Modern Web Woes - I'm mad at the internet

  • 🐕 I am attempting to get the site runnning as fast as possible. If you are experiencing slow page load times, please report it.
Razel said:
Out of curiosity, what sorts of novel projects would you like to see? One that I can think of off of the top of my head is something like SoundCloud, where artists can share/promote/sell their music without relying on third party labels, service providers, etc.
Click to expand...
A SoundCloud alternative sounds cool to me. Just about anything that has a concept of status updates and followers should be a good fit, including basic things like comments sections on personal websites.

Git is something I'd like to see tackled. Git itself is built around decentralisation, but all the stuff on top of git (publishing, pull requesting, issue tracking, code reviewing) is completely centralised. An ActivityPub solution in this space would be much tighter with the philosophy of the underlying technology.

Next, let me off the web. I don't want my web browser to be my operating system. I have Emacs for that, and it should be as much a first-class citizen in the ActivityPub ecosystem as a browser.

And rather than bespoke ActivityPub clients, I want a single client that lets me work with arbitrary ActivityStreams, so I can approve a pull request, check out the latest release from my favourite music artist, and tell the gunt that he's got a micropenis all from the same place.
 
  • Like
Reactions: JJLiautaud
Ahriman said:
Redpill me on IPFS. Is it as good as it sounds?
Click to expand...
What about it sounds good? I'm not really thrilled by a fancier iteration of "What if every link was a torrent instead of a URL".
The resilience sounds nice in theory, but in practice I'm not sure it's better than conventional archiving. What happens if one little block of the archive goes missing? How long would it be before anyone even knows it's gone? What if the last guy who had it didn't even know he had it? What guarantees that any particular content will stay up if no one has any specific commitment to make sure it's still up?
As for resistance against government interference, I'm skeptical of that too. Remember Napster? Governments can and will go after individual users P2P sharing forbidden content if the political will is there. I assume you're not intended to be running this on Tor at all times.
Finally, how does this work if you're running a node? Who decides what's on your node, is it automatically distributed? Could The Glowies easily poison your machine with CP, Tiananmen Square footage, and so on? What's to stop administrators or governments from publishing a list of "bad hashes" and saying anyone mirroring these hashes (or federating with someone who does) will get their social credit score docked?

There may well be answers to all these already, I don't know. Those are just my impressions from reading the website - I didn't dig down into the specs or implementations.
 
  • Agree
  • Thunk-Provoking
Reactions: Toolbox, ScatmansWorld, Return of the Freaker and 2 others
Kosher Dill said:
What about it sounds good? I'm not really thrilled by a fancier iteration of "What if every link was a torrent instead of a URL".
The resilience sounds nice in theory, but in practice I'm not sure it's better than conventional archiving. What happens if one little block of the archive goes missing? How long would it be before anyone even knows it's gone? What if the last guy who had it didn't even know he had it? What guarantees that any particular content will stay up if no one has any specific commitment to make sure it's still up?
As for resistance against government interference, I'm skeptical of that too. Remember Napster? Governments can and will go after individual users P2P sharing forbidden content if the political will is there. I assume you're not intended to be running this on Tor at all times.
Finally, how does this work if you're running a node? Who decides what's on your node, is it automatically distributed? Could The Glowies easily poison your machine with CP, Tiananmen Square footage, and so on? What's to stop administrators or governments from publishing a list of "bad hashes" and saying anyone mirroring these hashes (or federating with someone who does) will get their social credit score docked?

There may well be answers to all these already, I don't know. Those are just my impressions from reading the website - I didn't dig down into the specs or implementations.
Click to expand...
I dunno, I saw it floating around every now and then as an alternative to the bloatnet we use these days but haven't looked deep into it.


I'll check out this talk later.
 
Kosher Dill said:
What about it sounds good? I'm not really thrilled by a fancier iteration of "What if every link was a torrent instead of a URL".
Click to expand...
I think torrents and magnet links are way cooler for distributing static content than what Tim Berners Lee invented. HTTP was never just about distributing static content though, so it's not a good comparison.

IPFS is what happens when you think torrents and magnet links are really cool, and then imagine turning that into a global content-addressed monotonic filesystem, which you can navigate with a browser. The Bittorrent project had floated this idea before IPFS with Maelstrom, and now with BTFS. The idea is pretty obvious and obviously awesome, provided they can make it decently efficient (IPFS has been horrible on this front). I hope it takes off in one form or another.

Kosher Dill said:
The resilience sounds nice in theory, but in practice I'm not sure it's better than conventional archiving. What happens if one little block of the archive goes missing? How long would it be before anyone even knows it's gone? What if the last guy who had it didn't even know he had it? What guarantees that any particular content will stay up if no one has any specific commitment to make sure it's still up? As for resistance against government interference, I'm skeptical of that too. Remember Napster? Governments can and will go after individual users P2P sharing forbidden content if the political will is there. I assume you're not intended to be running this on Tor at all times. Finally, how does this work if you're running a node? Who decides what's on your node, is it automatically distributed? Could The Glowies easily poison your machine with CP, Tiananmen Square footage, and so on? What's to stop administrators or governments from publishing a list of "bad hashes" and saying anyone mirroring these hashes (or federating with someone who does) will get their social credit score docked?
Click to expand...
When running IPFS, you decide what you want to serve by "pinning". This is where it differs from GNUnet, where you automatically serve whatever you've downloaded. There's a big tradeoff there. I'd expect GNUnet to be more resilient if widely adopted, in terms of not losing data, because it's got more seeders by default. GNUnet includes an anonymous protocol so that you're not guilty of distributing some evil that you unknowingly fetched.

As for resilience, something like IPFS doesn't have to replace an archive. You can still have multiple archive services that agree to store the entirety of the filesystem and serve as permanent seeders. Missing pieces can be detected by crawling the DHT, and archives can be automatically mirrored if there's worry that the men with guns are going to come knocking.

That's the theory anyway. I've read suggestions that there are just fundamental inefficiencies with DHTs that might mean you're always better off centralising. I hope that's not the case
 
Last edited:
  • Informative
  • Like
Reactions: Super Dilator Duel 3D, Fission Mailed, Toolbox and 3 others
cecograph said:
IPFS is what happens when you think torrents and magnet links are really cool, and then imagine turning that into a global content-addressed monotonic filesystem
Click to expand...
Well, whenever I think of torrents, I immediately think of a 200GB file that's useless because it's missing one 4KB chunk in the middle.
Torrents work fine for things like Linux distros where there are always a billion nerds seeding them, but I don't think they scale down well to the low end where it's only a few people. If I had to chose between a few unreliable seeders and a few unreliable web hosters, I'd obviously choose the web hosters.

cecograph said:
When running IPFS, you decide what you want to serve by "pinning".
Click to expand...
Ah, this is the part I was missing. It looks like you do also serve up whatever's in your "most recently requested" cache whether you like it or not, but I imagine that's easy to deal with one way or another.

I guess the question is: why would I invest in the computing resources to pin my data in multiple copies all over the place (or hire a pinning company that does this), over just using a traditional web CDN? It looks like at least a few of these pinning services do in fact use a CDN behind the scenes - why not cut out the middleman?
Or put another way: if the CDN is already going to distribute the 20 HTTP requests generated by loading this page out to its servers in some optimal way, is it really that much better to torrent the data from 200 randoms from IPFS instead?
 
  • Like
Reactions: Super Dilator Duel 3D
Kosher Dill said:
Well, whenever I think of torrents, I immediately think of a 200GB file that's useless because it's missing one 4KB chunk in the middle.
Torrents work fine for things like Linux distros where there are always a billion nerds seeding them, but I don't think they scale down well to the low end where it's only a few people. If I had to chose between a few unreliable seeders and a few unreliable web hosters, I'd obviously choose the web hosters.
Click to expand...
Linux distros will always have seeders because the maintainers of those distros will always be seeding. The original content providers want the torrent to stay healthy. The rest of torrent traffic is for pirated material, and the copyright holders of that content want no part in it (they want it shut down).

Kosher Dill said:
Ah, this is the part I was missing. It looks like you do also serve up whatever's in your "most recently requested" cache whether you like it or not, but I imagine that's easy to deal with one way or another.
Click to expand...
That sounds like it must be new then. Unless my memory is screwy, when I was trying it out a few years ago, I don't think they ran any policy like that.

Kosher Dill said:
I guess the question is: why would I invest in the computing resources to pin my data in multiple copies all over the place (or hire a pinning company that does this), over just using a traditional web CDN? It looks like at least a few of these pinning services do in fact use a CDN behind the scenes - why not cut out the middleman?
Or put another way: if the CDN is already going to distribute the 20 HTTP requests generated by loading this page out to its servers in some optimal way, is it really that much better to torrent the data from 200 randoms from IPFS instead?
Click to expand...
The ideal is that you're not investing resources to pin: your downloaders are volunteering to do that for you, and you can trust them to do this because the only authority on the content's validity is the hash that was asked for. It's just like torrents.

If you're a web company, this is all very stupid, and you should pay for traditional infrastructure. You can recoup your costs by mining user data and running the SEO playbook to get paid shoving as much advertising in your users' faces as possible. In other words, Modern Web Woes.
 
Last edited:
cecograph said:
The ideal is that you're not investing resources to pin
Click to expand...
Right, but I'm just not seeing that as viable for the small-time publisher - to bring it back around, the ones squeezed out by those modern web woes. If you have content that's obscure or unpopular, you're going to have to pay one way or another, whether that's web hosting or pinning.

Perhaps the root of all woes is that nothing scales down well to the individual user anymore.
 
  • Like
Reactions: JJLiautaud
Kosher Dill said:
Right, but I'm just not seeing that as viable for the small-time publisher - to bring it back around, the ones squeezed out by those modern web woes. If you have content that's obscure or unpopular, you're going to have to pay one way or another, whether that's web hosting or pinning.

Perhaps the root of all woes is that nothing scales down well to the individual user anymore.
Click to expand...
That's not my understanding. My understanding is that, even if you're super small fry, you can pin your initial content with the dribble of bandwidth provided from your home ISP to the raspberry pi in your bedroom. If people find your content and like it, some will pin it, and use their bandwidth to distribute it. Ideally, this should be incentivised, or we should pick something other than IPFS where you automatically P2P and we use a protocol which keeps peers anonymous.

So even if your shitty blog is served from a raspberry pi in your bedroom, you don't run the risk of a hug-of-death if, God forbid, you get on the front page of a news aggregator. By the time that happens, most people will be downloading your blog post from peers who've pinned the content, not from you.

This is all in principle, and I'm not sure it's close to proven.

This might not be the model of a small corporate publisher. But, in principle, it's the model for the people who produce almost all the stuff I watch on youtube, on most of the blogs I read, and the model for the unmonetised web of yester-yore.
 
  • Like
Reactions: Super Dilator Duel 3D, ScatmansWorld and Coffee Shits
I mentioned it before, but that sans serif "Current Year" font soyed sites like to use - especially Twitter - is almost as annoying as using Comic Sans for anything but comics can be.
 
  • Like
Reactions: Puff
I think it's a combination of two factors. One, designing everything to be crammed down onto a tiny screen means that soulless sans-serifs are naturally favored. And two, real fonts are expensive to license for web or app use. Most professional entities are just going to go with the open-source free Google stuff or stick with "Browser Default".
 
  • Like
  • Informative
Reactions: Smaug's Smokey Hole and ToroidalBoat
You must log in or register to reply here.
Back