Your frustrations with the state of CSS 2.0 at the time were quite common, and part of what CSS Zen Garden aimed to inspire people to transcend. Now, CSS has evolved to the point where none of these hacks are necessary. You can now write pure, semantic HTML, and pure CSS, and implement full-fidelity designs that work across all modern browsers. That was the whole dream of the CSS Zen Garden, which wasn't fully realized at the time, but now it is.

I've unslopped it now. Thanks for the feedback.

I used AI to get the post out quickly, but I've gone through and manually re-written it today.

I appreciate the feedback. I did use AI to help me put this post together quickly. I've gone through today and manually re-written it. No more AI slop.

Absolutely. When I first started building websites in the early 2000s, I was building site networks, where individuals would want their own themable websites quickly spun up. They all used the same backend system and components, each with unique designs and custom stylesheets. This pattern is incredibly common, and a lot of themable website building platforms are built on it.

Contributions are welcome :) https://github.com/mozmeao/springfield

Yes, not only this, but in my experience, several older browsers actually crash or freeze with too many @import statements in the CSS. While those browsers weren't necessarily a priority to support with modern styles, they styles they recieved still needed to be minimally accessible, so crashing the browser was a deal breaker.

> The CSS Zen Garden worked because everyone was using a single markup file.

CSS Zen Garden was a proof of concept built around a single markup file. But if you extend that concept to hundreds of dynamic pages and a well built CSS design system, that's where the real power of CSS shows.

> In real life, it doesn't really work.

In real life it works incredibly well, which is what CSS Zen Garden originally showed, and what we've proven by applying these concepts to Firefox.

> If anything is out of place in the markup, the properties may not apply as intended. Similarly, if the CSS doesn't account for DOM structures, then you need modify rules or add new ones.

When stylesheets are written with an understanding of the Cascade and specificity, this isn't the case at all.

Personally, I love the current Firefox.com design and see the simplicity as a feature. And it's designed to be responsive to any screen size. A design like this would have been incredibly challenging to build with CSS 2.0 in 2008. That's the point of this post. But if you find the design boring, just like the CSS Zen Garden, you can design your own user stylesheet to re-style the whole site, hundreds of pages in dozens of languages, into any design you want, with one stylesheet, because the styles are cleanly separated from the markup. Have fun!

This is a reasonable reaction. Thank you for pointing that out. I've updated the post with some thoughts on this. Basically, think the radically different styles part of the CSS Zen Garden was only how the dream was shown, but it wasn't the substance of the dream itself. The substance of the dream was to show the power of CSS, and what was possible with separation of concerns and browser support. And that's what I think we showed with what we build for Firefox.

I agree completely. Personally, Tailwind never really resonated with me, though I've used it several times, and we did even briefly consider it for this project. I always felt that it was an approach created by people who never understood the paradigm of CSS, and didn't want to. Since my career started with CSS, I've always been more of a purist.

I appreciate the feedback. I did use AI to help me put this post together quickly. I've gone through today and manually re-written it. No more AI slop.

Correct, it's not part of Protocol. It's in this repo: https://github.com/mozmeao/springfield

Mostly, I have other, more important stuff going in my life. I did write a lot of this post myself, but used AI to flesh out some of the details because otherwise between my perfectionism and tendency to overthink blog posts, I would have never gotten around to publishing this.

Yes, you'll find it in this repo: https://github.com/mozmeao/springfield

Search for "flare" if you're looking for the CSS.

Yes! Thank you. I linked to the site from my post as well. This is a great resource for anyone currently learning CSS.

In my extensive travel experience, more than half the time the boarding time listed at the gate isn't even correct.

"Anthropic had built its brand around promoting AI safety, emphasizing red lines it said it wouldn’t cross. Its usage guidelines contain strict limitations that prohibit Claude from facilitating violence, developing or designing weapons, or conducting mass surveillance."

I can't say that I fully trust this at face value, but I will say, at least at face value, that this commitment to non-violence is something I wish more tech companies in history had made. Whether it's an authentic commitment or just PR remains to be fully seen.

You're right. And it's also important to be mindful that the LLMs can also translate between human intent and formal queries incorrectly, so they still shouldn't be fully trusted even when integrated with a more deterministic system.

Yeah, really dumb move on the part of Chase bank. They'd previously marketed their accounts as being geared towards international travelers, but now their cards can't be used in much of the world.

US "freedom" is more propaganda than reality, and that's becoming more and more apparent.

I'm very skeptical of using AI in this way. I've given Claude access to calendars and travel plans and asked it to do similar analytical tasks cross referencing documents that would take days for me to do manually. Since it was about my own plans and life that I knew well, it was possible for me to spot subtle errors that seemed correct at the surface level but actually weren't the conclusions I would make. I've attempted these types of tasks 10-20 times with similar experiences each time. In the end, it's made me very skeptical, like your wife. I don't trust any AI output without a thorough review. Hallucinations are still a frequent problem.

TLDR: You can invite people to do stuff.

It would be really interesting if we find out that a simple 10 minute daily massage of the lymph nodes in the neck significantly prevents Alzheimer's.