16 ms·

When will web browsers be complete?

144 points by fallat 6y ago

bloody-crow 6y ago
Quick comment on the presentation. This post could very easily be markdown and therefore rendered in a nice readable non-monospaced font. Instead it's a .txt and it's really painful to read. Edit: here's a rendered markdown conversion that is easier on the eyes: https://shrib.com/?v=nc#CapeStarling-6v1L9J https://shrib.com/?v=nc#CapeStarling-6v1L9J
- tleb_ 6y ago
  How is that easier to read? https://wtf.roflcopter.fr/pics/SpRqNBvB/0QUVpbN4.png https://wtf.roflcopter.fr/pics/SpRqNBvB/0QUVpbN4.png
  - bloody-crow 6y ago
    Ugh, it looked better on my screen.
- eCa 6y ago
  I recommend the raw rendering, followed by the browser’s reader mode. But I agree in principle.
fallat 6y ago
OP here (about to sleep): this was just some 2am ramblings that I found could invoke some interesting conversation. Already there are some great conversations. Night!
err4nt 6y ago
W3C does not publish all web standards, notably it doesn't publish the three biggest technologies used in web browsers: - HTML is worked on by WHATWG - CSS is worked on by CSSWG - JavaScript is worked on by TC39 And IE wasn't killed by features moving too fast, Microsoft made Edge browser and still does. Mozilla hasn't been killed by Google or competition, it's been kept alive in large part by Google (I think in some years as much as 90% or more of its total income coming from Google) so they get hundreds of millions—the problem at Mozilla is bad management for the last decade. I'm not sure how well this vision represents reality as it already is today.
- why_only_15 6y ago
  Microsoft makes a browser but the point people are typically making here is that Edge is just a reskin of Chromium.
  - err4nt 6y ago
    Between IE and the current Chromium-based Edge browser, was born, lived, and died another browser engine named EdgeHTML. IE was 2 browser families ago from Microsoft.
    - Gaelan 6y ago
      I thought EdgeHTML was just Trident with some legacy code ripped out. Edit: admittedly, "just" is probably underselling it a bit, but it's still a far cry from a completely new engine.
      - gsnedders 6y ago
        It was very much a fork of Trident, yes. I believe there was significant refactoring done after the legacy code was ripped out.
    - coldtea 6y ago
      Yes, but that was killed because Chrome moves too fast, and maintaining EdgeHTML wasn't easy for MS. Parent said: IE wasn't killed because the web moves too fast -- and used the presence of Edge (or EdgeHTML) as the argument. But both IE and EdgeHTML were killed because Chrome moved too fast for MS to be worth it to maintain its own engine...
- gsnedders 6y ago
  > - CSS is worked on by CSSWG The CSS WG is a W3C WG, and they're very much published as W3C specifications.
  - err4nt 6y ago
gbh444g 6y ago
"There is not much left for a Web browser to cover" - this reminded me of the famous "physics is mostly complete at this point." Web (browser) is the only cross-platform operating system, even if some standards say otherwise. I'm sure we'll see docker-style virtual web environments, cross-machine abstractions to simulate datacenters for distributed computing (GPU and such), PWA-style webapps bootable from USB sticks and functioning as operating systems, various crazy extensions for the upcoming neuralink-style tech, virtual reality frameworks, holographic displays and so on and so forth. Physics is never complete at any point.
- catmanjan 6y ago
  The fact that we're comparing a type of software to a school of science speaks volumes
  - jupp0r 6y ago
    Is it? You could make the same point about engineering, literature, the arts, sociology, ... you get the idea.
    - tannhaeuser 6y ago
      I'm not getting the idea. The argument, as I understand it, was that a browser should aspire to be a standardized and relatively simple app enabling a large variety of backends to deliver content, not a never-ending "social experiment" to become overwhelmed by a single party extracting almost all value off the web, and becoming the only party with the resources to implement browsers.
      - jupp0r 6y ago
        The notion that the world wide web is "finished" at some point is just flawed. Today we do video conferencing, 3D gaming, biometric authentication, photo editing and all sorts of other things on the web. Browsers will evolve along the lines of users needs and developers innovation. If this stops at some point, they will become the new Gopher and something else will come along to replace them.
rado 6y ago
Wish they were stable first. There is undefined and buggy behavior, e.g. sub pixel scrolling, scroll container padding etc.
tannhaeuser 6y ago
> there is not much more distance to cover [relative to native toolkits] As I see it, the web as originally conceived was "ready" around 2003, and then again around 2009 when CSS media queries became mainstream to adapt to mobile usage. For me, the excessive push for browser APIs as a desktop replacement since then, and the proliferation of frameworks is a generational thing, driven not by technology but by big data, SaaS (prospect of extracting recurring revenue from customers, rather than one-time license fees), and a new generation of developers getting into the game by leaving the status quo behind, solely for their own economical benefit. Google has masterfully played this game, and let the idea of open web standards bend the minds of developers who thought they were clever. Moz partnering with them under WHATWG seemed like a good idea at the time, but just made the concept of a web standard arbitrary and "whatever Chrome does" over time, leading to Moz's demise and that of every other "browser vendor". It'll take some stance and political will to reclaim the web.
- cageface 6y ago
  The alternative is an internet completely locked down in native apps in proprietary and unaccountable app stores. I'll take a bloated web browser over a corporate monoculture any day.
  - konjin 6y ago
    Sorry which is which? Because right now Chrome engine has more control of the web then IE did in it's heyday when Microsoft was ruled to be a monopoly.
    - vntok 6y ago
      Are you aware that Blink, Webkit and V8 are all open source?
      - pjmlp 6y ago
        With most people working out of Apple, Microsoft and Google salaries, good luck getting a bunch of random weekend coders to pick up and counterattack an alternative browser based on that open source.
tutfbhuf 6y ago
Never.
benjaminjosephw 6y ago
I think the disruptive innovation for the mono-culture in browser development will be web platforms that enable meaningful and broadly accessible end-user programming. The established players have no incentive to pursue this objective (since it undermines central ownership of data/experience) but, if executed well, could be an extremely valuable proposition for users. Google's current trajectory won't take us there but maybe others can.
ksec 6y ago
Let's talk about Web Page for a minute, or Web Page with some interactivity and animations. Are there any Web Page, render in Web Browser, both the page itself, and the standard itself, that gives the smoothness of say opening a PDF, or even better, a Native App. Does the Apple Store Web Page for selling iPhone felt as smooth as their Apple Store App on your iOS? Jank ( micro- pause ) are still everywhere. Both the Browser and the WebPage itself. And this isn't Web App, just simple Web Pages. I say web browsers ( and the Web Standards ) is far from complete.
- the_gipsy 6y ago
  I don't remember the app store app being smooth. I do agree about the "janks".
boltefnovor 6y ago
Internet Explorer is finished.
mschuetz 6y ago
Web browsers are just getting started, in my opinion. Browsers already are the most effective way to share content with a massive audience, and WebGPU and WebXR are going to lead to a whole lot of new types of content.
makach 6y ago
It is only when they are no longer needed that you can state that they have been completed. There will be no web browsers in the future! What comes after web browsers? Only exciting times...!
dmitriid 6y ago
The main problem with the current state of web browsers is that it's now impossible for a third party to develop a competing browser from scratch (or nearly from scratch). The already existing set is unimaginably huge, and it's a moving targets as browsers (especially Chrome) is throwing more and more features into it.
- lazyjones 6y ago
  That's on purpose, of course. And everyone who suffers from it, willingly tolerates it.
- tannhaeuser 6y ago
  And WHATWG (and CSS WG) rejects the idea of HTML profiles (defining reasonable subsets) and even versions, or anything forward-compatibility for that matter, with supporters standing by and cheering.
- anonymou2 6y ago
  that's why project Gemini was created: https://gemini.circumlunar.space/ https://gemini.circumlunar.space/
onion2k 6y ago
We will reach feature-parity with our local Operating Systems. The author states this in a way that makes it sound like it's possible, but that betrays an assumption that operating systems aren't moving forwards and innovating as well. The web will always lag behind the new features that come to OSs, and therefore browsers will always need to be changing in order to give web users access to those features.
- enos_feedler 6y ago
  Not only are operating systems moving forward in their own direction, they are picking up the missing pieces that were traditionally reserved for the web. The two most important being URLs for deep linking between apps (universal links) and installing/fetching without redirect to a download page (app clips). At some point we will cut out the store publishing step for certain classes of same apps running in safe runtimes (wasm with some native bindings with safe data management). Then what is the difference? You might say compatibility across platforms. But compatibility really is broken in browsers today. So....
nikitaga 6y ago
Chrome OS being open source is not a threat to Google for several reasons: The usefulness of the browser is that it's a cross-platform platform. You can run the same web apps on Windows, Mac, Linux, even mobile devices. Turning the browser itself into an OS-level platform will just create another platform that will need to compete with all existing platforms while having nothing unique to offer – after all, the same web apps run just fine on existing platforms, by their very design. What does ChromeOS have that MacOS or Linux do not have and can't have? You could argue that someone could build a competitor to iOS and Android using ChromeOS, but why wouldn't the same someone build a competitor using... Android? It's open source too. Because if they take Android source without Google's blessing, they would be left without Google maps, docs, and gmail. But Google can just as easily exclude mobile browsers from accessing their web services. In today's world it would be perfectly fine for Google to only offer their services through native apps on mobile, and it's up to them which platforms they choose to publish those apps on. Oh and then there's Widevine. You need Google's blessing to decrypt that, and even if you get it, every content provider that uses Widevine DRM can block access to their content from your browser for any reason (you can't bypass that). How far will your rebel browser platform go without being able to play e.g. youtube? Currently they only DRM certain types of videos, but what would stop them from flipping the switch and applying DRM to every video that is monetized on youtube, and blocking your browser on the basis that it allows adblockers that effectively prevent funding of those videos? The notion that Google is somehow – anyhow – in a weak position is utterly ridiculous. Their moat is stronger than ever.
- pjmlp 6y ago
  With the Chrome monoculture, the Web is effectively ChromeOS nowadays, specially with the increasing amount of APIs that are Chrome only, which although other browsers refuse to implement really doesn't matter as their market share keeps decreasing towards zero.
  - hesarenu 6y ago
    If chrome advances it does not matter. The lowest common denominator would be safari specifically ios safari. Until it adopts we would have to code to it's standard.
qppo 6y ago
I can't read this plaintext in a mobile browser window so clearly there's some work left to do. Or maybe they've done too much work. Who knows.
- kevsim 6y ago
  Reader mode in safari worked OKish for me.
madrox 6y ago
When this essay started with "As someone who has used the Web since 2007" I immediately felt old, so thanks for that. As someone who has been using the Web since 1994 (and I feel like I was late to the game), I can assure the author their perspective has a lot of recency bias and this isn't something to worry about, in the general sense. Or worry about it, because it already happened a while ago. Take your pick. Local operating systems, which this essay says is the gold standard of feature sets, are not standing still. If you think of how long it will take for browsers to reach current parity, go back that amount of time and see how far the OS has come. But let's consider, for a moment, that browsers catch up. Let's just look at native development in general, since this has had access to all local OS features since before the web. Has that never changed? It's changing all the time. We're finding higher abstractions to work with. We're making it easier. I highly doubt Grid and Flexbox as they exist today will still be how web layout works in 20 years. Google isn't even the first to scare the world. This essay could've been written in 2002...about Internet Explorer. Stop laughing. It held even more dominance than Chrome does today, and was closed source. Somehow we survived...which is good, because VRML sucked. However, even with all that aside, people have been rolling their own browsers for years. People fork chromium today. The dark web exists. Maybe in 20 years lots of people will be streaming everything from a private cloud supercomputer vis-a-vis Stadia. The internet is no more homogeneous than all the people of the world. Everyone is finding new ways to access it all the time. This is why open standards are more important than browsers. It means anyone can build their own at any time based on the documentation and it can all read the same documents. So I wouldn't worry about this coming to pass in quite the way you think. Though running your browser in a VM isn't the worst idea.
- cft 6y ago
  In 20 years it will be impossible to manually browse to an unapproved "unsafe" URL in a general purpose browser.
  - ktpsns 6y ago
    Well, its hardly possible to manually browse an unapproved unsafe URL even today. Ever tried to browse a site with a self-signed certificate? In most browsers you have to do a lot of extra clicks to get in. But there will always be special purpose browsers. There is "TOR browser" today, and we will have "vintage web browsers" the same way we have CLI/TLI browsers still today (thinking of elinks and lynx), or the same way as there are still FTP clients and Gopher clients.
Hurtak 6y ago
"When will operating systems be complete?" "When will car technology be complete?" ... The answer is they wont be in short term, and they will be made obsolete in the long term.
- yodelshady 6y ago
  Cars don't suffer feature creep. We've optimised them - removed the horse, switched to steel construction, added electrical subsystem - but really there's been no major functional change in cars, ever.
  - dwheeler 6y ago
    Um, what? Our expectations about cars have definitely changed over time. I don't think it's even legal to sell new cars without seatbelts, something that did not exist when cars started. Even low end cars in the US have added features compared to low end cars of the past, such as power windows, power steering, power mirrors, and so on. It's true that adding features is slower in cars, but that is because cars are partially hardware. It takes much longer to add features to hardware, because making copies is not free. But that does not mean they are unchanged.
  - rusticpenn 6y ago
    I am not sure if you are sarcastic.
    - yodelshady 6y ago
      Nope. Cars have advanced, but if they were webtech, we'd be wondering why they've not implemented OrbitalMechanics support yet. OceanMechanics would be old hat.
  - TeMPOraL 6y ago
    ...added electrical subsystem, replaced combustion engine with batteries and electric motors, we're adding various degree of autonomy. In parallel, we've replaced sane controls with frustrating and dangerous touchscreens, we've added countless types of comfort features, we've improved security, we've computerized engine controls, we've DRMed the car - ostensibly to prevent theft, but actually to route more money from aftermarket towards the manufacturers. Etc. Cars do suffer feature creep, and not all of the features are beneficial to their owners. Just like with the web.
marcus_holmes 6y ago
I've seen the same cycle happen repeatedly in tech (there are many cycles, but this is one of them): systems are centralised on a server, then the client gets more powerful and features of the system are pushed to the client. This gets too much and the client slows down, so the features are centralised to a server again. I can easily see a world where Chrome/ium becomes too bloated to work well, and a "new web" of thin, light, browsers with drastically reduced features but increased speed appears. They'll work with "fat web" servers but throw away half the features (so we'll end up designing two versions of each website). Eventually most of the functionality of the old "fat web" will migrate across - obeying the 90/10 rule (the 10% of features that provide 90% of the value). With 90% of the value, plus increased speed, everyone will move to "thin web" browsers. Chrome/ium will be left where ActiveX is now - some non-bleeding-edge organisations will use it because it has features that didn't make it to the "thin web" and replacing it is a pain, but it's obsolete and no-one really uses it.
- madrox 6y ago
  Stadia is a great example of this
  - Tom4hawk 6y ago
    Stadia is just Google trying to centralized another thing. Their pricing is ridiculous. Group of people that are ready to pay monthly fee, pay full price for games (and remember - no used copies[big market on consoles] and no competing stores[PCs]), live so close to the Stadia servers that latency is not a problem and have a good internet connection is unbelievably small.
- nwah1 6y ago
  Google AMP and Facebook Instant Articles are doing this.
  - marcus_holmes 6y ago
    Yeah, you're right. I hoped it wouldn't come from the giants, but as a grassroots tech thing, but I guess we don't live in that world any more.
kristopolous 6y ago
The HTML/web model is still fundamentally incomplete. Links are one way, things don't flow based on concept but based on this clunky chunk called a page. We still only have one functional semantic for cross linking the clunky chunks called the hyperlink. The only way we have to find things are to use companies that have managed to visit and index all the sites, not through some search and index interface but by processing the page that the human sees and then trying to rank them making the web ultimately heavily centralized. Then there is the authorship meta layer that doesn't exist to interleave content from the same author, say by email or some other identifier. Instead we've centralized that into temporal structures that vanish when you look back, like in a dream in a "let 1000 flowers bloom" method where all you ever end up with are just a bunch of silly flowers. Then there is the permanence feature where the internet is like a river that's always moving. Someone else has to rely on generosity to yet again, centralize the historicals so hopefully things can exist at least as long as bad ink on terrible paper. Otherwise a couple generations of thought will be in metal garbage heaps I could have written all this, word for word in 1993 but instead it's 2020, Honestly, I'd say solutions to 70% of the fundamental problems are still unsolved nearly 30 years later.
- tannhaeuser 6y ago
  Regarding two-way links etc., XLink/XInclude, based on SGML and HyTime concepts, was proposed by Sun back then in 1998 or so. Thankfully, that didn't take off.
- xorcist 6y ago
  Incomplete compared to what? There are no hard dependencies between web sites and links fail all the time. The web didn't succeed in spite of this, but because of this. The decision to use one-directional links was a good design choice in retrospect. It has also kept the web agile enough to keep evolving.
- slightwinder 6y ago
  > Links are one way Is it even possible to change this? We did had (or still have?) backlinks in the blogging-world (trackback/pingback), but this is some active mechanism which simple does not scale well for physical reasons. > things don't flow based on concept but based on this clunky chunk called a page The base is a document, which describes the content and interface. "Page" is just the naming for some. Some others are called "App". What else is their? > We still only have one functional semantic for cross linking the clunky chunks called the hyperlink. We do have other formats than HTML. HTML had different approaches to include more stuff. Did not work for reasons. AT the end all dreams shatter on reality.
anymouse2 6y ago
When Microsoft insists everybody is up to standard.
anymouse2 6y ago
When Microsoft insists everyone is standard.
ppf 6y ago
When Google itself cannot keep up with the moving target it has created, then it will be complete. I call it the Singularity.
amelius 6y ago
They will be complete when we realize that they can be better implemented as generic virtual machines.
smitty1e 6y ago
This seemed a key point: > We can logically follow this to a future of a "split Web". There's going to be the "Google Era" Web that will live for a long time, and another iteration of the Web where us technologists will try our best to create something more to our liking. All human endeavors tend toward Towers of Babel. Alternates shall arise and topple Google.
peterwwillis 6y ago
The browser becomes the OS and then that OS will get a browser. Turtles.
amadeuspagel 6y ago
This focus on chrome os gets it completely backwards. Google doesn't expand chrome's capabilities because they need it for chrome OS, google invests in chrome and chrome OS, because it supports their search engine business. Google wants anything you might want to do with your computer to start by searching google, not some kind of app store. See: https://www.gwern.net/Complement https://www.gwern.net/Complement
- gwern 6y ago
  I would also point out the problem of "bitcreep" ( https://www.gwern.net/Holy-wars https://www.gwern.net/Holy-wars ): Google can keep Chrome-OS and Chrome 'FLOSS' indefinitely much the same way it has kept Android's FLOSSness largely nominal - by simply having complex enough, slowly shifting, incompatible behavior, everything gets sucked in around it. FLOSS is slow, and can only catch up if there is a stable target. (Stephenson covers this in "In The Beginning Was The Commandline" too.) Add in the service-fication of Android, and now the key code and systems aren't even FLOSS, as the FLOSS code exists only as a partial bootloader for the services that every user (and every website or application they use?) require.
02020202 6y ago
i am still baffled that we haven't moved away from this old html thing and made new protocols and new ways to draw UI over internet already. html protocol and markup language was never designed to do what we're doing with it today to begin with. and don't get me started with javascript. i guess backwards compatibility is more important than solid architecture and advancing the technology.
zokier 6y ago
Personally I feel browsers got feature complete already somewhere around 2011, Firefox 4 was a great release and did pretty much everything I would want from browser. Last really important spec feature I feel was video element, which was added around that time. Got us rid of need for crappy plugins. Honestly I'm not sure what I'd miss if I had to go back to Fx4.
- BlueTemplar 6y ago
  Yeah, IMHO browsers got worse in recent years, first with Opera 12 being killed off, then Firefox losing most of the power of its features/ plugins... (I still can't find anything nearly as good as Tab Groups !)
chrismorgan 6y ago
> So what is the Web? Well we can agree the Web is a conglomerate of standards proposed by the W3C. No, I can’t agree with that because it hasn’t ever been true. For its first decade, it was the place where most relevant standards were proposed, but that still didn’t define the web; you had to look at what browsers did rather than what specs said, especially in such days as IE’s domination, because it varied quite substantially from the “standards”, and more importantly people depended on these variations and extensions to the “standards”. But then W3C tried to take the web further in a direction that the actual people with power (browser manufacturers) didn’t like—more XML everywhere, and little to no activity on HTML itself—and so they split and formed WHATWG which has since taken over progressively more and more of the actual important standards (e.g. HTML, DOM, URL, Fetch, Encoding), so that W3C itself is increasingly irrelevant, though there are still important things that are done through W3C, e.g. the CSS Working Group (CSSWG); and if you’re willing to stretch it, you could count the Web Incubator Community Group (WICG), but that’s more hosted by W3C than part of W3C. Nowadays, if you say, “what is the web?”, you must include specifications from at least IETF (e.g. HTTP), WHATWG (e.g. HTML), W3C (e.g. CSS) and Ecma (e.g. JavaScript); but actual browser behaviour (an extremely vague concept) must also be considered too, because it has a big impact on what the web is.
- totallymike 6y ago
  Decent point. If my recollection is accurate, a number of browsers or other implementations must already implement something before it can become standard.
- fergie 6y ago
  I have to confess that I didnt know anything about WHATWG until I read this comment- I always assumed that HTML was defined by W3C.
  - chrisseaton 6y ago
    > I always assumed that HTML was defined by W3C Yes I think that's the impression they try to create.
  - 6y ago
api 6y ago
Web browsers won't ever be complete. They will bloat until the standard is abandoned and something leaner replaces it, like everything else.
coldtea 6y ago
Well, web browsers once meant hypertext and browsing, that is reading and navigating hypertext. That was complete (or close) 10-15 years ago, in the sense that it could do render most things print magazines could do with good enough html/css APIs. But that's not what people (including companies making SaaS, which are also stakeholders here), want from the web. They want an app delivery mechanism mediated through a web browser. So web browsers will be complete the same time when OS-level frameworks, UIs, and libs are complete, so when computing is "complete". That is, never.
pmontra 6y ago
Having used the web for twice as many years as the author did I feel optimistic about betting that web browsers will be complete only when nobody will use them anymore. We're very good at inventing new stuff and OSes and browsers have to keep up.
ForHackernews 6y ago
As soon as Google has finished killing Mozilla and Safari?
pessimizer 6y ago
They'll be complete when users have absolutely no way to interact with a web page other than what was explicitly intended by the company serving the web page, and when every possible interaction between the user and their computer is reported back to the the web page they're accessing. There will be three browsers: The "Safe Browser" which will be used by 85% of people, the "Smart Browser" which will be used by 5% of people, and the "Rebel Browser" which will be used by 10% of people, most of them in their teens and 20s. They will each have different skins (and t-shirts), but all be identical under the hood. The Rebel Browser will allow you to consent to accessing porn, and when the Smart Browser reports that it is the Smart Browser to sites its users visit, they will only be served articles, ads and search results that make them feel smarter than everyone else. The Safe Browser will have an "Important Business" mode that will allow you to consent to accessing porn, but it will examine your face first to ostensibly determine if you're a serious adult. The Rebel browser will also silently examine your face, and for the same reason - to track your porn preferences. The day it will be feature-complete is when sites get an API for bricking your computer, which is routed through an algorithmic federal judge.
- userbinator 6y ago
  ...and there will be a 4th one used by almost 0% of people, which is illegal to use along with the general-purpose computer that's required to use it, but provides the user complete control: https://www.gnu.org/philosophy/right-to-read.en.html https://www.gnu.org/philosophy/right-to-read.en.html https://boingboing.net/2012/01/10/lockdown.html https://boingboing.net/2012/01/10/lockdown.html IMHO the web started to go down this path in the late 2000s. Recent developments (manifestv3, DoH, youtube-dl, etc.) are only further indications of the continuing trend towards increasing corporate control and dumbing-down of the users (so they become less likely to oppose.)
- MacsHeadroom 6y ago
  This is presented in a sort of hyperbolic way. But I do really think this is the future- and not as far off as people think. But not for nefarious reasons. The road to hell is paved with good intentions. Average users value performance, cost, and battery life. The cost efficiency and capabilities of cloud compute will always outpace that of consumer mobile devices. So remote compute with zero user control is inevitable. A stadia like experience for everything is what the overwhelming majority of consumers want- they just don't know it yet.
crazygringo 6y ago
> As we move forward, eventually, there will be nothing left to propose for the Open Web Platform. We will reach feature-parity with our local Operating Systems. But operating systems continue to innovate/improve/change at roughly the same rate as browsers these days, no? So it seems like a total logical fallacy that the browser would ever be complete. People always come up with new ideas and better solutions for things users need, new hardware arises that needs software support and input, etc.
smrtinsert 6y ago
As soon as operating systems are!
Ericson2314 6y ago
No professionally-written software is every complete, thanks to Conway's law and the need to have a job under capitalism. Simple as that.
c-smile 6y ago
Why plural? Probably "the web browser" is more correct term these days, no? The idea of "browser as OS" or "OS as browser" is not that new. That move started by "active desktop" concept on Windows 95 (https://en.wikipedia.org/wiki/Active_Desktop https://en.wikipedia.org/wiki/Active_Desktop). And speaking about browser... It's main goal is to serve following tasks: 1. To provide safe browsing experience. 2. To present content to the user. Note that the order of these two tasks matters. Initial active desktop effort has died because order of these two tasks was wrong at their first attempt. From that point of view, "browser as OS" just increases attack surface. And so more and more isolation layers will be added into browsers making them less effective and performant than alternatives. Consider WebAssembly for example. That was desperate move to provide an option to run some computation-heavy code inside browsers with at least comparable speed of native code. But it always be slower than native solutions. Java vs native code, we've been there, seen that. To be short: native OS and solutions will always be more performant than the one emulated in the browser. Just because of the order of tasks. In this sense the browser will never be complete. It may become asymptotically close to the OS but will never reach that point. And note the price: on that way it will become asymptotically close to the size of OS.
- emteycz 6y ago
  Soon CPUs and GPUs will be so fast and so efficient and so cheap that it doesn't matter.
  - c-smile 6y ago
    That's never ending story. As soon as you will have better GPU you will have better native native games that use HW at maximum extent. They will work at the magnitude of times better than what WebGL emulation layer will offer you. So hi-end games will always be native.
shams93 6y ago
Well for Safari the answer is never, like Apple have said no to any browser features that compete with their native apis so no web midi no web usb etc...
- shaicoleman 6y ago
  Indeed, all these features are marked as "Not Considering" https://webkit.org/status/ https://webkit.org/status/
BlueTemplar 6y ago
One aspect is how IPv4's quirks (and Google's domination plans) have pushed the Internet to push more and more things over HTTP(S). IPv6 will hopefully change that, and native programs will get another reprieve from WebApps. (I also have to point out that WebApps still mostly suck - and are they even used yet on mobile/wearables/VR ?)
nottorp 6y ago
When it doesn't make business sense to embrace and extend any more.
fallat 6y ago
I will take the replies and create a response post tonight.
mdip 6y ago
> ... used the Web since 2007 Am I really that old? Something that was stated, but not directly, is what's driven a lot of the latest additions to the browser: Cross-platform development. The operating system provided the abstractions around various hardware implementations in the past, which meant being able to (possibly) run your software, if re-compiled, on different hardware platforms (it reality, some OSes weren't available except on a single platform and others were frequently not straight forward to get compiling on different platforms). The browser is providing abstractions to the same hardware around various OS implementations. Add in WebAssembly, and you can pick from a number of languages -- including those that were traditionally used to write apps not designed to run in browsers. But to the quote: I was cleaning out my basement, which quickly became a "treasure hunt" of sorts since I've been storing things that I took from my parents' house when I moved out. My favorite find was a CD-ROM that was basically a CD-based Search Engine of the Internet. It was "Designed for Windows 95" and was probably a free disc given to me by a vendor rep when I worked at CompUSA in my teens. I'm tempted to see how many of these actual sites still exist (I'm sure none of the deep links do). Incredibly, these things probably sold very well back then. It even came with a browser (neither Mosaic nor Netscape; no clue who's it is).
willseth 6y ago
The author has very limited knowledge of the web, its history, or the W3C's ambitions (not to mention WHATWG, IETF, ECMA, etc.). His view of the web in general is bafflingly myopic, with an almost completely arbitrary understanding of the businesses discussed. It should come as no surprise that the conclusions he draws are non sequitur. So why has this been upvoted? Why is this on HN at all? It's ignorant nonsense. This should be obvious to anyone who's spent more than a few hours researching these topics. Please stop taking no-nothing randos seriously.
- billyjobob 6y ago
  It does say "As someone who has used the Web since 2007" so I guess it's written by a child? So probably we shouldn't be too harsh judging them.
- ritchiea 6y ago
  Lots of users vote based on the headline before actually reading the article. Lots of users even comment without reading the article. HN is an extremely large community these days.
Existenceblinks 6y ago
I have a complete unlovely thought; web browser needs a rewrite with thin web specs (imagine javascript the good part book meme). People think it's impossible to rewrite, but I believe it's only because it's a mess. Inventing a new programming language and new DOM api that works greatly with webassembly probably worth it, consider that the next 10 years the web is still not going anywhere. You may laugh at Flash .. but hey do you love "hydration" thing? I'd love to have an isomorphic?(terrible name) template engine that rendered as text as well as client program binary. Do you love OS based form control that can't be styled? I'd love a complete separate controls from OS, clean the contract between widget and actual system api. Just stop caring (coercing) how it looks, just provide api to enforce how its functionality works, we already have these bunch of aria- thing. The question is what block the progress of making it complete, what kind of problem? I believe it's not a technical problem. Maybe incentive?
- acoard 6y ago
  It's both technical and incentive. The problem is that so much of technology is standing on the shoulders of giants. If we re-write, how much of the giants do we decide to lop-off? For example, could you get Google, Apple, Mozilla, to all agree where to draw the line? Should we still use HTTP? Should everything use QUIC? An entirely new standard? What about DNS? But let's say they do agree. Then what, they create entirely new browsers (I'll call Chrome2, Safari2, etc). Maybe they don't even use the DOM or HTML at all, but an entirely new setup. A side note – if the web was re-built today it probably wouldn't have "View Source." As you said, I imagine it'd lean into WASM and "isomorphic" code, which could just be WebGL/canvas/some new paradigm/etc. So then what... everyone says abandon my original website and download a new browser to access the new one? Because that's what throwing out backwards compatibility looks like. And that seems like a non-starter to most people. Browsers are too big to fail; too big to be thrown out and re-written. Our hope is incremental, not revolutionary. Things like WASM in browsers let us pursue new paradigms while also being backwards compatible. Lastly, I'd say the incentives for a new web would be worse than an old web. The original web was written by tinkerers and tech startups like Netscape/Mozilla. The new web would be written by multibillion dollar companies who like and know how to build moats. It would have microtransactions (not necessarily bad by itself) and I worry would look more like an app store.
tabtab 6y ago
I believe the argument is that Chrome's code-base is becoming the de-facto browser standard, which will allegedly simplify things by making a stable platform. But it's been hard to predict browser trends. Look how Chrome came out of nowhere and knocked Microsoft's off the top. Netscape also got sucker-punched. Who knows what will punch next. Web standards still have some glaring gaps that I'm not sure the current approach can tame. I propose "the web" be split into three standards: 1. Media/Games/Art/Video 2. Documents (existing HTML may be good enough with minor tweeks) 3. Productivity: CRUD and Data oriented interactive GUI By splitting, each standard can FOCUS on doing what it does best rather than the watered-down one-size-fits-all we have now. A given browser may support all 3, somewhat like how Java Applets and Flash could run inside of HTML documents. (They may be pluggins and/or independent browsers.) But unlike Applets & Flash, the standards wouldn't try to be full virtual OS's. Dump as much on the server as you can to keep client standards lean, and reduce the need for scripting by including common behaviors. For example, a common GUI idiom is for a button to activate a window or form. Rather than rely on scripting, that action could be built into the GUI markup: <button openTarget="myWindow"> Click to open My Window </button> Or maybe: <button label="Click to open My Window et. al."> <action> <setFocus target="myWindow"> <setFocus target="anotherWindow" modal="false"> </action> </button> The point is that if common GUI idioms and behaviors are built into the GUI markup standard, scripting (typically JavaScript) is not needed as often. The GUI Markup Language would be state-friendly and natively interactive, unlike HTML.
moritzwarhier 6y ago
I hope this doesn't classify as trolling; But isn't it ironic that the author uses a GitHub gist to share this plain text essay? Had to click "View Raw" to read it on mobile. I know the essay is about more topics than just documents vs apps. Still that shows to me that "The principle of least power" doesn't easily apply to technology choices of web users. GitHub was probably more convenient for the author than uploading a txt file somewhere.
akrymski 6y ago
Alternative end game: the browser dies slowly, gradually being replaced by native apps. We're already seeing this: consumers prefer native apps on their phones for virtually everything: shopping, socialising, reading news. Surely the future us will laugh at web browsers as some ancient tech that used these weird markup languages to layout content that's monetised with ads into pages that you'd navigate back and forth in order to emulate an application? New web browser engines are now impossible to implement due to the sheer size of specs. Surely we are nearing the end game?
- sfink 6y ago
  Consumers prefer native navigation (go straight to an app, not home -> browser -> app) to apps that in practice are native wrappers around web views.

144 points by fallat 6y ago

Quick comment on the presentation. This post could very easily be markdown and therefore rendered in a nice readable non-monospaced font. Instead it's a .txt and it's really painful to read. Edit: here's a rendered markdown conversion that is easier on the eyes: https://shrib.com/?v=nc#CapeStarling-6v1L9J https://shrib.com/?v=nc#CapeStarling-6v1L9J

tleb_ 6y ago
How is that easier to read? https://wtf.roflcopter.fr/pics/SpRqNBvB/0QUVpbN4.png https://wtf.roflcopter.fr/pics/SpRqNBvB/0QUVpbN4.png
- bloody-crow 6y ago
  Ugh, it looked better on my screen.
eCa 6y ago
I recommend the raw rendering, followed by the browser’s reader mode. But I agree in principle.

fallat 6y ago

OP here (about to sleep): this was just some 2am ramblings that I found could invoke some interesting conversation. Already there are some great conversations. Night!

err4nt 6y ago

W3C does not publish all web standards, notably it doesn't publish the three biggest technologies used in web browsers: - HTML is worked on by WHATWG - CSS is worked on by CSSWG - JavaScript is worked on by TC39 And IE wasn't killed by features moving too fast, Microsoft made Edge browser and still does. Mozilla hasn't been killed by Google or competition, it's been kept alive in large part by Google (I think in some years as much as 90% or more of its total income coming from Google) so they get hundreds of millions—the problem at Mozilla is bad management for the last decade. I'm not sure how well this vision represents reality as it already is today.

why_only_15 6y ago
Microsoft makes a browser but the point people are typically making here is that Edge is just a reskin of Chromium.
- err4nt 6y ago
  Between IE and the current Chromium-based Edge browser, was born, lived, and died another browser engine named EdgeHTML. IE was 2 browser families ago from Microsoft.
  - Gaelan 6y ago
    I thought EdgeHTML was just Trident with some legacy code ripped out. Edit: admittedly, "just" is probably underselling it a bit, but it's still a far cry from a completely new engine.
    - gsnedders 6y ago
      It was very much a fork of Trident, yes. I believe there was significant refactoring done after the legacy code was ripped out.
  - coldtea 6y ago
    Yes, but that was killed because Chrome moves too fast, and maintaining EdgeHTML wasn't easy for MS. Parent said: IE wasn't killed because the web moves too fast -- and used the presence of Edge (or EdgeHTML) as the argument. But both IE and EdgeHTML were killed because Chrome moved too fast for MS to be worth it to maintain its own engine...
gsnedders 6y ago
> - CSS is worked on by CSSWG The CSS WG is a W3C WG, and they're very much published as W3C specifications.
- err4nt 6y ago

gbh444g 6y ago

"There is not much left for a Web browser to cover" - this reminded me of the famous "physics is mostly complete at this point." Web (browser) is the only cross-platform operating system, even if some standards say otherwise. I'm sure we'll see docker-style virtual web environments, cross-machine abstractions to simulate datacenters for distributed computing (GPU and such), PWA-style webapps bootable from USB sticks and functioning as operating systems, various crazy extensions for the upcoming neuralink-style tech, virtual reality frameworks, holographic displays and so on and so forth. Physics is never complete at any point.

catmanjan 6y ago
The fact that we're comparing a type of software to a school of science speaks volumes
- jupp0r 6y ago
  Is it? You could make the same point about engineering, literature, the arts, sociology, ... you get the idea.
  - tannhaeuser 6y ago
    I'm not getting the idea. The argument, as I understand it, was that a browser should aspire to be a standardized and relatively simple app enabling a large variety of backends to deliver content, not a never-ending "social experiment" to become overwhelmed by a single party extracting almost all value off the web, and becoming the only party with the resources to implement browsers.
    - jupp0r 6y ago
      The notion that the world wide web is "finished" at some point is just flawed. Today we do video conferencing, 3D gaming, biometric authentication, photo editing and all sorts of other things on the web. Browsers will evolve along the lines of users needs and developers innovation. If this stops at some point, they will become the new Gopher and something else will come along to replace them.

rado 6y ago

Wish they were stable first. There is undefined and buggy behavior, e.g. sub pixel scrolling, scroll container padding etc.

tannhaeuser 6y ago

> there is not much more distance to cover [relative to native toolkits] As I see it, the web as originally conceived was "ready" around 2003, and then again around 2009 when CSS media queries became mainstream to adapt to mobile usage. For me, the excessive push for browser APIs as a desktop replacement since then, and the proliferation of frameworks is a generational thing, driven not by technology but by big data, SaaS (prospect of extracting recurring revenue from customers, rather than one-time license fees), and a new generation of developers getting into the game by leaving the status quo behind, solely for their own economical benefit. Google has masterfully played this game, and let the idea of open web standards bend the minds of developers who thought they were clever. Moz partnering with them under WHATWG seemed like a good idea at the time, but just made the concept of a web standard arbitrary and "whatever Chrome does" over time, leading to Moz's demise and that of every other "browser vendor". It'll take some stance and political will to reclaim the web.

cageface 6y ago
The alternative is an internet completely locked down in native apps in proprietary and unaccountable app stores. I'll take a bloated web browser over a corporate monoculture any day.
- konjin 6y ago
  Sorry which is which? Because right now Chrome engine has more control of the web then IE did in it's heyday when Microsoft was ruled to be a monopoly.
  - vntok 6y ago
    Are you aware that Blink, Webkit and V8 are all open source?
    - pjmlp 6y ago
      With most people working out of Apple, Microsoft and Google salaries, good luck getting a bunch of random weekend coders to pick up and counterattack an alternative browser based on that open source.

tutfbhuf 6y ago

Never.

benjaminjosephw 6y ago

I think the disruptive innovation for the mono-culture in browser development will be web platforms that enable meaningful and broadly accessible end-user programming. The established players have no incentive to pursue this objective (since it undermines central ownership of data/experience) but, if executed well, could be an extremely valuable proposition for users. Google's current trajectory won't take us there but maybe others can.

ksec 6y ago

Let's talk about Web Page for a minute, or Web Page with some interactivity and animations. Are there any Web Page, render in Web Browser, both the page itself, and the standard itself, that gives the smoothness of say opening a PDF, or even better, a Native App. Does the Apple Store Web Page for selling iPhone felt as smooth as their Apple Store App on your iOS? Jank ( micro- pause ) are still everywhere. Both the Browser and the WebPage itself. And this isn't Web App, just simple Web Pages. I say web browsers ( and the Web Standards ) is far from complete.

the_gipsy 6y ago
I don't remember the app store app being smooth. I do agree about the "janks".

boltefnovor 6y ago

Internet Explorer is finished.

mschuetz 6y ago

Web browsers are just getting started, in my opinion. Browsers already are the most effective way to share content with a massive audience, and WebGPU and WebXR are going to lead to a whole lot of new types of content.

makach 6y ago

It is only when they are no longer needed that you can state that they have been completed. There will be no web browsers in the future! What comes after web browsers? Only exciting times...!

dmitriid 6y ago

The main problem with the current state of web browsers is that it's now impossible for a third party to develop a competing browser from scratch (or nearly from scratch). The already existing set is unimaginably huge, and it's a moving targets as browsers (especially Chrome) is throwing more and more features into it.

lazyjones 6y ago
That's on purpose, of course. And everyone who suffers from it, willingly tolerates it.
tannhaeuser 6y ago
And WHATWG (and CSS WG) rejects the idea of HTML profiles (defining reasonable subsets) and even versions, or anything forward-compatibility for that matter, with supporters standing by and cheering.
anonymou2 6y ago
that's why project Gemini was created: https://gemini.circumlunar.space/ https://gemini.circumlunar.space/

onion2k 6y ago

We will reach feature-parity with our local Operating Systems. The author states this in a way that makes it sound like it's possible, but that betrays an assumption that operating systems aren't moving forwards and innovating as well. The web will always lag behind the new features that come to OSs, and therefore browsers will always need to be changing in order to give web users access to those features.

enos_feedler 6y ago
Not only are operating systems moving forward in their own direction, they are picking up the missing pieces that were traditionally reserved for the web. The two most important being URLs for deep linking between apps (universal links) and installing/fetching without redirect to a download page (app clips). At some point we will cut out the store publishing step for certain classes of same apps running in safe runtimes (wasm with some native bindings with safe data management). Then what is the difference? You might say compatibility across platforms. But compatibility really is broken in browsers today. So....

nikitaga 6y ago

Chrome OS being open source is not a threat to Google for several reasons: The usefulness of the browser is that it's a cross-platform platform. You can run the same web apps on Windows, Mac, Linux, even mobile devices. Turning the browser itself into an OS-level platform will just create another platform that will need to compete with all existing platforms while having nothing unique to offer – after all, the same web apps run just fine on existing platforms, by their very design. What does ChromeOS have that MacOS or Linux do not have and can't have? You could argue that someone could build a competitor to iOS and Android using ChromeOS, but why wouldn't the same someone build a competitor using... Android? It's open source too. Because if they take Android source without Google's blessing, they would be left without Google maps, docs, and gmail. But Google can just as easily exclude mobile browsers from accessing their web services. In today's world it would be perfectly fine for Google to only offer their services through native apps on mobile, and it's up to them which platforms they choose to publish those apps on. Oh and then there's Widevine. You need Google's blessing to decrypt that, and even if you get it, every content provider that uses Widevine DRM can block access to their content from your browser for any reason (you can't bypass that). How far will your rebel browser platform go without being able to play e.g. youtube? Currently they only DRM certain types of videos, but what would stop them from flipping the switch and applying DRM to every video that is monetized on youtube, and blocking your browser on the basis that it allows adblockers that effectively prevent funding of those videos? The notion that Google is somehow – anyhow – in a weak position is utterly ridiculous. Their moat is stronger than ever.

pjmlp 6y ago
With the Chrome monoculture, the Web is effectively ChromeOS nowadays, specially with the increasing amount of APIs that are Chrome only, which although other browsers refuse to implement really doesn't matter as their market share keeps decreasing towards zero.
- hesarenu 6y ago
  If chrome advances it does not matter. The lowest common denominator would be safari specifically ios safari. Until it adopts we would have to code to it's standard.

qppo 6y ago

I can't read this plaintext in a mobile browser window so clearly there's some work left to do. Or maybe they've done too much work. Who knows.

kevsim 6y ago
Reader mode in safari worked OKish for me.

madrox 6y ago

When this essay started with "As someone who has used the Web since 2007" I immediately felt old, so thanks for that. As someone who has been using the Web since 1994 (and I feel like I was late to the game), I can assure the author their perspective has a lot of recency bias and this isn't something to worry about, in the general sense. Or worry about it, because it already happened a while ago. Take your pick. Local operating systems, which this essay says is the gold standard of feature sets, are not standing still. If you think of how long it will take for browsers to reach current parity, go back that amount of time and see how far the OS has come. But let's consider, for a moment, that browsers catch up. Let's just look at native development in general, since this has had access to all local OS features since before the web. Has that never changed? It's changing all the time. We're finding higher abstractions to work with. We're making it easier. I highly doubt Grid and Flexbox as they exist today will still be how web layout works in 20 years. Google isn't even the first to scare the world. This essay could've been written in 2002...about Internet Explorer. Stop laughing. It held even more dominance than Chrome does today, and was closed source. Somehow we survived...which is good, because VRML sucked. However, even with all that aside, people have been rolling their own browsers for years. People fork chromium today. The dark web exists. Maybe in 20 years lots of people will be streaming everything from a private cloud supercomputer vis-a-vis Stadia. The internet is no more homogeneous than all the people of the world. Everyone is finding new ways to access it all the time. This is why open standards are more important than browsers. It means anyone can build their own at any time based on the documentation and it can all read the same documents. So I wouldn't worry about this coming to pass in quite the way you think. Though running your browser in a VM isn't the worst idea.

cft 6y ago
In 20 years it will be impossible to manually browse to an unapproved "unsafe" URL in a general purpose browser.
- ktpsns 6y ago
  Well, its hardly possible to manually browse an unapproved unsafe URL even today. Ever tried to browse a site with a self-signed certificate? In most browsers you have to do a lot of extra clicks to get in. But there will always be special purpose browsers. There is "TOR browser" today, and we will have "vintage web browsers" the same way we have CLI/TLI browsers still today (thinking of elinks and lynx), or the same way as there are still FTP clients and Gopher clients.

Hurtak 6y ago

"When will operating systems be complete?" "When will car technology be complete?" ... The answer is they wont be in short term, and they will be made obsolete in the long term.

yodelshady 6y ago
Cars don't suffer feature creep. We've optimised them - removed the horse, switched to steel construction, added electrical subsystem - but really there's been no major functional change in cars, ever.
- dwheeler 6y ago
  Um, what? Our expectations about cars have definitely changed over time. I don't think it's even legal to sell new cars without seatbelts, something that did not exist when cars started. Even low end cars in the US have added features compared to low end cars of the past, such as power windows, power steering, power mirrors, and so on. It's true that adding features is slower in cars, but that is because cars are partially hardware. It takes much longer to add features to hardware, because making copies is not free. But that does not mean they are unchanged.
- rusticpenn 6y ago
  I am not sure if you are sarcastic.
  - yodelshady 6y ago
    Nope. Cars have advanced, but if they were webtech, we'd be wondering why they've not implemented OrbitalMechanics support yet. OceanMechanics would be old hat.
- TeMPOraL 6y ago
  ...added electrical subsystem, replaced combustion engine with batteries and electric motors, we're adding various degree of autonomy. In parallel, we've replaced sane controls with frustrating and dangerous touchscreens, we've added countless types of comfort features, we've improved security, we've computerized engine controls, we've DRMed the car - ostensibly to prevent theft, but actually to route more money from aftermarket towards the manufacturers. Etc. Cars do suffer feature creep, and not all of the features are beneficial to their owners. Just like with the web.

marcus_holmes 6y ago

I've seen the same cycle happen repeatedly in tech (there are many cycles, but this is one of them): systems are centralised on a server, then the client gets more powerful and features of the system are pushed to the client. This gets too much and the client slows down, so the features are centralised to a server again. I can easily see a world where Chrome/ium becomes too bloated to work well, and a "new web" of thin, light, browsers with drastically reduced features but increased speed appears. They'll work with "fat web" servers but throw away half the features (so we'll end up designing two versions of each website). Eventually most of the functionality of the old "fat web" will migrate across - obeying the 90/10 rule (the 10% of features that provide 90% of the value). With 90% of the value, plus increased speed, everyone will move to "thin web" browsers. Chrome/ium will be left where ActiveX is now - some non-bleeding-edge organisations will use it because it has features that didn't make it to the "thin web" and replacing it is a pain, but it's obsolete and no-one really uses it.

madrox 6y ago
Stadia is a great example of this
- Tom4hawk 6y ago
  Stadia is just Google trying to centralized another thing. Their pricing is ridiculous. Group of people that are ready to pay monthly fee, pay full price for games (and remember - no used copies[big market on consoles] and no competing stores[PCs]), live so close to the Stadia servers that latency is not a problem and have a good internet connection is unbelievably small.
nwah1 6y ago
Google AMP and Facebook Instant Articles are doing this.
- marcus_holmes 6y ago
  Yeah, you're right. I hoped it wouldn't come from the giants, but as a grassroots tech thing, but I guess we don't live in that world any more.

kristopolous 6y ago

The HTML/web model is still fundamentally incomplete. Links are one way, things don't flow based on concept but based on this clunky chunk called a page. We still only have one functional semantic for cross linking the clunky chunks called the hyperlink. The only way we have to find things are to use companies that have managed to visit and index all the sites, not through some search and index interface but by processing the page that the human sees and then trying to rank them making the web ultimately heavily centralized. Then there is the authorship meta layer that doesn't exist to interleave content from the same author, say by email or some other identifier. Instead we've centralized that into temporal structures that vanish when you look back, like in a dream in a "let 1000 flowers bloom" method where all you ever end up with are just a bunch of silly flowers. Then there is the permanence feature where the internet is like a river that's always moving. Someone else has to rely on generosity to yet again, centralize the historicals so hopefully things can exist at least as long as bad ink on terrible paper. Otherwise a couple generations of thought will be in metal garbage heaps I could have written all this, word for word in 1993 but instead it's 2020, Honestly, I'd say solutions to 70% of the fundamental problems are still unsolved nearly 30 years later.

tannhaeuser 6y ago
Regarding two-way links etc., XLink/XInclude, based on SGML and HyTime concepts, was proposed by Sun back then in 1998 or so. Thankfully, that didn't take off.
xorcist 6y ago
Incomplete compared to what? There are no hard dependencies between web sites and links fail all the time. The web didn't succeed in spite of this, but because of this. The decision to use one-directional links was a good design choice in retrospect. It has also kept the web agile enough to keep evolving.
slightwinder 6y ago
> Links are one way Is it even possible to change this? We did had (or still have?) backlinks in the blogging-world (trackback/pingback), but this is some active mechanism which simple does not scale well for physical reasons. > things don't flow based on concept but based on this clunky chunk called a page The base is a document, which describes the content and interface. "Page" is just the naming for some. Some others are called "App". What else is their? > We still only have one functional semantic for cross linking the clunky chunks called the hyperlink. We do have other formats than HTML. HTML had different approaches to include more stuff. Did not work for reasons. AT the end all dreams shatter on reality.

anymouse2 6y ago

When Microsoft insists everybody is up to standard.

anymouse2 6y ago

When Microsoft insists everyone is standard.

ppf 6y ago

When Google itself cannot keep up with the moving target it has created, then it will be complete. I call it the Singularity.

amelius 6y ago

They will be complete when we realize that they can be better implemented as generic virtual machines.

smitty1e 6y ago

This seemed a key point: > We can logically follow this to a future of a "split Web". There's going to be the "Google Era" Web that will live for a long time, and another iteration of the Web where us technologists will try our best to create something more to our liking. All human endeavors tend toward Towers of Babel. Alternates shall arise and topple Google.

peterwwillis 6y ago

The browser becomes the OS and then that OS will get a browser. Turtles.

amadeuspagel 6y ago

This focus on chrome os gets it completely backwards. Google doesn't expand chrome's capabilities because they need it for chrome OS, google invests in chrome and chrome OS, because it supports their search engine business. Google wants anything you might want to do with your computer to start by searching google, not some kind of app store. See: https://www.gwern.net/Complement https://www.gwern.net/Complement

gwern 6y ago
I would also point out the problem of "bitcreep" ( https://www.gwern.net/Holy-wars https://www.gwern.net/Holy-wars ): Google can keep Chrome-OS and Chrome 'FLOSS' indefinitely much the same way it has kept Android's FLOSSness largely nominal - by simply having complex enough, slowly shifting, incompatible behavior, everything gets sucked in around it. FLOSS is slow, and can only catch up if there is a stable target. (Stephenson covers this in "In The Beginning Was The Commandline" too.) Add in the service-fication of Android, and now the key code and systems aren't even FLOSS, as the FLOSS code exists only as a partial bootloader for the services that every user (and every website or application they use?) require.

02020202 6y ago

i am still baffled that we haven't moved away from this old html thing and made new protocols and new ways to draw UI over internet already. html protocol and markup language was never designed to do what we're doing with it today to begin with. and don't get me started with javascript. i guess backwards compatibility is more important than solid architecture and advancing the technology.

zokier 6y ago

Personally I feel browsers got feature complete already somewhere around 2011, Firefox 4 was a great release and did pretty much everything I would want from browser. Last really important spec feature I feel was video element, which was added around that time. Got us rid of need for crappy plugins. Honestly I'm not sure what I'd miss if I had to go back to Fx4.

BlueTemplar 6y ago
Yeah, IMHO browsers got worse in recent years, first with Opera 12 being killed off, then Firefox losing most of the power of its features/ plugins... (I still can't find anything nearly as good as Tab Groups !)

chrismorgan 6y ago

> So what is the Web? Well we can agree the Web is a conglomerate of standards proposed by the W3C. No, I can’t agree with that because it hasn’t ever been true. For its first decade, it was the place where most relevant standards were proposed, but that still didn’t define the web; you had to look at what browsers did rather than what specs said, especially in such days as IE’s domination, because it varied quite substantially from the “standards”, and more importantly people depended on these variations and extensions to the “standards”. But then W3C tried to take the web further in a direction that the actual people with power (browser manufacturers) didn’t like—more XML everywhere, and little to no activity on HTML itself—and so they split and formed WHATWG which has since taken over progressively more and more of the actual important standards (e.g. HTML, DOM, URL, Fetch, Encoding), so that W3C itself is increasingly irrelevant, though there are still important things that are done through W3C, e.g. the CSS Working Group (CSSWG); and if you’re willing to stretch it, you could count the Web Incubator Community Group (WICG), but that’s more hosted by W3C than part of W3C. Nowadays, if you say, “what is the web?”, you must include specifications from at least IETF (e.g. HTTP), WHATWG (e.g. HTML), W3C (e.g. CSS) and Ecma (e.g. JavaScript); but actual browser behaviour (an extremely vague concept) must also be considered too, because it has a big impact on what the web is.

totallymike 6y ago
Decent point. If my recollection is accurate, a number of browsers or other implementations must already implement something before it can become standard.
fergie 6y ago
I have to confess that I didnt know anything about WHATWG until I read this comment- I always assumed that HTML was defined by W3C.
- chrisseaton 6y ago
  > I always assumed that HTML was defined by W3C Yes I think that's the impression they try to create.
- 6y ago

api 6y ago

Web browsers won't ever be complete. They will bloat until the standard is abandoned and something leaner replaces it, like everything else.

coldtea 6y ago

Well, web browsers once meant hypertext and browsing, that is reading and navigating hypertext. That was complete (or close) 10-15 years ago, in the sense that it could do render most things print magazines could do with good enough html/css APIs. But that's not what people (including companies making SaaS, which are also stakeholders here), want from the web. They want an app delivery mechanism mediated through a web browser. So web browsers will be complete the same time when OS-level frameworks, UIs, and libs are complete, so when computing is "complete". That is, never.

pmontra 6y ago

Having used the web for twice as many years as the author did I feel optimistic about betting that web browsers will be complete only when nobody will use them anymore. We're very good at inventing new stuff and OSes and browsers have to keep up.

ForHackernews 6y ago

As soon as Google has finished killing Mozilla and Safari?

pessimizer 6y ago

They'll be complete when users have absolutely no way to interact with a web page other than what was explicitly intended by the company serving the web page, and when every possible interaction between the user and their computer is reported back to the the web page they're accessing. There will be three browsers: The "Safe Browser" which will be used by 85% of people, the "Smart Browser" which will be used by 5% of people, and the "Rebel Browser" which will be used by 10% of people, most of them in their teens and 20s. They will each have different skins (and t-shirts), but all be identical under the hood. The Rebel Browser will allow you to consent to accessing porn, and when the Smart Browser reports that it is the Smart Browser to sites its users visit, they will only be served articles, ads and search results that make them feel smarter than everyone else. The Safe Browser will have an "Important Business" mode that will allow you to consent to accessing porn, but it will examine your face first to ostensibly determine if you're a serious adult. The Rebel browser will also silently examine your face, and for the same reason - to track your porn preferences. The day it will be feature-complete is when sites get an API for bricking your computer, which is routed through an algorithmic federal judge.

userbinator 6y ago
...and there will be a 4th one used by almost 0% of people, which is illegal to use along with the general-purpose computer that's required to use it, but provides the user complete control: https://www.gnu.org/philosophy/right-to-read.en.html https://www.gnu.org/philosophy/right-to-read.en.html https://boingboing.net/2012/01/10/lockdown.html https://boingboing.net/2012/01/10/lockdown.html IMHO the web started to go down this path in the late 2000s. Recent developments (manifestv3, DoH, youtube-dl, etc.) are only further indications of the continuing trend towards increasing corporate control and dumbing-down of the users (so they become less likely to oppose.)
MacsHeadroom 6y ago
This is presented in a sort of hyperbolic way. But I do really think this is the future- and not as far off as people think. But not for nefarious reasons. The road to hell is paved with good intentions. Average users value performance, cost, and battery life. The cost efficiency and capabilities of cloud compute will always outpace that of consumer mobile devices. So remote compute with zero user control is inevitable. A stadia like experience for everything is what the overwhelming majority of consumers want- they just don't know it yet.

crazygringo 6y ago

> As we move forward, eventually, there will be nothing left to propose for the Open Web Platform. We will reach feature-parity with our local Operating Systems. But operating systems continue to innovate/improve/change at roughly the same rate as browsers these days, no? So it seems like a total logical fallacy that the browser would ever be complete. People always come up with new ideas and better solutions for things users need, new hardware arises that needs software support and input, etc.

smrtinsert 6y ago

As soon as operating systems are!

Ericson2314 6y ago

No professionally-written software is every complete, thanks to Conway's law and the need to have a job under capitalism. Simple as that.

c-smile 6y ago

Why plural? Probably "the web browser" is more correct term these days, no? The idea of "browser as OS" or "OS as browser" is not that new. That move started by "active desktop" concept on Windows 95 (https://en.wikipedia.org/wiki/Active_Desktop https://en.wikipedia.org/wiki/Active_Desktop). And speaking about browser... It's main goal is to serve following tasks: 1. To provide safe browsing experience. 2. To present content to the user. Note that the order of these two tasks matters. Initial active desktop effort has died because order of these two tasks was wrong at their first attempt. From that point of view, "browser as OS" just increases attack surface. And so more and more isolation layers will be added into browsers making them less effective and performant than alternatives. Consider WebAssembly for example. That was desperate move to provide an option to run some computation-heavy code inside browsers with at least comparable speed of native code. But it always be slower than native solutions. Java vs native code, we've been there, seen that. To be short: native OS and solutions will always be more performant than the one emulated in the browser. Just because of the order of tasks. In this sense the browser will never be complete. It may become asymptotically close to the OS but will never reach that point. And note the price: on that way it will become asymptotically close to the size of OS.

emteycz 6y ago
Soon CPUs and GPUs will be so fast and so efficient and so cheap that it doesn't matter.
- c-smile 6y ago
  That's never ending story. As soon as you will have better GPU you will have better native native games that use HW at maximum extent. They will work at the magnitude of times better than what WebGL emulation layer will offer you. So hi-end games will always be native.

shams93 6y ago

Well for Safari the answer is never, like Apple have said no to any browser features that compete with their native apis so no web midi no web usb etc...

shaicoleman 6y ago
Indeed, all these features are marked as "Not Considering" https://webkit.org/status/ https://webkit.org/status/

BlueTemplar 6y ago

One aspect is how IPv4's quirks (and Google's domination plans) have pushed the Internet to push more and more things over HTTP(S). IPv6 will hopefully change that, and native programs will get another reprieve from WebApps. (I also have to point out that WebApps still mostly suck - and are they even used yet on mobile/wearables/VR ?)

nottorp 6y ago

When it doesn't make business sense to embrace and extend any more.

fallat 6y ago

I will take the replies and create a response post tonight.

mdip 6y ago

> ... used the Web since 2007 Am I really that old? Something that was stated, but not directly, is what's driven a lot of the latest additions to the browser: Cross-platform development. The operating system provided the abstractions around various hardware implementations in the past, which meant being able to (possibly) run your software, if re-compiled, on different hardware platforms (it reality, some OSes weren't available except on a single platform and others were frequently not straight forward to get compiling on different platforms). The browser is providing abstractions to the same hardware around various OS implementations. Add in WebAssembly, and you can pick from a number of languages -- including those that were traditionally used to write apps not designed to run in browsers. But to the quote: I was cleaning out my basement, which quickly became a "treasure hunt" of sorts since I've been storing things that I took from my parents' house when I moved out. My favorite find was a CD-ROM that was basically a CD-based Search Engine of the Internet. It was "Designed for Windows 95" and was probably a free disc given to me by a vendor rep when I worked at CompUSA in my teens. I'm tempted to see how many of these actual sites still exist (I'm sure none of the deep links do). Incredibly, these things probably sold very well back then. It even came with a browser (neither Mosaic nor Netscape; no clue who's it is).

willseth 6y ago

The author has very limited knowledge of the web, its history, or the W3C's ambitions (not to mention WHATWG, IETF, ECMA, etc.). His view of the web in general is bafflingly myopic, with an almost completely arbitrary understanding of the businesses discussed. It should come as no surprise that the conclusions he draws are non sequitur. So why has this been upvoted? Why is this on HN at all? It's ignorant nonsense. This should be obvious to anyone who's spent more than a few hours researching these topics. Please stop taking no-nothing randos seriously.

billyjobob 6y ago
It does say "As someone who has used the Web since 2007" so I guess it's written by a child? So probably we shouldn't be too harsh judging them.
ritchiea 6y ago
Lots of users vote based on the headline before actually reading the article. Lots of users even comment without reading the article. HN is an extremely large community these days.

Existenceblinks 6y ago

I have a complete unlovely thought; web browser needs a rewrite with thin web specs (imagine javascript the good part book meme). People think it's impossible to rewrite, but I believe it's only because it's a mess. Inventing a new programming language and new DOM api that works greatly with webassembly probably worth it, consider that the next 10 years the web is still not going anywhere. You may laugh at Flash .. but hey do you love "hydration" thing? I'd love to have an isomorphic?(terrible name) template engine that rendered as text as well as client program binary. Do you love OS based form control that can't be styled? I'd love a complete separate controls from OS, clean the contract between widget and actual system api. Just stop caring (coercing) how it looks, just provide api to enforce how its functionality works, we already have these bunch of aria- thing. The question is what block the progress of making it complete, what kind of problem? I believe it's not a technical problem. Maybe incentive?

acoard 6y ago
It's both technical and incentive. The problem is that so much of technology is standing on the shoulders of giants. If we re-write, how much of the giants do we decide to lop-off? For example, could you get Google, Apple, Mozilla, to all agree where to draw the line? Should we still use HTTP? Should everything use QUIC? An entirely new standard? What about DNS? But let's say they do agree. Then what, they create entirely new browsers (I'll call Chrome2, Safari2, etc). Maybe they don't even use the DOM or HTML at all, but an entirely new setup. A side note – if the web was re-built today it probably wouldn't have "View Source." As you said, I imagine it'd lean into WASM and "isomorphic" code, which could just be WebGL/canvas/some new paradigm/etc. So then what... everyone says abandon my original website and download a new browser to access the new one? Because that's what throwing out backwards compatibility looks like. And that seems like a non-starter to most people. Browsers are too big to fail; too big to be thrown out and re-written. Our hope is incremental, not revolutionary. Things like WASM in browsers let us pursue new paradigms while also being backwards compatible. Lastly, I'd say the incentives for a new web would be worse than an old web. The original web was written by tinkerers and tech startups like Netscape/Mozilla. The new web would be written by multibillion dollar companies who like and know how to build moats. It would have microtransactions (not necessarily bad by itself) and I worry would look more like an app store.

tabtab 6y ago

I believe the argument is that Chrome's code-base is becoming the de-facto browser standard, which will allegedly simplify things by making a stable platform. But it's been hard to predict browser trends. Look how Chrome came out of nowhere and knocked Microsoft's off the top. Netscape also got sucker-punched. Who knows what will punch next. Web standards still have some glaring gaps that I'm not sure the current approach can tame. I propose "the web" be split into three standards: 1. Media/Games/Art/Video 2. Documents (existing HTML may be good enough with minor tweeks) 3. Productivity: CRUD and Data oriented interactive GUI By splitting, each standard can FOCUS on doing what it does best rather than the watered-down one-size-fits-all we have now. A given browser may support all 3, somewhat like how Java Applets and Flash could run inside of HTML documents. (They may be pluggins and/or independent browsers.) But unlike Applets & Flash, the standards wouldn't try to be full virtual OS's. Dump as much on the server as you can to keep client standards lean, and reduce the need for scripting by including common behaviors. For example, a common GUI idiom is for a button to activate a window or form. Rather than rely on scripting, that action could be built into the GUI markup: <button openTarget="myWindow"> Click to open My Window </button> Or maybe: <button label="Click to open My Window et. al."> <action> <setFocus target="myWindow"> <setFocus target="anotherWindow" modal="false"> </action> </button> The point is that if common GUI idioms and behaviors are built into the GUI markup standard, scripting (typically JavaScript) is not needed as often. The GUI Markup Language would be state-friendly and natively interactive, unlike HTML.

moritzwarhier 6y ago

I hope this doesn't classify as trolling; But isn't it ironic that the author uses a GitHub gist to share this plain text essay? Had to click "View Raw" to read it on mobile. I know the essay is about more topics than just documents vs apps. Still that shows to me that "The principle of least power" doesn't easily apply to technology choices of web users. GitHub was probably more convenient for the author than uploading a txt file somewhere.

akrymski 6y ago

Alternative end game: the browser dies slowly, gradually being replaced by native apps. We're already seeing this: consumers prefer native apps on their phones for virtually everything: shopping, socialising, reading news. Surely the future us will laugh at web browsers as some ancient tech that used these weird markup languages to layout content that's monetised with ads into pages that you'd navigate back and forth in order to emulate an application? New web browser engines are now impossible to implement due to the sheer size of specs. Surely we are nearing the end game?

sfink 6y ago
Consumers prefer native navigation (go straight to an app, not home -> browser -> app) to apps that in practice are native wrappers around web views.