Web Development

I’ve been building a website called 100,000,000 and I’ve recently opened it to a small number of people for early testing.

The concept is simple: each participant contributes one photo and one message, and the long-term aim is to build a very large public collection.

From a web-development perspective, one of the more interesting problems has been designing the site so the experience still makes sense if the collection becomes extremely large.

The current site is primarily server-rendered PHP/MariaDB with JavaScript-enhanced interfaces. Some of the things I’ve been working on include:

• separate desktop and mobile Gallery experiences
• bounded Gallery loading rather than attempting to preload the collection
• a mobile wall plus vertically navigable Scroll mode
• a shared image/message Viewer while preserving Gallery state
• responsive image layers so small Gallery tiles don’t request unnecessarily large assets
• search, filtering and direct submission-number lookup
• localisation across 17 interface languages, including RTL layouts
• on-demand message translation
• keeping public Gallery data separate from account/payment/moderation data
• moderation and payment before an entry becomes publicly eligible

I’m deliberately not claiming the current infrastructure has been proven at 100 million records. The approach at the moment is to keep the public interfaces bounded and scale from measured usage rather than trying to render or retrieve enormous datasets client-side.

The site is here: https://100000000.photos/

What I’d particularly appreciate from other web developers is feedback on the actual experience:

• Does the purpose of the site make sense quickly?
• Does the Gallery feel intuitive on desktop/mobile?
• Are there any obvious UX problems or confusing interactions?
• Anything noticeably slow, broken or awkward?
• Are there architectural or frontend decisions you’d question after using it?

It’s still a very early public test, so critical feedback is useful.

An Auckland convenience store is in the midst of an ongoing online battle after inadvertently allowing its website to be edited by the general public.

The store […] found itself advertising, among other things, a “pair of crusty socks”, a “Princess Diana Commemorative Plate”, and “National Parks to the highest bidder”.

The store’s site uses Base44, a “no-code AI platform” that allows users to build web applications, but it has been left open for the public to edit.

At one point, while items were being added for sale, the website’s operator appeared to be arguing with the public, with one item saying “stop hacking us”. An item popped up in response saying “don’t use an AI site builder then.”

Airline websites are slow. 95% of them fail Google’s Core Web Vitals on mobile.

HTML has a <geolocation> element now. While it’s only supported in Chrome’n’friends for now, it’s progressive-enhancement-able. I covered it in our 2026 HTML article and also blogged about its interesting enforced accessibility.

But the most interesting story about this element is: permissions recovery.

Mostly meaning: you said no, but now you wanna say yes.

The buttons are supposed to only be clicked when someone completes a series of tasks, bots clicking the buttons throws off the count. I don’t have my own domain it is nested under the main neocities.org domain.

If you’ve encountered a web platform feature that behaves differently across browsers, a standards-based capability that’s difficult to rely on today, or an area where better interoperability would unlock new opportunities, we want to hear from you.

The proposal submission period is open now and will remain open through September 23, 2026.