Hacker News Reader

Comment by superkuh | Hacker News Reader

Everything there is correct except the response. Don't give up on the web. Just stop using websites that implement user hostile things like web components. It sucks but not nearly as much as switching to a new protocol. The corporate web (like github/youtube/etc) will become inaccessible but there's plenty of real websites out there. At least until the search engines start filtering them out for not leashing themselves to some TLS cert feudal lord.

On this subject, see also: https://linuxreviews.org/Modern_Web_Standards_Are_Leaving_Ni... "Modern Web Standards Are Leaving Niche Web Browsers Behind"

View on HN

I think giving up on the web is doubly wrong, because these days, the web is constantly changing. Just this week we've seen how QUIC/HTTP3 is a big thing, and it's a leap from HTTP2, and a huge leap from 1.1. We've seen small revolutions all over the place. Yeah, it's not clear which ones will "stick" but I think there's still time for meaningful change that doesn't require boiling the ocean.

Isn't QUIC/HTTP3 a protocol that improves the performance of pipelining which is something that very bloated webpages benefit from? Isn't that something we should move away from?

Sometimes you want to load a bunch of thumbnails. Pipelining is pretty good.

Though you can pipeline just fine on HTTP/1.1 if people would actually enable it.

Pipelining is something that helps everyone other than the gigantic new breed of "single page web apps" that are actually just single gigantic JS files.

We should definitely definitely definitely be trying to help pipelining. Unless you send only one resource, pipelining is probably a huge win.

And it definitely, from the server side, makes things way easier. As you can get a bunch of requests & start satisfying them in whatever order works for you. This isn't like HTTP 1.1 Pipelining, which was in order: now connected systems can out of order send whatever data they have, without having to buffer it & wait for the first asked for data to clear the pipe.

This is massively massively helpful tech we are creating. Attaching it to "very bloated webpages", & implying this is something "we should move away from" is a high level of shade that is not at all in any way deserved nor fair nor accurate.

I was arguing less for a specific tool, technology, or whatever, and more that the web as platform is constantly in flux and so "throwing it away" makes no sense.

You don't stop using the Interstate system bc a stretch of road is bad.

Speaking specifically as a Gemini enthusiast:

We don't want "the web, but less of it". Do you not realize that "less of it" is still a whole hell of a lot of web? Take a look at Netsurf: 13 years of development and half a million lines of C and it's probably the most advanced rendering engine after the big three. And yet, one of my websites, sourcehut.org, renders like this:

https://l.sr.ht/_2pv.png

On WebEngine, it should look like this:

https://l.sr.ht/bSdI.png

And this is one of the most conservative websites on the modern internet. The problem with using the web, but less of it, is that the web is still incredibly complicated and even conservative websites require a massive engineering undertaking to render properly, and it just takes one link to a more liberal page to ruin the experience entirely.

Gemini is a separate, self-contained network, based on a much simpler philosophy. I wrote my Gemini client in two afternoons, and it's already pleasant to use with all Gemini pages.

We don't want the web, but less. We want something else entirely.

> The problem with using the web, but less of it, is that the web is still incredibly complicated and even conservative websites require a massive engineering undertaking to render properly

The main problem is that "less of it" isn't well defined, so any website, while still appearing to require little of the web, requires its own subset. If there is a clear demarcation about which features are supported and which aren't, then websites can be built targeting this smaller subset, and everyone would be happy. There is no need to reinvent the wheel and start totally from scratch. Reading the Gemini specification, it seems like a NIH of HTTP. What's wrong about a HTTP 1.0 server?

If you start with the web but less, you're never going to shake the web. You're still fundamentally buying into the same ecosystem. I don't want to be a subset of a broader ecosystem which I hate. I want a new ecosystem.

Is there anything on Gemini that prevents centralization?

Because lets suppose this will happen, everyone now goes for the sweet land of Gemini.

At the beginning it will be just like the web in the nineties, full of indies web sites, a dozen open source clients, that implements the protocol, yay!

Profit right? Here now enter the 2000's of Gemini.. startups begin to grow stronger most of them mediating and selling the content of others that they get for free.

The individual gemini sites begin to fade, the startups create Gemini clients with a lot of new features starting to turn impossible for new implementations.

(Now its the time we start to ask ourselves, where did we see that movie?)

So whats the point of a new protocol or standard, if it doesnt solve the real issues we are facing nowadays, and the only solution it provides its to manipulate people emotions that we should get back to that romantic era of the internet, the 'belle epoque'.

"Its hard to solve the modern issues? get your time machine pal, lets get to the nineties and start everything again without solving the core issues, and if by chance we happen to thrive, just let those big problems to be solved by your grandson"

There is only one thing we should be doing right now: Decentralize the internet, making "blue seas" for new things to thrive making the moves of FAANGS to centralize and control data, information and computation futile by getting people out of a game they control.

A solution like Gemini would reset the game, yes, but only to have the same problems later, giving the creators were not trying to solve them in the first place.

Gemini is trying to solve a problem by using denial, and as we know, it doesnt actually solve the problems. It just make them disappear temporarily to hit us again in a near future.

You can make a Gemini client in two weeks? thats great, this is probably what you would take to make browsers in the nineties..

>startups create Gemini clients with a lot of new features starting to turn impossible for new implementations

I don't think this is going to happen. We're pushing very hard for a culture which embraces the limitations of the protocol. Anyone who even thinks about extensions has the fury of the mailing list wrought upon them as soon as they propose them.

And Gemini is so simple that it's difficult to monetize. There's not much corporate incentive here. I think that corporations will continue to use the web, because the web was basically hand-designed for corporate needs - even going so far as to do so at the expense of the users.

>You can make a Gemini client in two weeks?

Two afternoons. And a stated goal of the specification is not to add substantial changes.

We can decentralize the internet and serve Gemini over our new, decentralized space. We can work on more than just one of THE MOST IMPORTANT PROBLEMS™ at a time, you know.

> We can decentralize the internet and serve Gemini over our new, decentralized space. We can work on more than just one of THE MOST IMPORTANT PROBLEMS™ at a time, you know.

People are creating "killer apps" because the web evolved to its current state. Theres a lot of bloat of course, preventing new browsers to emerge and this is a real issue.

But the serious effort would be to scrap most of this bloat and create a leaner web standard that would be easier to implement and yet offer advantages for people developing in it. Seriously i dont know WHY Tim Berners Lee think the solution is something like Solid, but not to create a leaner version of his own creation.

Which developer will be ok to trade what he/she have today for http0.9+markdown? None, because is too little. Maybe the Linux manuals will find they are more popular there..

The creators will not want to add the solutions to problems that will appear, but this will just make it a small niche.

Im trying hard to launch a solution to decentralization that at the same time looks good for the developer and the user, and its really hard, because the status quo is very sophisticated. Its David vs. Goliath and there's no way to get there except by facing Goliath.

If the guys behind this at least optimize it to be "a protocol for terminal/console browsers" i'm pretty sure that they would find a niche to live and start, but not as something that when looking over the shoulders of the current web presents no advantages to developers or users.

They say they are not competing with the web, and that's ok, but they should at least do it in a way were the thing they created manage to exist. Otherwise it just looks like noise.

Sorry if i take this too serious, but its because we are all facing real danger now, the world we are heading to doesnt look pretty if we keep in this direction.

And more noise right now, just takes peoples attention from where we are really needing anyone help, and i think its a disservice, at least right now. Because we get more divided, and we are already just David, it will be even harder by being half-of-David.

Its cool if people dont care, or are in another phase in life, but at least how about to play in another playground, why waste peoples time in a arena and a time we need all the effort we can get?

> Anyone who even thinks about extensions has the fury of the mailing list wrought upon them as soon as they propose them.

If a central entity decided they could capture a new market by shipping “Gemini++: The One With DRM So You Can Watch Netflix”, I’m not sure why the “fury of the mailing list” would be a concern for them.

Given that users pick their clients, the folks pushing for fewer features would end up splitting the userbase between “people who agreed with them” and “people who didn’t”.

For an example of how this goes if a medium gets popular, the best example may in fact be HTTP/HTML/etc.

At what point is it okay to depend on a feature?

> not leashing themselves to some TLS cert feudal lord

Really?

You can get one free and change at any time.

You can change what? How many companies are giving away SSL certificates?

You can change if and who you get a cert from.

If you're really that worried about Let's Encrypt turning evil, you could stop having one when that actually happens. Or get a free one from many DNS providers. Or pay $5 for one. There is zero lock-in, and keeping traffic encrypted is a good thing.

And just for encryption, you can self-sign.

Obviously that doesn't work for the web, but "TLS cert feudal lord" doesn't track for me at all.

Self-signing is exactly what I've done for all my sites.

But browsers are getting more and more restrictive about what they will allow their users to see. Right now technically ignorant people will be so confused and frightened by warnings they click away. That's okay, I don't care about those people. But for many it's a deal breaker. I give it about 3 years before the big corporate browsers stop allowing HTTP at all. Search engines already deprioritize HTTP and eventually they'll just delist. And it won't hurt their income one bit.

The number of sites that don't HTTP and just 304 to HTTPS is already frightening.

Good. It's more important to cater to general privacy/security than to the kind of guy that already has a server running but can't be bothered to install certbot.

For a non-expert setting up a website from scratch, they can get the whole stack set up automatically and it won't make a difference to them. Especially since they're probably on shared hosting; shared hosting is much closer to the ideal of having the website be a folder of html than running an entire server.

Sure, as long as we have the tooling available for students to get there too (which, I think we do, or pretty quickly can).

Sometimes the solution to a problem that you should not have only gets in the middle of solving the root issue, giving some people fall for the motto "problem solved, look the other way" where the initial problem gets completely forgotten.

This is the reason why i fear that if we dont act fast for some things, younger generations without the proper historical context, will only follow through a path that treat them like human cattle, only to serve the feudal FAANG rulers.

The cloud + dumb web/app clients duo is really dangerous to us all, and unfortunately this is the path we are heading to.

No control, no freedom, no power, no voice, working as a dumb colony of mere tech consumers.

Centralization should be the enemy of anyone who fight for those values.

http://john.ankarstrom.se/html2/

>Sustainability is also the one of the many reasons why my web site is served over HTTP and not HTTPS. The problem with HTTPS is that you need to get a certificate from a trusted authority, which either costs money or is free but needs to be updated every month, and even if you pay for it, you need to renew the certificate every year or so. If the operation of your blog depends on a monthly certificate renewal from Let's Encrypt – even if the renewal is set up to be automatic – it is less sustainable than serving your site over HTTP. HTTP will always be around, will always be supported and will never require a certificate.

>Another problem with HTTPS is that it is very difficult to revert back to HTTP once you have started using HTTPS. Every outside link to your web site will be incorrect, and you can't redirect HTTPS traffic to HTTP without a valid certificate. In other words, once you start using HTTPS, you will be forever dependent on the certificate renewal process.

Even the people moving away from the regular web will most likely be switch-hitting. Using MegaGloboChromeWeb for all that stuff where you have to do it, but also trying separate approaches.

Gemini might be one of them, for applications where hyperlinked text offerings make sense.

But betting too much on a single protocol is what got us in this place.

For example, this site makes no sense to me at all, technically speaking. There's nothing HN offers that couldn't be done with Usenet or a mailing list. Alas...

Reply on news.ycombinator.com