Hacker News Reader

> That doesn't mean we can't change the paradigm, but to do that, we have to stop defending the current model.

Here's the crux of our disagreement. You believe that the web is such a broken application platform that it is possible to convince enough vendors and people to get behind a better solution. However, I (despite your presumptuous implication that I'm a millenial), have been around long enough to know that will never happen. Web standards will continue iterating, and companies will continue building apps on the web, even the most powerful app platforms today such as iOS and Android for all their market power can not stop this force. The reason is because it's a platform that works. The man-millenia behind the web can not be reproduced and focused into a single organized effort. You might as well argue that we replace Linux with Plan 9, it doesn't matter how much passion you have and how sound your technical argument is, Linux, like the web, is entrenched. It's gone beyond the agency of individual humans and organizations to become an emergent effect.

That's not to say that the web might not some be supplanted by something better, but it won't come because of angry engineers wringing their hands about how terrible the web is. It will come from something unexpected that solves a different problem, but in a much simpler and more elegant way, and over time it will be the thin edge of the wedge where it evolves and develops into a web killer.

Maybe I'm just cynical and lack vision, perhaps you can go start a movement to prove me wrong. I'll happily eat my hat and rejoice at your accomplishments when that time comes.

"You believe that the web is such a broken application platform that it is possible to convince enough vendors and people to get behind a better solution. However, I...have been around long enough to know that will never happen."

"That's not to say that the web might not some be supplanted by something better..."

Whomever wrote the first paragraph of your comment should get in touch with the person who wrote the second paragraph.

OK, seriously, though, let's summarize:

1) Person says "web development sucks, here's why: $REASONS"

2) You reply: "it's the only truly cross-platform development environment ever"

3) I (and others) reply: "no, it really isn't. it isn't even a development environment, by any reasonable measure."

Now you're putting words in my mouth about convincing vendors and starting movements. I'm not trying to start a revolution here, just trying to counter the notion that we can't do any better than the pile of junk we've adopted. You don't have to love your captors!

I have no idea if someone will come up with a revolutionary, grand unified solution tomorrow, but I know that this process starts with the acknowledgement that what we have sucks, and that we have lots of examples of better solutions to work from. Hell...just having a well-defined set of 1995-era UI components defined as a standard would be a quantum leap forward in terms of application development.

The irony is I understand your qualitative opinion of the web, and I generally agree with it. What I believe makes you unable to see my argument is an inability to separate technical excellence from the market dynamics that govern adoption.

Declaring the web "not even a development environment" is just absolutist rhetoric that can in no way further the conversation. If you define "development environment" as a traditional GUI toolkit then your're just creating a tautology to satisfy your own outrage.

This is a great discussion. What is it about the English language that makes it so much easier to oppose someone than express nuances in general opinion? I would like to see more discussions like this based at implementation level, surely something valuable and innovative is being grasped at by both sides.

Lets be honest, the web is a developer environment in the same way that a paper aeroplane is a passenger plane. I mean I'm sure its possible to create a 747 from paper, but do you really want to?

Why is it that a new javascript framework pops up each week? Its because the web as a developer environment is deficient. Despite it being standardised so much stuff doesn't work without kludges in each browser

> It will come from something unexpected that solves a different problem, but in a much simpler and more elegant way, and over time it will be the thin edge of the wedge where it evolves and develops into a web killer.

So.. app stores?

It has already begun. The most popular webapps (Facebook, Twitter etc.) already have native clients in Android and iOS. I believe the majority of people already prefer and use the native FB/Twitter apps more often than accessing the FB/Twitter websites. So it's already obvious that native apps must be more convenient.

Right now however, app stores are a little clumsier to navigate compared to browsers.

For webapps:

• you have to open the browser,

• type in the address OR

• use a web search if you don't know the exact address.

But for apps:

• you have to open the app store,

• search for the app,

• potentially filter through unofficial third-party software,

• download the app, possibly after entering your credentials,

• navigate to the app icon,

• authorize any security permissions on startup (in the case of Android or badly-designed iOS apps.)

We just need the Big Three (Apple/Google/Microsoft) to actively acknowledge that app stores can supplant the-web-as-application-platform, and remove some of those hurdles.

Ideally an app store would be akin to searching for a website on Google.com (or duckduckgo.com) with a maximum of one extra click or tap between you and the app.

Apps should also be incrementally downloadable so they're immediately available for use just like a website, and Apple already has begun taking steps toward that with App Thinning.

Ultimately there's no reason why the OS and native apps shouldn't behave just like a web browser, because if web browsers keep advancing and evolving they WILL eventually become the OS, and the end result will be the same to what I'm suggesting anyway.

Currently though, both the native OS side and the web side exist in a state of neither-here-nor-there, considering how most people actually use their devices.

Not so much works, as they have found a way to monetize as a rent.

I'm in the middle of reading through The Unix-Haters Handbook and I must say that the arguments against web-as-application-platform (and the attempts to defend it) and are eerily similar to what this book reports about arguments against Unix that were circulated in the 80s and early 90s. Sadly, the fact Unix managed to a) win, and b) fuck up the computing world so badly that people don't even realize how much we've lost doesn't make me hopeful about the future of the web.

Bad stuff seems to win because it's more evolutionarily adapted than well thought out stuff. This happens to hold for programming languages too.

Your takeaway of computers going from niche industry to the single largest driver of global economic activity is that the bad stuff won? What an incredibly myopic conclusion.

The recent cross-communication between JavaScript, Elm, and Clojure has been incredibly fruitful but hasn't been noticed by the bitter die-hards. And really, almost all of it could've happened literally 15 years ago with Lisp if the Lisp community hadn't been dismissive, arrogant douchebags that considered JavaScript a worthless toy language.

What's truly sad is that some people would rather be abstractly right while producing nothing of value than work with the dominant paradigm and introduce useful concepts to it.

> Your takeaway of computers going from niche industry to the single largest driver of global economic activity is that the bad stuff won? What an incredibly myopic conclusion.

This did not happen thanks to Unix; if anything, you'd probably have to be grateful to Microsoft and Apple for introducing OSes that were end-user-usable. There's a reason the "year of Linux on Desktop" never happened and is always one year from now.

The point of The Unix-Haters Handbook, which also applies very much to modern web is that the so-called "advancement" didn't really bring anything new. It reinvented old things - things we knew how to do right - but in a broken way, full of half-assed hacks that got fossilized because everything else depends on it.

> And really, almost all of it could've happened literally 15 years ago with Lisp if the Lisp community hadn't been dismissive, arrogant douchebags that considered JavaScript a worthless toy language.

I don't know where you're getting that from, but it's probably a good opportunity to remind you that JavaScript was supposed to be Scheme twice, both time it didn't happen because Netscape wanted a Java-looking solution right fucking now to compete first with Java, and then with Microsoft, and somehow no-one thought to pause for the moment and maybe do it right.

(Also don't blame Lisp community for the fact that companies reinvented half of Lisp in XML. Rather ask yourself why most programmers think the history of programming is a linear progression of power from Assembler and C, and why they remain ignorant of anything that happened before ~1985.)

JavaScript got a bad rep because a) it was terribly broken (less so now), and b) because of all the stupid stuff people were writing in it those 15 years ago. But the current problems of the Web are not really the fault of JavaScript, but of the community moving forward at the speed of typing, without stopping for a second and thinking if those layers on layers on layers of complexity are actually needed or useful. Simple landing pages are now built on frameworks that are more complex than what we used to call "Enterprise Edition" 10 years ago.

Unix isn't for end users, it's for developers to build on top of to give things to end users. Linux on desktop already happened years and years ago if you work for a tech company, and that's probably about as far as it needs to go.

http://steve-yegge.blogspot.com/2006/04/lisp-is-not-acceptab...

This is Steve Yegge's understanding of the Lisp community, and I should clarify that I don't think the XML monstrosities we all work with are "all their fault", but that, on the whole, the Lisp community and enterprise coders were mutually antagonistic.

Javascript after ES3 really wasn't broken at all, just most people coding in it didn't know how to take advantage of it. No language can prevent someone dedicated to bad code architecture from writing bad code. A Lisp programmer would've found a lot of comfortable features and powerful patterns and been able to share them, but most of their efforts were wasted denouncing everyone outside of their tiny sect. The end result was that most people learning JavaScript were taught how to code like it was broken Java because most of the resources were written by Enterprise Java devs who didn't understand what a fundamentally impoverished language Java is.

Thanks for the link to that post. I'd also advise to read through its comment though - some people there, especially Pascal Costanza, point out quite a lot of problems that basically reduce it to ranting of a person who doesn't understand the language and the culture he's writing about ;).

Also, the influx of enterprise patterns into JavaScript is quite a recent phenomenon - personally, I blame Google (who, for a reason I can't understand to this day, embraces enterprise-level Java as their primary platform for everything...), but regardless, the problem with JavaScript culture is mostly that of very fast growth coupled with lack of experience and (probably unwilling) ignorance of the past. Since this community basically controls the Internet, it's hard for voices expressing some restraint and thoughtfulness to get trough the noise.

And I really do recommend The Unix-Haters Handbook. Funny thing is - over a decade ago, when I was acquainting myself with the Linux world (after many years of DOS and Windows experience), I've been noticing and complaining about various things that felt wrong or even asinine. Gradually I got convinced by people I considered smarter than me that those things are not bugs but features, they're how a Good Operating System works, etc. Only now I realize that my intuition back then was right, but I got Stockolm-syndromed to accept the insanity. Like most of the world. The sad thing is, there were better solutions in the past, which once again shows how IT is probably the only industry that's totally ignorant of its own history and constantly running in circles.

What part of UHH do you think is not obsolete? I dislike many aspects of Unix, but it seems like 90% of the book is either wrong, meaningless invective, or obsolete. It's hard to view it as of more than historical value -- which is a shame, because Unix is far from perfect.

To clarify, I'm about half of the way through the book, and I have found a couple arguably-correct points. Complaints about Usenet and Unix vendors have mostly gone the way of the dodo. The author consistently ignores any distinction between the filesystem and the rest of the OS, which may have been accurate at the time of writing, but hasn't been true for decades. Similarly, we don't distinguish between a given shell and Unix as a whole, even though the book makes explicit mention that other shells exist. And why there is a chapter on Usenet passes understanding. Suggesting that shell expansion be handled by an external library is equally bizarre. So far the valid points are: * command input syntax is inconsistent. I don't know that this has a feasible solution, but it is true. * tar specifically sucks * sendmail configuration sucks

The real crime of UHH is that it merely hates, it does not instruct. When we do find valid criticisms, there is no suggestion for how to fix things, or how other OSes are better at the same role. I've resigned myself to read the entirety, but for all the authors' complaints about not learning anything from history, one can only feel like they have themselves to blame.

If you know C++/Perl/PHP but don't know ES6, websockets, Node, etc, would you not say that your opinion about the web being a bad environment for rich applications might be in some way colored by your personal economic interests?

A common defensive mechanism among people with outdated skills is to try to delegitimize new frameworks and technologies in the hopes of convincing the broader community not to use things they don't know.

I'm inferring this from your arguments being driven by analogies and insinuations rather than concrete critique. It's not my intention to attack you personally, but an aggressively dismissive attitude towards unfamiliar concepts should be properly contextualized.

As for React, isn't it more likely that you don't know React very well, have never looked at its internals, and in general don't feel like you have the time or ability to learn much about modern web development?

If you build a few projects with React and still dislike it, good! Your critiques will be a lot more valid and useful at that point, whereas right now...yeah.

I know C++, used to know Perl, and know JavaScript pretty well. The fact that you actually name WebSockets as a technology sort of sums up the issue we are discussing here. WebSockets is not something that can be compared to C++ or even Node. It's a dumb hack which justifies its existence primarily by allowing web apps to circumvent corporate firewall policies.

The web is a joke of an app platform. Those of us who have wider experience of different kinds of programmings see some web devs struggling with this concept and conclude, I think quite reasonably, that the only plausible explanation is lack of experience. This is not due to "outdated skills" - I daresay everyone criticising the web as a platform in this thread has, in fact, written web apps. It's the opposite problem. It's to do with developers who haven't got the experience of older technologies having nothing to compare it too, so "web 2.2" gets compared to compared to "web 2.0" and it looks like progress.

And in case you're tempted to dismiss me too, I recently tried to write an app using Polymer. It sucked. The entire experience was godawful from beginning to end. Luckily the app in question didn't have to be a web app, so I started over with a real widget toolkit and got results that were significantly better in half the time.

I want to disagree with you more here, but Polymer really does suck.

I would be interested in a detailed explanation of why websockets are a "dumb hack." Duplex streams much more closely map to what web apps actually need to do than dealing with an HTTP request-response cycle. In what way is streaming a hack and making requests that were originally designed to serve up new pages not a hack?