Post

How broken is today's frontend culture? A friend points out that the median *mobile* page is now larger than a copy of DOOM (2.6 MiB vs. 2.48 MiB), the 75th percentile page is more than 2 DOOMs, and the P90 mobile page is 4.5x the size of DOOM: httparchive.org/reports/page...

The people that need to engage with this type of post most bounce off of headlines like this. I love these blog posts, but the typical defensive reactions are so common at this point that I wonder how we can better get people to rethink their positions here?

e.g. literally in this same thread, commenter points out that pages / features are more complex. They're right (to a degree), and yet miss the point that the end-result is still a poor user experience at the margins. I've been struggling with this comms issue now. Do you have any thoughts?

My approach has been to lean into the data, which works for some, but in much of the discourse polarizes further and makes people *more* defensive/look for strawman arguments. I think the comms issue is more nuanced than it might appear prima facie.

FWIW I don't have answers here, just observations as someone else closely monitoring the narrative around frameworkism vs performance, etc.

My goal is to seed the ecosystem with the information we need to understand the scale of the problem because solutions will likely be structural, not individual. Preaching regulation, not recycling bins, so it's largely immaterial if individuals bounce off it: infrequently.org/2025/11/perf...

The information is incredible, as always, thanks. I agree solutions need to be structural, though is there not room for both? I've been involved in activism of some form for 15 years and there's usually a need for both individual (e.g. grassroots) and structural (regulatory) movement.

To be sure, that doesn't mean *you* need to do both. I love the info, as I said! I'm more trying to point out that I see a lot of grumpy people on both sides talking past each other, hence less polarization and adding clarity to the issues may be beneficial.

(Reading this back I realize I sound a bit preachy/overly academic, sorry, not my intent.)

To be honest, I feel what had the biggest real-world impact here is Astro, which managed to position and market itself as a sort of Next.js replacement for the same crowd, kinda sneaking in MPAs through the backdoor into the mainstream discourse.

I agree Astro is a blessing. I actually think there's a lot happening here for people who are tuned into the space. Astro and Marko 6 come to mind, but also lots of interest outside the immediate JS-ecosystem (Lustre, 11ty, etc) + more focus on progressive enhancement (Svelte, Enhance, etc).

Hopefully there's enough of a butterfly effect causing people to ask questions about how they build web experiences.

Definitely a lot happening. But not sure what it will take to shift what a lot of devs think of the default/boring stack from a React SPA to something more sensible. Even Lustre has on their Quickstart guide: "This guide will focus on the most common use case: SPAs"

Right agreed, that's where my questions elsewhere in this thread (and my general frustration) is. TBF to Lustre, I think their audience may (?) be more multiplayer, high interactivity apps reliant on the BEAM (similar to Phoenix LiveView). Less so the average "app" that Avg B-corp needs.

But the problem isn't as easy as just showing the data. IME, you (often) have to convince both business and technical leaders to pay both the opportunity costs and migration/rewrite costs of moving away from the React-centric culture. Even if you can manage this, there will be pushback.

And this pushback happens (again IME), even if you can *prove* the costs are worth it / business critical, both from a technical perspective (simpler, better tools, faster builds, whatever) and a business one (revenue, dev cycles, opportunity cost rebuilding/training).

That's not to say it's not worth pursuing, quite the opposite. My point is that there's complexity here, and I hope we can find better ways to communicate this complexity productively without continuing to see the web "lose" due to the herd mentality around mainstream React/Vercel.