HTML and the Web Used to be Something a Child Could Learn, "Modern" Web is a Puzzle of Frameworks, Bloat, and Worse
When the Web was more like Gemini Protocol
Yesterday in Techrights: The Future of the Web is One Rendering Engine or 'Flavours' of Chrome | Best Sites Are Not Optimised for Any Browser, They Work Equally Well With All of Them
"When I was about 14 I learned HTML," Ryan said. "There was a book. It was not difficult. It was just a simple markup language. I thought it was well made. Even a child could easily learn it. I much preferred HTML 3.2. I never wrote a single page of HTML 4. I could do everything I needed to in 3.2 and I could also call up NPAPI plug-ins if I had to, and tell the browser where to load those. I liked NPAPI because you could tell the browser to load this plug-in and run that content in it. It should be simple for people to write web pages. How they look is secondary. I did not ever use style sheets or JavaScript."
I myself did HTML when I was about 15 (Ryan is a tad younger than myself; turns out we were in Disneyworld on the same month). I made my first 'Web site' in Geocities just months before I was in Disneyworld and showed it to my aunt. Back then the Web was relatively new. There were not so many sites, but those which existed were lean by today's standards and worked OK on 486 PCs with just 16 megabytes (not gigabytes) of RAM. For the whole PC, not the Web browser. If there was any JavaScript in pages, it would have to be compact (it was also a relatively new thing), as JavaScript too bloated would freeze PCs.
Ryan recalled: "I developed a strong preference for early versions of Opera. I was amazed that the browser was so small it would technically fit within a DD 3.5" floppy disk yet it had such complete support for web standards. Internet Explorer 5 was over 110 times that big and many web standards were unimplemented, or broken. Opera ran on many platforms. I could even run it under BeOS. IE supported Windows, and certain patch levels of Solaris and HP-UX. The UNIX ports were a complete mess. The release notes of IE 5 UNIX for Solaris say what version of Solaris and which patches it will run on. If you patch Solaris with anything other than those patches, or install a newer Solaris, IE 5 will just segfault. So basically what that meant was if you wanted IE 5 to run on Solaris, you had to install those exact patches, then just let security problems pile up and live with bugs in the OS that had been patched, because once you patched the OS again, IE 5 would segfault. This shows the amazingly shitty design of Microsoft software and the MainWin porting job and the WISE program. It's just incomprehensible why patching the Solaris kernel would cause IE 5 to not run. When does applying kernel patches, minor patches, cause software to not run? I ran Solaris for some time. I never had any stability problems from it. It was quite impressive. Probably crashed less than even Linux did. That's not really surprising when you think about it. Solaris ran on systems that HAD to stay working reliably. Linux has always been sort of quick and dirty and prone to hacks. Usually nowhere near as bad as Windows. But the coding quality standards are sometimes questionable compared to other unix systems."
He spoke of security problems. Well, back then the real security problems did not come from the browser, as opposed to something one could download with a browser. Microsoft would change that with ActiveX controls and other nonsense. The security problems in browsers would become more commonplace over time. Basically, browsers were allowed to do too much to the whole system and, in turn, Web sites one visits (or clicks to land on) would be able to do loads of nasty things.
The solution isn't (or wasn't) to rewrite programs as the security problems were due to bad design.
As Ryan put it: "I don't think Rust is the right way to go about dealing with this. I think that they need to just do code quality audits on existing components. The kernel shouldn't be this chimera of different programming languages either. Then you get people who know one but not the other and can't effectively work outside their area. So far Apple has kept Rust out of Webkit. I use WebkitGTK fairly often and I appreciate that everything in there and Epiphany is using C, GObject stuff, and GTK."
For me, Gemini Protocol works rather well. I get a lot of information and perspectives from Gemini capsules. Clicking on Web sites nowadays results in slowdown, sometimes even a system hanging. If attempts are made to suppress ads, bloat etc. some sites would simply deny access.
In a lot of ways, Gemini Protocol is like the Web of the mid-90s. Yes, images are supported as well, depending on the Gemini client. The security problems aren't quite there because there is very little a capsule can do to your machine. █