I'm fed up with it, so I'm writing a browser

So, obviously jaytaph is doing this for his own reasons, which is more than fair.

I've looked into the browser space, and its history, how we got here, and where else we might have gone, and might still get to.

Regarding browser history, there's the question of how we ended up with a single monolithic application which handles the many different elements of Web interaction, from HTML rendering to JS interpretation to identity and security to managing bookmarks and history, to multimedia (video and audio playback), and the often strongly conflicting roles of document reading and discourse versus commerce and transactions versus media playback versus general application support. (I've long thought that those four principle roles might best be split out.)

One of the more interesting (and obscure) early browsers was ViolaWWW, initially released in 1992, which was a graphical browser with scripting and stylesheet support, and introduced or provided an early exemplar of bookmarking, browser history, images, tables, frames, and other features that now define the modern browser.

https://en.wikipedia.org/wiki/ViolaWWW>

https://web.archive.org/web/20160303174535/http://www.viola....> (original seems dead: http://viola.org/>)

I've long wondered why browsers don't offer the option to simply cache and store content locally rather than rely on remote lookups for all access. There are some sites for which this doesn't work especially well, but for many static sites, the ability to browse offline and search locally would be a powerful bonus. Similarly, why isn't it possible to run a search across content from my recent browser history rather than across the entire Web? If I know I'd read something in the past hour, or day, or week, constraining a search to recently-accessed content would be a strong bonus.

(How to deal with content changing over time by saving snapshots and diffs of websites could also be interesting.)

Or why isn't the notion of a discussion --- threaded, flat, sorted, weighted, whatever --- a first-class entity within the Web world? There have been an untold number of reinventions of what was essentially Usenet-plus-something, and the notion that discussions must be fully specified by each individual website (including HN) makes increasingly less sense to me.

I've toyed around with the notion of the Web as a filesystem-addressible space. Plan9OS with its 9P protocol has something like this. The notion is that Internet and Web space could be navigated as a filesystem using both existing and newly-developed shell and other tools to directly access and manipulate objects.

A modular approach to Web tools would mean that we could disaggregate HTML parsing, various proxy services, caching and storage, ad-blocking, privacy indirection (Tor, VPN, or other forwarding mechanisms), interactions (editing and such), among a suite of tools. Some of this now exists, but the hurdles to general use are still fairly high. Parking something like Readability.js directly in the mainstream content parsing flow would be an interesting option.

Working off Readability, I've had a notion that a set of predefined document types, to which user-selected styles could be assigned, would be an interesting change-up to the present Web. A specific URL would specify what its document type was (article, index, gallery, discussion, multimedia, catalogue, application, custom, ...), and the person accessing the site could define what style they prefer for viewing the site. The templates would be fairly tightly defined, but would allow for limited branding elements, think of how pre-printed paper letterhead looked back in the day.

Another Web element that might be interesting to disrupt would be search. If websites themselves could provide indices to their content, crawlers would not be necessary (though audits to guard against keyword stuffing or other deceptive practices would remain). Indices could be distributed in a mechanism similar to DNS. Reputation and ranking would remain issues (though those might be achieved through third-party or federated services). But much of the present infrastructure that's presently required to build search engines, and the entrenched incumbent advantage, would be broken.

I'd love to see a new set of Web architects play with ideas such as this, and am trying to do so myself.

I didn't flag/mod the comment, but I think it errs towards conspiracy theories.

I'm not American, and I understand the American government does evil things sometimes, but it also does a lot of good things for its citizens. It builds roads, keeps the skies clear, funds education, health systems, sues tech giants when they go too far, tries to manage the speculation/money markets, etc. To say "undeniably often working against the best interests of its citizens" is very hyperbolic. Most would probably argue that the government does not do enough, even conservatives (ex: controlling abortion or bathrooms, not that I agree).

The news are also not there to demoralize you, not as a primary objective. Anxiety and negative news do sell ads. I find public radio such as NPR do a wonderful job of balancing point of views, giving the facts, and provide some hope by explaining the underlying issues and possible solutions. Where I live, public radio is the only non-ragebait media that constantly puts pressures on whatever government is in power, and explains various points of views (while admittedly being more centrist editorially, angering both hard-left and hard-right).

I totally agree however with doing whatever it is you are good at, and that's a way to work towards improving the world.

No one said that. At the time there was a massive gaping hole in the OS market begging to be filled. OS vendors were gouging everyone, and GNU/Hurd kept experiencing delays.

If I had to redesign the modern web and ditch the legacy requirements, I’d pick a design with a series of simple docs and vm’s that would look like a squeak/smalltalk environment. Directories and search engines would just be a plugin tool to index what you can retrieve with your permission levels. And session management would allow coexisting environments to run side by side like ide’s