Live data from Hacker News

Link rot and content drift are endemic to the web

theatlantic.com

201–210 of 220 posts

Re: Link rot and content drift are endemic to the web

#201

Earlier quoted context omitted.

I’m happy to give YouTube my money to provide a real alternative to ad-supported models. Plus the vast majority of videos I watch id never want to see again.

You're not really providing YouTube an alternative to ad-supported model - you're only proving that lots of obnoxious ads can convert people into paying customers.

Yes, and how dare a business desire paying customers.

Re: Link rot and content drift are endemic to the web

#202

Libraries (like Internet Archive for one) seem like the logical 'permalink' curators of information collections. I'm thinking, for example, of a large regional history site created by a a 501(c)(3) corporation. [ https://www.historylink.org/ ] Such efforts deserve a guaranteed permanent home with permanent funding. Another example: hundreds of 'Old-time-radio' programs and early TV shows have only emerged and survive…

I think about used book stores, libraries, the library of congress, and just shelves in people's homes. All of that allows books to survive, when they're printed enough, and people actually want them.

Lots of books, magazines, newspapers do not survive because nobody wants to read them. We don't keep a super-archive of every piece of paper ever published, because we acknowledge that it's OK for things to die. Important things we keep, useless things we throw away. It's irrational to hoard garbage.

So really, all we need is to proactively decide what we do want to save, and save it as soon as it's published. Maybe also provide a "digital shelf" for people to keep their own copies, and a standard way to search for and distribute that information, like NNTP + Gnutella for e-books (along with some features to avoid being sued a la DMCA). The rest should be allowed to rot like fallen logs in the woods.

Re: Link rot and content drift are endemic to the web

#204
post #168

Earlier quoted context omitted.

I set up a server specifically for downloading Youtube videos from all my playlists on a daily basis just for this reason. At some point I got fed up at seeing all the missing videos on my playlists (and not even knowing what was removed)

What do you use for automation? My first instinct is cron jobs running youtube-dl scripts against a set of playlist URLs, but I'd be interested to hear more.

Yep, you basically just described the setup, nothing super fancy :)

Cron job which runs a batch script at midnight which feeds youtube-dl some playlists URLS's which it downloads to a HDD. I also have nextcloud running which has access to the directory the videos are saved to, so I can easily share them if I want to.

Re: Link rot and content drift are endemic to the web

#205
post #201

Earlier quoted context omitted.

You're not really providing YouTube an alternative to ad-supported model - you're only proving that lots of obnoxious ads can convert people into paying customers.

Yes, and how dare a business desire paying customers.

My point is that converting is a signal promoting more ads, not less.

Re: Link rot and content drift are endemic to the web

#206

Hashes are the answer. How they get implemented in solving this problem is the question

What specifically do you hash? How do you hash content which is programmatically determined and changes on every page load? How do you account for the same work in multiple versions, translations, or updates, strictly using hashes? (Note that a chained hash, e.g., a git history, is not a strict use of hashes, though it most definitely does use hashes.)

>programmatically determined and changes

Ethereum's Solidity is a great real world example. The instructions are compiled into a binary, and that binary can be addressed via its cryptographic digest.

Changes in state are easily represented as instructions.

>multiple versions, translations, or updates, strictly using hashes?

The same way a git repo does it now: Good software design.

>is not a strict use of hashes

Huh? I'm not sure what your trying to convey. Are you referring to the fact that git uses diffs between commits? At any commit, a repo may be re-hashed and the current state completely represented by a digest.

Of all things to worry about being represented as a digest, software seems to be one of the smallest concerns.

Re: Link rot and content drift are endemic to the web

#207

Earlier quoted context omitted.

What specifically do you hash? How do you hash content which is programmatically determined and changes on every page load? How do you account for the same work in multiple versions, translations, or updates, strictly using hashes? (Note that a chained hash, e.g., a git history, is not a strict use of hashes, though it most definitely does use hashes.)

>programmatically determined and changes Ethereum's Solidity is a great real world example. The instructions are compiled into a binary, and that binary can be addressed via its cryptographic digest. Changes in state are easily represented as instructions. >multiple versions, translations, or updates, strictly using hashes? The same way a git repo does it now: Good software design. >is not a strict use of hashes Huh?…

Git chain hashes.

Merkle trees are useful.

They include hashes. They are not simply hashes.

Content-addressable storage robust against variations is another approach.

Re: Link rot and content drift are endemic to the web

#208

Earlier quoted context omitted.

>programmatically determined and changes Ethereum's Solidity is a great real world example. The instructions are compiled into a binary, and that binary can be addressed via its cryptographic digest. Changes in state are easily represented as instructions. >multiple versions, translations, or updates, strictly using hashes? The same way a git repo does it now: Good software design. >is not a strict use of hashes Huh?…

Git chain hashes. Merkle trees are useful. They include hashes. They are not simply hashes. Content-addressable storage robust against variations is another approach.

I'm at a loss for your criticism.

Didn't the original post say, "How they [hashes] get implemented in solving this problem is the question"?

Merkle trees implement cryptographic hashes.

>Git chain hashes.

Git chains hashes? Yes, of course, along with diffs.

How the permanent web will implement digests in various applications is the question. Whatever the answer is for link rot, the guaranteed solution will implement and depend upon cryptographic hashing algorithms. There are already many examples of this class of problem being solved by utilizing hashes.

Re: Link rot and content drift are endemic to the web

#209

Earlier quoted context omitted.

I, too, was fascinated by a "web without browsers" (or with other kinds of browsers, really) until i stepped into the community and realized that community was a proto-nazi cesspit full of misogynistic attitudes. Maybe now that times have passed, people have died who posthumously admitted their preferences for white supremacy (and heavily bitcoin-supported that), and whole projects have been renamed, there can be a m…

> I, too, was fascinated by a "web without browsers" (or with other kinds of browsers, really) until i stepped into the community and realized that community was a proto-nazi cesspit full of misogynistic attitudes. Could you expand on that? It seems a bit out of (anti-)left field, so to speak.

The name and logo of Weboob were specifically chosen to be a provocative pair of boobs. There were attempts to sexualize and feminize all kinds of things through "boobs" throughout the project. You could find comments about "faggots" and "niggers" in the code. All in all it was not a very friendly community in my standards. If you wanna know more about the recent revelations/scandals, you can look up Pankkake's suicide notes and the big Bitcoin donations he made to the US alt right and neo-nazi networks (he's from France).

> It seems a bit out of (anti-)left field, so to speak.

Sorry but english is not my native language. What does that mean?

Post reply on HN