Hacker News Reader

Comment by vwpolo3 | Hacker News Reader

Agreed, also while they can not be censored by the administrator of the website, they can still be censored (or moderated) by the Plugin Owner, right?

View on HN

If this got popular, and an administrator wanted to disable comments, one thing that they could do is add a unique random prefix to every request to a URL. The web server would just generate a random prefix for each link it produces, and strip the random prefix from the path in every request it receives.

For example, HN's front page contains a set of URLs for each item. This item would be https://news.ycombinator.com/b149012c-56f4-4859-82b8-1f3f83e... for one request, and https://news.ycombinator.com/73e78d90-cc25-4091-8ffa-94783a2... for the next request. Both requests would get mapped to the same page, and serve up the same content (except with another uuid inserted into links).

Every comment would go into a unique SHA1 bucket and never seen again.

If the content is to be found via search engines, then this may not be advisable, and the plugin owner could try the same techniques that say Google would to use the canonical URL.

Yeah, perhaps a better solution would be for the comments to be stored in something like ipfs? Though if i understand ipfs right, it works like torrents, meaning that unpopular comments will disappear over time.

This is the thing that I was thinking about. That a blockchain-like solution would better fit the goal of the extension. And I don't think this because I worry about censorship or free speech. I think the main benefit of this extension is that it creates an option to comment on sites where to creator of the site didn't make commenting possible.

However I do worry about moderation and trolls. Allowing people to make as much noise as they can is almost as damaging to free speech as censorship. I'm not really familiar with IPFS, what does it mean that unpopular comments disappear? Is it like seeding the only torrent (comments) that you like?

>Allowing people to make as much noise as they can is almost as damaging to free speech as censorship.

The thing is, to see that noise you would first have to be on the specific link, want to see comments on it, then go out of your way to install the addon if you haven't already. This is so much better for the people running the websites as they do not have to even acknowledge the comments' existence and are not obligated to moderate anything. Anyone who wants to comment or see them should know what they are getting into. They can grow a skin or uninstall the addon.

I'd be more concerned about spam than speech I dislike.

Is there any kind of rate-limiting to stop bad actors from making a million comments per second?

Not right now. But I'm hoping to come up with an equivalent of captcha that doesn't need 3rd parties. That way any comment will need a few seconds of the commenter's time (not CPU time).

Yeah AFAIK it works like torrents.

About the moderation, etc, this is really opt-in so you use it by your own choice and by "going in" you pretty much have to expect that like any other place where moderation is light or non-existent you'll see stuff you dislike.

>I'm not really familiar with IPFS, what does it mean that unpopular comments disappear? Is it like seeding the only torrent (comments) that you like?

Uploading to IPFS is like creating a torrent for any file you share. Someone always needs to be seeding that file for it to remain accessible to people.

So what OP meant by unpopular comments disappear is that nobody would be seeding them.

In fact, I seriously considered ipfs. Then I tried to patch its severe connectivity problems with webtorrent dht. I even considered dat. But internet consists of symmetric NATs, that make any such DHTs hardly possible. An extension based on js-ipfs, would need 1 GB of memory and would constantly hog CPU with some mysterious computations. It would have to keep tens of tcp connections open to maintain at least a resemblance of connectivity. Even then, if you published a comment via "ipfs add" on your laptop and tried to discover it via "ipfs get" on your phone, it would be completely unable to do so even after an hour. Yes, it is this bad. And even then, assuming ipfs add/get worked perfectly with negligible load on CPU and memory, you'd still to keep the list of comment hashes on some data server because in ipfs the key principle is immutability. No, ipns is not a solution. It does allow to change a single hash pointer, but you need to use a private key to do so. This won't scale beyond a single data server. I had big hopes on ipfs, but it really isn't there and given the fundamental problem - inability to implement fast DHT in the world of Symmetric NATs - I don't see how it can become better.

Reply on news.ycombinator.com