Comment by cxr | Hacker News Reader

Comment by cxr | Hacker News Reader

Parent Full thread

Two under-utilized properties of URLs are also that:

- there's a near-infinite supply of them

- they support forward declaration

Together the practical upshot is that if you're having a conversation with someone or responding during the Q&A of a talk or whatever and you want to be able to say, "Yeah, we thought about that, and we have some information about it on our site—just visit acmeinitiative.example.com/skub," except that you haven't already written the /skub article yet, that doesn't preclude you from being able to say in the moment (i.e. live) that /skub is effective immediately now the designated handle for such an article, it's where the article will appear once you do write it, and it's how any interested party should retrieve it once it does appear—whenever that is. (The same goes for articles published by other people/organizations and other third-party resources that you want to reference—just mint a URL from your namespace on-the-fly and then whenever you get a chance, set up a redirect to whatever it is you wanted to link to.)

There are so many recordings (podcasts episodes, etc.) that I've listened to involving smart, technical people who definitely control their own domains but don't think to take advantage of this. Usually they sort of mumble some description that you might be able to use to find whatever they're talking about, or they manage to only get half the words in the title wrong when they're trying to recall it for the host, and then you and every other interested listener has to individually squander time and attention if you want to track it down. It results in a huge waste of collective energy.

The UX is not good enough yet. We would have to 1) show that it is reserved to people going on the link and 2) offer good/enough ways to be notified once the link becomes online and 3) need to know the likelihood of the link to actually work in the future based on prior commitment.

AND I probably forgot a couple of issues.

...and URLs being forever cuts both ways. Want to reorganize your domain's structure? Better set up 301s to forward the old address, forever, and hope (or avoid) any overlap between the old schema and the new.

Transfer of ownership? Well, all those hyperlinks from other sites don't know.

I think the idea is that if you were recording a podcast it wouldn’t be live (I know the parent used the word live but I think they meant “live” during recording or in conversation while the episode is being created), so you are free to make references to soon to be declared URLs.

You just have to make sure you have populated the content at the location you are referencing before you upload or publish your episode for your listeners.

> You just have to make sure you have populated the content at the location you are referencing before you upload

You don't even have to do that, though. It suffices to say, "I'll wire these things up on Monday."

This is not a criticism of URLs per se, but I find it troubling that the mere act of visiting one is something that we have to warn people away from. It's like:

> Don't eat moldy food, else you'll get sick (unless it's bleu cheese, yum)

Ok, that sounds like good advice, and easy do follow.

> Don't look in the direction of modly food, else you'll get sick

That's unreasonable. What kind of madman wrote the rules for this universe?!

When I try to imagine a better way, it usually disallows the referents of a URL-equivalent from changing after they're created--that way trust bestowed once can be reused without prompting the user a second time for the "same" thing.

For that reason, I'm not a fan of placeholder URL's like you're describing. The instability of URLs to me feels like more bug than feature.

I don't know what you're talking about at the end, but you're definitely applying an inconsistent (double) standard at at least one point in your comment. The status quo is one where the reference is either completely unresolvable, or the referent accessible only after some effort that would have been better avoided. Forward-declared identifiers don't exacerbate any of these issues. Meanwhile, the set of things ameliorated by them is non-empty.

Basically I'd rather be using cryptographic hashes of the page instead of URLs so that if I trust the hash then I have an obvious mechanism for determining if I should trust the payload that the hash refers to.

Stability like that would severely limit the places that a malicious payload could hide, and it would enable users to compare notes about what is or is not trustworthy.

If the identifier can exist before its referent, then any such verify-the-payload-given-the-id activity becomes much more complicated because we now have to wonder if we're getting different versions of the page for the same identifier (e.g. like when airlines present different prices based on which browser you've used even though you used the same link in each. I'm trying to dream up a web where that's not possible).

Hashes are not memorable. I can’t give my gramma a hash, but I can give her example.com/recipe123. If she trusts example.com, or me, she should be able to trust the content. If she does neither, a hash will not save her because she needs to have seen the content first to make a decision to trust it.

It's fairly common to create "link" objects which have both a human readable component and a URL component. Not much would be lost if the URL got less readable, we'd just have to be more diligent about associating a human readable string with the link. This could be done automatically if the content happens to provide its own "name" field. Otherwise you'd just have to give your link a name.

Links which are displaying an ad-hoc name can show up in one color. Links which display whatever the content names itself can show up in another color. We can have different fonts for whether people you trust have flagged the content as trustworthy or whether the've flagged it as malicious. Nobody needs to see the hash itself. But none of that works if a link might resolve to different content at different times.

As far as needing to see the content before you know you can trust it... There's no harm in fetching malicious data and taking a peek at it. Just don't act on it.

If we train people to not even look at the threats then they're not going to have a feel for what threats actually look like.

I think that is too much infrastructure for just a link. I think links are fine the way they are today. There are other problems with the web, but it's not links. Also, simetimes taking a look is the same as acting on it. Clicking on an link in outlook is often acting on it, if it's some script or whatever it is that people do on outlook.

If it's content addressed you can gossip content between users, so I'd say it's far less infrastructure because you don't need servers.

But I know what you mean, it's a lot for the user to manage. I'm just looking for something drastic to change because the web as it is makes me feel like a rat in a maze. I'm trying to figure out how to leave notes on the walls for the other rats.

> That's unreasonable. What kind of madman wrote the rules for this universe?!

My chain of thought went: "QR code containing a bobby drop tables! What would a human version of this be? Viral memes, in the original sense, that cause psychic damage? Oh wait, photosensitive epilepsy is a thing."

I think there was such a thing in Snow Crash

https://www.infinityplus.co.uk/stories/blit.htm

That was great, thanks

as a podcaster i feel this, but 1) its hard to look up precise references without interrupting conversation flow (im optimistic llms will help here) and 2) some people would want to tell u to search their name more because that helps The Algorithm

It's like you didn't understand anything that I wrote.

Reply on news.ycombinator.com