Lightnews — Scholar-powered news

LightNews

Ben Schwarz

@benschwarz.bsky.social

3.2K followers 170 following 720 posts

Founder & Product developer @ calibreapp.com. Into sustainability, online privacy, cycling, photography, music, hiking, native country and creative pursuits 🌞

Posts Media Videos Starter Packs

Pinned

Ben Schwarz @benschwarz.bsky.social · Apr 29

🆕 Core Web Vitals test tool update:

Now you can view up to 40 weeks (~9 months) of metrics for your sites.

Data fetched **fresh daily** from Google directly.

Enter URL, get a free #webperf dashboard 👇
calibreapp.com/tools/core-web-vitals-test

Screenshot of Calibre's free Core web vitals test tool. Largest Contentful Paint metric is displayed for calibreapp.com, observed at 2.13 seconds.

Ben Schwarz @benschwarz.bsky.social · 1h

It’s clear that dynamic sites are harder to make fast and efficient, but I do not agree that we should universally decide to judge performance based on the underlying technology or delivery mechanism. If it’s slow, it’s slow.

Ben Schwarz @benschwarz.bsky.social · 1h

Sure, but if it isn’t hard, why are so many sites presenting with poor CWV?

Also, I’d challenge that Next.js is a lot less popular than you believe it to be. It’s popular in a small circle. Beyond that? Not at all.

Ben Schwarz @benschwarz.bsky.social · 1h

Yep, definitely not comprehensive.

Also, users do not care what framework website is written in. They care about how it performs. That's a fair comparison IMO.

Ben Schwarz @benschwarz.bsky.social · 1h

I get that the HTTP Archive data isn't a clear benchmark and that in some cases comparison is imperfect. I think the most important detail to focus on here is OUTCOMES.

Results show that Next.js sites **do not** perform favorably in the wild for real users. I don't think that should be ignored.

Ben Schwarz @benschwarz.bsky.social · 1h

Not at all, I didn’t feel criticised whatsoever!

I didn’t include nuxt because I’ve never used it, don’t know anyone who has used it, and don’t plan to use it. I’m sure it’s popular, but not in my circles

Ben Schwarz @benschwarz.bsky.social · 5h

I bet you were SO STOKED when you figured out what you could get "for free". Is it available to see somewhere? I'd love to see it

Ben Schwarz @benschwarz.bsky.social · 6h

Wait til you see how it sums two numbers together 🤯

I think you get what I'm saying

Ben Schwarz @benschwarz.bsky.social · 6h

If you want to drive a car that makes more than 70dB of tailpipe exhaust noise, you have to pay a monthly premium levy to the local EPA, and you aren't allowed to drive it between 10pm and 7am.

Zach Weinersmith @zachweinersmith.bsky.social · 13h

Do you have any extremely niche, but serious, ethical stances?

Ben Schwarz @benschwarz.bsky.social · 6h

I see a lot of people discovered the <output> tag for the first time this week.

My question is: why are you excited by it? I really can’t think of a pattern where it would be actively useful? It seems too limited to be valuable.

Ben Schwarz @benschwarz.bsky.social · 8h

When I was still renting, I had rental agents pretending that it was COMPULSORY to use their subscription-like rent payment system.

I reminded them that what they were doing was misleading and not actually the case, and just continued to pay the same way. They would lie to their own mothers.

Reposted by Ben Schwarz

Ben Schwarz @benschwarz.bsky.social · 9h

My take is that either Next.js doesn't make it easy enough to create sites that perform well, or it's just not good technology.

Having used Next.js for several years (along side lots of other frameworks), and being super experienced in this area, I can tell you:

It isn't a good framework.

Ben Schwarz @benschwarz.bsky.social · 9h

Based on what I've seen from the people involved, the founders of Next.js and Bun are two groups that I want nothing to do with. There's plenty of receipts out there if you take a look. These are bad people.

Ben Schwarz @benschwarz.bsky.social · 9h

In this case, websites need to have "enough" traffic to be considered in the dataset.

There's a lot more websites that aren't reported in this dataset, but it's a decent sample size all the same.

Ben Schwarz @benschwarz.bsky.social · 9h

I'm not sure "bias" is the correct term, but IMO it doesn't matter if the sites were optimised or not.

What we can see from the numbers, is that in terms of **outcomes**, Next.js sites perform badly.

Ben Schwarz @benschwarz.bsky.social · 9h

I'm sorry I didn't include a broader set of frameworks in my original post, but it's clear to see there's better (and worse) options out there.

Next.js is popular and that's why I chose to feature it against some popular competitors! I've never used Nuxt, and probably won't either.

Ben Schwarz @benschwarz.bsky.social · 9h

My take is that either Next.js doesn't make it easy enough to create sites that perform well, or it's just not good technology.

Having used Next.js for several years (along side lots of other frameworks), and being super experienced in this area, I can tell you:

It isn't a good framework.

Ben Schwarz @benschwarz.bsky.social · 9h

Yes! It's true, but there is some nuance here: It's about outcomes.

It's possible to build a fast site with Next.js, however, if we look at the scoreboard, we can see in most cases, people don't.

Could be the content type, 3rd parties, audience, knowledge of team or other factors.

Ben Schwarz @benschwarz.bsky.social · 9h

Small numbers are still proof that people can build and launch sites that perform well in the wild with Solid. That's a great endorsement.

Ben Schwarz @benschwarz.bsky.social · 10h

Is the percentage grouped by origin?

Ideally it would be, so that each origin is on a level playing field in terms of its contribution to the % of good user sessions.

Ben Schwarz @benschwarz.bsky.social · 11h

Thanks for sharing all this Barry, I didn’t even have to ask 🤙

While this system of measurement is not perfect, it highlights that when Next.js, React or Lit are present, performance outcomes are POOR.

That could the the type of the site, audience and other factors… but also: numbers don’t lie!

Reposted by Ben Schwarz

Jake Archibald @jakearchibald.com · 23h

I dug into progressive image rendering and found a bunch of common assumptions aren't quite true.

➡️ JPEG & WebP progressive-renders worse in Safari
➡️ AVIF _does_ support progressive
➡️ But JPEG XL in Safari doesn't
➡️ JPEG XL decodes much slower than AVIF
⬇️ and more

jakearchibald.com/2025/present...

The present and potential future of progressive image rendering

Exploring progressive image rendering across JPEG, PNG, WebP, AVIF, and JPEG XL.

jakearchibald.com

Ben Schwarz @benschwarz.bsky.social · 22h

The main issue with next.js is that instead of delivering raw HTML, content is (usually) rendered client side.

Performance suffers because client-side content rendering is slow & more “expensive” than it could be.

TLDR; it’s a slow, inefficient way to render a website and it shows!

Ben Schwarz @benschwarz.bsky.social · 23h

Are you saying “once the SPA has loaded its fast”?

Ben Schwarz @benschwarz.bsky.social · 1d

According to HTTP Archive, Next.js sites see the **worst** Core Web Vitals performance when compared to other popular frameworks

httparchive.org/reports/tech...

Screenshot from https://httparchive.org/reports/techreport/tech?tech=ALL,Next.js,Ruby+on+Rails,Astro,SvelteKit,Svelte,React,VitePress&geo=ALL&rank=ALL&page=1

Ben Schwarz @benschwarz.bsky.social · 2d

Great post Ryan!