Comment by sylware

What's the article is missing: this is directly related to the complexity of file formats and protocols.

There are 2 webs:

- the web site, to serve noscript/basic (x)html, namely basic HTML forms which can be augmented with <video> and <audio> nowadays, namely it serves web _pages_. It was made super modular, you have browsers not handling CSS, and it is fine for _pages_ with a semantic 2D table (implicit navigation even for braille browsers). Web engines there, are more than reasonable to write an alternative of, even a plain CSS renderer (look at netsurf browser), only text (lynx/edbrowse/etc), graphic (links2/elinks/etc). In the end, 'HTML' is not perfect (like CSS), a bit of a mess actually, that's why they tried an XML representation, a failure because it was literaly sabotaged by... "Big Co" or in the web realm, the 'whatng cartel': I remember their web engines were a pain to use xhtml to develop even a simple page... but not with html... curiously. That said, mistakes were made also on the "w3c" side: the 'semantic web', a real abomination of delirious complexity, which I think is what actually made people jump on the "whatng" train, what a disaster. Now, HTML has been back with its weird(shabby?) parsing, but this was kind of 'cleaned up' and much more rigorously defined.

- the web app: the abomination. Basically, only gigantic and insanely complex software can make a web app work (including their SDK), aka only the web engines from Big Co, here 'the whatng cartel'. It is getting worse, it is said web apps are more and more requiring only one web engine to 'properly work' (often gogol blink), and suspicions are very strong at this is made _on purpose_ (I remember the day when gmail.com did disable their noscript/basic (x)html web interface... then POP3 not a long time ago... I guess you all see where this is going). In this realm, there is near ZERO possibility to create a _real-life_ alternative without a bunch a developers laser focused on that for one billion years. I have been wishing for an alternative web engine I could build from source with a simple SDK, does not exist, and even the few attempts here and there are _not_ doing that: they lean towards super complex syntax computer languages (c++ and similar), hence a failure right from the start.

The 'web3'? A lean javascript engine (for instance quickjs, but there are others), with a small set of OS basic abstraction APIs, and a few 'accelerated' specialized APIs (vector drawing, pixel drawing blitter, video decoding, glyph drawing, etc). First problem: nobody will agree an those interfaces (they would have to be as simple as possible), and the 'whatng cartel' will make sure their are useless...

Or a even simpler "HTML" (probably the same with CSS)? "markdown" like the article suggest? Would it have enough expressive power? Again, nobody will agree on the format and will want to make its own.

A good middle ground is to work with a 'subset' of HTML: rought on the edge, but would do a good enough job for nearly all online services out-there, whatever the platform. Nearly 100% of the online services were running on that a few years back, and with <video> and <audio>, it could be even closer than 100% nowadays.

And there is the danger of the 'mobile app only': there, the only way out is to regulate and enforce the availability of a small, stable in time, set of as simple as possible protocols and file formats to allow reasonable efforts at developping an 'app' for an alternative platform (elf/linux, *BSD, fooOS, etc).