The Next Stage of the Web Browser

In his post On compositional window management Alexander Bandukwala explores ways in which window managers could be improved. I can't help but draw parallels to modern web browsers. Applications reimplement the window managers due to the limited feature set. Websites reimplement the web browser in the same way.

I complained in the past, that I want the browser to be just a document reader. But let's face it; it's not. It's really just an SDK. But even for an SDK, it's pretty bad.

Websites keep reimplementing the browser using JS, because the browser sucks. Despite all their fancy advertising and redesigns, browser haven't really evolved for the last decade. Yea, they've implemented a bunch more ridiculous Javascript APIs, but what real added value do they provide to the end-user?

Here's some things I'd want in my browser:

  • You know all those cookie popups everywhere? What if they were just part of the browser? Currently, every website is allowed to save cookies by default. There's nothing preventing a website from saving cookies anyway when you tap the tiny "no" button. What if we reverse that? Whenever a website tries to save cookies, the browser displays a standard popup with a clear "no" button that actually denies the cookies.

  • You know how websites have those complicated menus? Yea, what if my browser just scraped their sitemap.xml and compiled a convient index of the site for me?

  • And while we're at it, why not render a table of contents for every page? You know, those things Wikipedia and numerous other sites are now inserting into their HTML, because our document readers—oh sorry, I meant universal SDKs—still don't have this feature?

  • You know how links you've already visited turn purple (unless the website overrides it). What if we would indicate the target location of the link in a similar way? Links to other pages on the same site would stay as-is, but outgoing links would get some kind of extra indication. Also, it would be nice to know the page size of the target page upfront, especially on metered connections.

  • Also bringing back RSS discovery and subscriptions properly would be pretty nice!

  • Having publishing built into the browser would truly democratize the Web. Instead of the convoluted form-based UIs we could have a simple HTML editor builtin that could publish using HTTP PUT if the target server supports it.

  • Imagine if supporting writers and other creative people was built-in to the Web? A browser-native button to send the author of the page/media element a few bucks would allow creators to passively earn money by creating cool stuff. Maybe we could put HTTP 402 Payment Required to good use?

  • And finally, how about actually useful building blocks for applications, instead of this JS+form madness we have now? The idea of building up the interface using a markup language such as HTML or XML is very good, but what if we could also utilize OS-native components in our designs?

Could somebody at Google or Mozilla maybe pretty please make (some of) this reality? /hj