A clean start for the web
The web is in need of some reinvention right now.
The web’s evolution over the last decade has mirrored the American economy. All of the essential indicators are going “up and to the right,” a steady stream of fundamental advances reassure us that there “is progress,” but the actual experience and effects for individuals stagnates or regresses.
The crisis affects platforms, creators, and consumers alike.
I’m going to try and dissect and diagnose this situation, a bit. You can skip forward if you just want to read my casual, unprofessional pitch for a reboot of the web. The idea is that we could choose a new lightweight markdown format to replace HTML & CSS, split the web into documents and applications, and find performance, accessibility, and fun again.
The platform collapse
The platform side is what changed last week, when Mozilla laid off 250 employees and indicated that it would affect Firefox development. Firefox wasn’t the #2 browser - that’s Safari, mainly because of the captive audience of iPhone and iPad users. But it was the most popular browser that people chose to use.
Chart from statcounter
This is a textbook monoculture. In one sense, it’s a victory for collaboration because nobody’s ‘wasting time’ on competing implementations and web developers can expect the same features and bugs across different browsers. But in a deeper way, it threatens one of the basic principles of how the web has evolved.
Specs & implementations
The web has evolved through a combination of specifications and implementations. Organizations like the WHATWG, W3C, and IETF have been collaboration spaces for independent developers, corporations, and academics to discuss potential new features of the web. Then, browsers would test those ideas out in a variety of implementations.
This was an interesting structural piece: it reassured us all that it was possible to follow along, and that a multi-participant web was one of our goals. It was frustrating to pull up caniuse and see blank spots, but the idea was that different browsers may take the lead in some areas, but everyone catches up eventually. Chrome was not always the first to jump on features, or the first to optimize.
It’s slower to collaborate than to work alone, but it was beneficial in some ways that we’ve lost now. Chrome has been moving extremely fast, adding new specifications and ideas at a startling rate, and it’s becoming one of the hardest pieces of software to replicate.
Mike Healy I think said it best:
Do you think the web has almost ‘priced itself out of the market’ in terms of complexity if only 1-2 organisations are capable of building rendering engines for it?
Not only is it nearly impossible to build a new browser from scratch, once you have one the ongoing cost of keeping up with standards requires a full team of experts. Read Drew DeVault’s Web browsers need to stop for that point, and keep reading all of Drew’s stuff.
What about Flow?Yep, there’s a browser called Flow, which may exist and may support a full range of web standards. If it does exist, I’ll be very excited about it, but it has been teased for almost a year now without any concrete evidence, so it could equally be vaporware.
The problem for creators
The web has gotten much harder to develop for.
The web has had about 25 years to grow, few opportunities to shrink, and is now surrounded by an extremely short-sighted culture that is an outgrowth of economic and career short-termism. There are lots of ways to do anything, and some of the most popular ways of building applications on the web are - in my opinion - usually ghoulish overkill.
For folks who just want to create a web page, who don’t want to enter an industry, there’s a baffling array of techniques, but all the simplest, probably-best ones are stigmatized. It’s easier to stumble into building your resume in React with GraphQL than it is to type some HTML in Notepad.
The problem for consumers
We hope that all this innovation is for the user, but often it isn’t. Modern websites seem to be as large, slow, and buggy as they’ve ever been. Our computers are barely getting faster and our internet connection speeds are stagnating (don’t even try to mention 5G). Webpage size growth is outpacing it all.
I don’t want to lay all of the blame at those web developers, though. Here’s a story from an old job that I find kind of funny. We were collecting some data from user interactions to answer simple questions like “do people click to upload or do they drag & drop?” So we enabled Segment, a tool that lets you add data-collection pipelines by including a single script. The problem, though, is that Segment offered a big page of on/off switches with hundreds of data providers & ad-tech companies on it. And, sure, enough, some folks closer to the business side started clicking all those buttons.
See, the problem with ads and data tracking is that you can, and who is going to say no? (In that instance, I said no, and added a CSP that would block new advertiser access at the page level.)
You cannot get a simple system by adding simplicity to a complex system. - Richard O’Keefe
Where do we go from here? Some of the smartest folks out there have been advocating for a major version revision of the web.
I am in no way qualified to speculate on a whole new web from scratch, but the air quality is scary so I’m skipping my run and it’s Saturday morning so here we are.
How do we make the web fun, participatory, and good?
My first thought is that there are two webs:
The document web
There is the “document web”, like blogs, news, Wikipedia, Twitter, Facebook. This is basically the original vision of the web, as far as I can understand it (I was 2). Basically CSS, which we now think of as a way for designers to add brand identity and tweak pixel-perfect details, was instead mostly a way of making plain documents readable and letting the readers of those documents customize how they looked. This attribute actually survived for a while in Chrome, in the form of user stylesheets, and still works in Firefox. Though it’s going to be a rough ride in the current web which has basically thrown away semantic HTML as an idea.
The “application” web
Then there’s the “application web”. This started as server applications, built with things like Django and Ruby on Rails and before them a variety of technologies that will live forever in corporations, like Java Servlets.
Backbone.js demonstrated that a lot of these applications could be moved into the browser, and then React and its many SPA-style competitors established a new order for the web – highly-interactive, quite complex, client-side applications.
The war between the parts of the web
I posit that this dual-nature is part of what gives the web its magic. But it’s also a destructive force.
The magic is that a simple blog can be creative expression, can be beautifully interactive. This one isn’t, but I’m just saying - it’s possible.
When I read blog posts from ‘traditional web developers’ who are mad that HTML & CSS aren’t enough anymore and that everything is complicated – I think this is largely that the application stack for building websites has replaced the document stack in a lot of places. Where we would use Jekyll or server-side rendering, we now use React or Vue.js. There are advantages to that, but for a lot of minimally-interactive websites, it’s throwing away decades worth of knowledge in exchange for certain performance perks that might not even matter.
The appeal of social networks
The appeal of social networks is partly because they let us create documents without thinking about web technology, and they provide guarantees around performance, accessibility, and polish that otherwise would take up our time. You don’t have to think about whether your last Facebook post will load quickly on your friend’s phone or whether your Instagram post will be correctly cropped and resized in the timeline - those things are taken care of.
To some extent, this doesn’t need to be something that only social networks provide, though: standards like RSS and services like Instapaper show that pleasing formatting and distribution can be done at the platform level and be provided on top of existing vanilla websites.
Document web 2.0
A unified theory of a new web that had just enough application characteristics and enough document characteristics to provide the sorts of hybrid interactive documents that we see today - now that would be cool. But the path to a splintered web is clearer and is what I’m thinking of first, so here’s some of that.
- Rule #1 is don’t make a subset. If the replacement for the web is just whatever features were in Firefox 10 years ago, it’s not going to be a compelling vision.
- Rule #2 is don’t make it compatible. If the replacement web lives alongside, undifferentiated from the current web, then you’ll never actually reduce complexity because replacement web browsers will still support everything, and people won’t be encouraged to leave the old web.
- Rule #3 is make it better for everyone. There should be a perk for everyone in the ecosystem: people making pages, people reading them, and people making the technology for them to be readable.
Okay, so let’s say we’re creating a new document web.
First, you need a minimal, standardized markup language for sending documents around. You might want to start with a lightweight markup language, which will ironically be geared toward generating HTML. Markdown’s strict specified variation, Commonmark, seems like a pretty decent choice. That’s the language I’ve written all my blog posts in, and the most popular language in its family. There are lots of great parsers and a big ecosystem of tools for Markdown.
Then, you need a browser. Mozilla has been working on a brand new browser for a while - Servo. That team got laid off last week, which sucks. That project includes standalone Rust crates for font rendering, and there’s a world-class Rust Markdown implementation, and a growing set of amazing application frameworks. Could you build a pure-Markdown-browsing browser that goes straight through this pipeline? Maybe?
I think this combination would bring speed back, in a huge way. You could get a page on the screen in a fraction of the time of the web. The memory consumption could be tiny. It would be incredibly accessible, by default. You could make great-looking default stylesheets and share alternative user stylesheets. With dramatically limited scope, you could port it to all kinds of devices.
And, maybe most importantly, what would website editing tools look like? They could be way simpler.
What could aggregation look like? If web pages were more like documents than applications, we wouldn’t need RSS - websites would have an index that points to documents and a ‘reader’ could aggregate actual webpages by default.
Application web 2.0
I feel like every time I mention something about the web, the automatic response is that WebAssembly might fix it. Maybe?
I don’t know. WebAssembly is pretty great, but should web applications just be rendered to a canvas, and every application brings its own graphics toolkit? Do we really want anti-aliasing differences between web applications? Applications-in-containers is a thing - look at Qubes - but it’s not really something that users should want. Anyone who has used Blender or Inkscape on a mac has some idea of how this goes.
Or is WebAssembly the new ‘core’ and we still render UIs with HTML? Or… create a shared linked library that WebAssembly apps can use that works roughly like SwiftUI, offering application-friendly layout conventions like constraints instead of document-centric ideas like line heights and floats?
The problem with imagining the application web is that it’s pretty expansive.
The worse the ‘Mac App Store’ and ‘Windows App Store’ and ’App Store’ and ’Play Store’ get, the bigger a cut those monopolies demand, the more it costs to be a Mac or Windows developer, the more that applications get pushed to the web. Sure, some applications are better on the web. But a lot are just there because it’s the only place left where you can easily, cheaply, and freely share or sell a product.
There was a time when we could install applications, give some sort of explicit agreement that something would run on our computers and use our hardware. That time is ending, and web pages now have rather complex ways of getting at everything from webcams to files, game controllers, audio synthesis, cryptography, and everything else that was once the domain of
.apps. This is empowering, sure, but is quite an unusual situation.
Who’s working on this?
- Beaker Browser is partly a reinvention of the internet – it’s the simplest way to use dat for decentralization, but they’re also experimenting with new kinds of documents and ways of authoring.
- Project Gemini is a really interesting, distinctly retro-flavored web alternative. (via Jesse)
- I’ve been pretty inspired by taizen, a command-line based Wikipedia browser. It shows how a text-first experience can be really fun.
What do you think?
There are a lot of other ways to look at and solve this problem. I think it is a problem, for everyone except Google. The idea of a web browser being something we can comprehend, of a web page being something that more people can make, feels exciting to me.
The markdown-centric approach feels very doable. I think the clearest rebuttal is that it ‘sucks all the fun out of the web,’ and there’s some truth to that. But the early web wasn’t fun in many conventional ways - you couldn’t quite create art there, or use it as much more than a way of sharing documents. But it was fun as heck, because sharing is fun and it was simple and flexible in some cool ways. So the key is to discover the small things that unlock the possibilities in this plan, if they’re there. Or find a different plan with ‘just enough fun.’
Social networks are universally more restrictive than web pages but also more fun in significant ways, chief amongst them being that more people can participate. What if the rest of the web have that simplicity and immediacy, but without the centralization? What if we could start over?