Hacker News new | past | comments | ask | show | jobs | submit login

I believe that the web could be so much more if competition between platforms was somehow decoupled from competition between UIs. It would be incredible if every social network had a marketplace of competing UIs, allowing for rapid front-end innovation.

Of course, many services provide APIs, so in theory this decoupling is already possible, but the problem is that many platforms rely on advertising & promoted content for their revenue, so they need to maintain control over the front-end and therefore won't allow people to build alternative UIs.

I don't know how to solve the revenue problem. I just wish we could find a way to have UI competition.




Before behemoths funded by free-but-spying software/services took over, protocols and standards were a thing and had some power, at least. They've (largely Google, but not just them) been slowly killing them. Can't risk making it easy for eyeballs to leave your platform (ugh) since that's what you're selling, and you can't just stick ads in (say) their RSS feed since that doesn't help you track what people click, let alone spy on them across other sites.

Basically protocols are what you want, but they don't make it easy enough to be shitty, so they're dying until we outlaw (yes, with actual laws) being shitty. Which may never happen.


I really want to try this.

A new Reddit

- but with the Wikipedia funding model

- with Snapchat's ephemerality

- an amazing API and all open source code

- all submissions cc-by-sa licensed

Since profit isn't a motive, it would encourage the proliferation of new frontends and clients. As well as really cool mashups (haven't heard that term in awhile - probably due to increasingly closed ecosystems.)

Without having to save user generated content you can save tremendously on servers. You don't have to scale other than enough to handle the fire hose. Everything could reside in memory. Just Redis and Rust.

The content gets scrubbed after a week. No sharded storage is necessary. But due to the license, others can persist content they like.

There's next to zero value in looking up old posts. I don't know why this hasn't been done.


> There's next to zero value in looking up old posts. I don't know why this hasn't been done.

I genuinely cannot tell if this is satire.

A lot of subreddits are bastions of information for sometimes very niche topics. I frequently run into reddit posts that are 6+ months old but still relevant to what I'm seeking knowledge about. Outside of the programming sphere, posts from years ago generally still hold value.

Saving posts would imply that I knew what Future Me wants to look up, or praying that someone was interested and cached it. Writeups/insightful discussion would be completely wasted, and I would have to find -another- website for potential caches of things relevant to what I'm looking for.

It would be like IRC where people delve into the same topics and arguments ad infinitum.

If I wanted an ephemeral social network (like snapchat) I would use an ephemeral social network, like snapchat.


> A lot of subreddits are bastions of information for sometimes very niche topics.

While I don't doubt that's the case, it's incredibly difficult to get at old content due to Reddit's horrible search experience. I never look up anything old there.


Reddit's search has always been terrible.

site:reddit.com or site:reddit.com/r/nichesubreddit <query> is how I actually find things. Lately I actually have begun to cache important things locally, but I wouldn't assume anyone else is doing that for the particular topics or niche subsection of a topic that I'm looking up.


>The content gets scrubbed after a week. No sharded storage is necessary. But due to the license, others can persist content they like.

This is what a * chan does. N pages of content max; anything more, and the oldest thread will be "dropped off" the site.

Combined with anonymous-by-default posting, it works fairly well.

>There's next to zero value in looking up old posts. I don't know why this hasn't been done.

This is not really true; there's tons of good information that gets lost. On most * chans, this was recovered by screenshotting the thread and sharing the image around; today there are auto-archiving sites for the larger * chans, and people just link to that. But the culture still prefers sharing (edited, MS Paint) screenshots around.


> Without having to save user generated content you can save tremendously on servers. You don't have to scale other than enough to handle the fire hose.

Or just get rid of the false binary between centralized control/data (silo model) and decentralized resource-consuming p2p apps (blockchain model).

Federation models provide a sane alternative and have proven scalable/maintainable in the long run (see email/xmpp).

For federated web protocols, see ActivityPub, Linked Data or WebSub. Outside the web, there's the old and wise XMPP.


There's many ways to achieve this. Usually any standard protocol will make it possible.

On the web specifically, micropub and activitypub were designed with interoperability in mind (for the microformats and activitystreams formats respectively).

We need to use more of these (or other) standard protocols in every backend project we make: our blogs, agendas..


Isn't that what Fediverse is all about or at least coming close to? The communication runs on ActivityPub (https://activitypub.rocks/) and anyone can setup their servers and clients as they wish.


maybe ActivityPub will reach some kind of critical mass.




Consider applying for YC's W25 batch! Applications are open till Nov 12.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: