Hacker News Reader

> Discovery of good sources of information such as blogs is hard. And I think the biggest problem is the lack of trust. Everyone wants to grab your attention [1]. So how do you know that others won't waste your attention?

> To solve this problem I am building https://linklonk.com that cultivates trust as you rate content. When you upvote a link you connect stronger to the feed that posted it (which could be a blog's feed) and to other users that upvoted this link before you. When you downvote - your connections to those who upvoted become weaker.

> The strength of your connections to other feeds and users represents how useful their content recommendations have been to you in the past and they could be used as a measure of how likely their future recommendations will be worth your time (ie, trust that they won't waste your time).

> The content is ranked according to the connection weights - so you get information from the sources that have shown to be content curators for you.

> I did a Show HN recently for this project that has more details: https://news.ycombinator.com/item?id=28405643

> [1] - like my comment here tries to draw your attention to my hobby project.

I find this so interesting. OP complains about people constantly trying to sell him something which is why he was turned off of blogs. And what happens? Someone tries to sell him something for his problem of too much selling.

Ah. What a world. Solution to too much selling is more selling.

LinkLonk is free, and it solves the problem. I've only used it for about 15 minutes, because in that time it fed me enough interesting stuff that I overwrote (and then closed) the tab.

This isn't somebody trying to sell something. It's somebody trying to help.

What does overwrote mean in this context? thanks

HN is a place full of people working on solutions to problems shared by a lot of people on HN, so it's common and normal to to offer that thing up to someone with the same problem.

Alternative framing: Solution to someone's problem is to propose a tool one has built that helps fix it.

Pretty sure there's an xkcd for this.

I like the idea. I have a sort of feature request premised on the assumption you end up having weights attached to the index to determine the priority of output. Could you make these weights exportable?

In that way, you could have curated content. Like if I find someone that has really similar interests to me I could import their weights and see the web via their prioritisation. Similarly, (countering a problem I always have with google search bubbles) I could explain to my friend how to navigate to a site I found via search if they import my weights.

Edit: being able to manually modify my own weights would probably be helpful as well to decay sell outs.

I had an idea similar to this as well - to help people kickstart recommendations for their friends. I'm thinking of creating a personal url for each user (e.g., linklonk.com/u/lonk). If someone upvotes that url - they establish connections to the sources that user is connected to.

When you want to introduce someone to LinkLonk you could share with them your personal url.

As for decaying sellouts, wouldn't downvoting the content they upvoted do what you want?

By the way, every time someone you are connected to upvotes something, your connection to them becomes slightly weaker. So if you simply ignore recommendations from sources with a lower signal-to-noise ratio (or high volume) - their recommendations will eventually fade away from the "For you" page.

Can I dump my OPML into this, and you recommend me feeds based on what I'm already subscribed to?

This is a really nice idea. If I may make a suggestion, scrolling down takes a lot of scrolling because each link takes up so much vertical screen space on account of really big whitespace gutters, size of the thumbnail, and the general layout. I wind up being able to see only 3 on a screen compared 16 on HN! I would be much more likely to adopt this regularly if you changed the layout to pack more previews in to a vertical screen of space. HTH

Would it work if we had an option for a "compact" mode that shows items without no thumbnail and description?

Or would you change something else?

I don't know if I can get rid of the bottom part with the buttons since this is where the collection selector is - where you pick what collection the item you like should go to.

yeah a more compact mode would be good!

> The strength of your connections to other feeds and users represents how useful their content recommendations have been to you in the past and they could be used as a measure of how likely their future recommendations will be worth your time (ie, trust that they won't waste your time).

Your intention is noble, but this is still based on network effects and a positive feedback loop. I doubt you can beat the social medias in their own game.

True, the network effects of the traditional social media are strong. LinkLonk needs a critical mass of users for network effects to kick in.

To make it worthwhile for the users before the project reaches that point, I am trying to make LinkLonk a better tool that helps you keep track of the content you liked (kind of a bookmarking service) and helps you follow RSS feeds in (kind of a feed reader).

If you have other ideas of how to make it more useful before we have enough users - let me know.

> If you have other ideas of how to make it more useful before we have enough users - let me know.

I have an idea; maybe you can enable people to publish a list of links that are curated by themselves, in the form of a RSS feed. This way your user can have a sense of ownership.

Thanks for the idea! LinkLonk already has a concept of a "list of links" - collections. When you upvote a link you put it into one (or more) of your collections.

What you are suggesting is similar to another idea I had (https://linklonk.com/item/9146000221282140160): "An option to generate a publicly visible url for any of your collections. This could help you share your collection of liked items with other people. And it could help LinkLonk get new users."

Adding an RSS feed for these publicly accessible collections would be a natural extension.

Thanks for considering it. my email address is derek _at_ 3qin.us. Shoot me an email if you want; I may have more ideas.

The initial version of sharing collections by link is done: https://linklonk.com/collection/programming-yUlhkwDXFb

RSS would be the next step.

RSS for everything would be epic!!

You are still going get splash damage from people who trust people who fall for plausibly non-clickbait clickbait, as is the top item for me right now [1]. As this system gets larger and more overlapping you are going to have to work more diligently to undo bad follows to get tinier amounts of content referrals.

The discovery problem should be solved on page and spread that way, to remain decentralised and free from the all the centralised incentives that lead to clickbait. You almost have to punish hosts that appear anywhere centralized lest they get a taste for clickblood.

[1] https://linklonk.com/item/2226616383778586624

If you see content that is not worth your time recommended to you then you can downvote it to stop trusting those who found it useful. It is all in your control as a user.

I didn't quite get what are the centralized incentives that lead to clickbait. Is it a comment about LinkLonk or centralized systems in general?

Centralization and decentralization have trade-offs. Right now the downvotes on LinkLonk are private. The upvotes are semi-public, but not available in raw form. If you wanted to implement the LinkLonk's algorithm in a decentralized manner then you would need all hosts to make the upvote and downvote data available to all other participants. Maybe homomorphic encryption can solve this problem.

My goal is to build a proof of concept in my spare time and the centralized approach seems to be the right option in my situation.

I don't think trust can be reliably built with elements of gamification like this. The deal with gamification is that people learn to game it out, which erodes the kinds of sincere or honest interactions you're trying to cultivate.

In the olden times blogs earned trust by cultivating a reputation. The reputation was earned by having an audience that trusted them and recommended them. Cross-linking content to other blogs, guest blogging, being included on a blogger's 'blogroll,' etc. were all ways they expanded their audience.

It was slower and had much less reach, but it also focused more on "building an audience" rather than "driving traffic." We, fundamentally, don't trust content, so mechanisms that operate on validating atomized bits of content are going to fall flat. We trust people and institutions. If you want to build trust it has to work on the agents producing the content rather than the content itself. Segmenting content up into atomized bits is what creates the erosion of trust in the first place. It's something timeline driven social media feeds do specifically because it makes it difficult to parse genuine buzz from advertising, which makes the ads more effective. But that's the opposite of trustworthiness.

This is sort of a perfectionist perspective. Search engines use the same “gamification” and suffer the same problems you’re “predicting” but that doesn’t mean you don’t use search. It does mean it’s an arms race between the engine and the abusers. Weighting the agents instead of the content is no different than a popularity contest and is essentially an “appeal to authority” (or a lot like cancel culture). Just because someone has weird opinions about X doesn’t mean they can’t be brilliant about Y. If you ranked the content then their X content can sink and their Y content can rise. The problem with Twitter, etc. is that their incentives are not as aligned with their users’ goals as we would like. Probably the best part about blogging was that it wasn’t centralized and so wasn’t subject to one person’s definition of what those trade-offs should be. But, of course, now we’re trying to discuss fixing one of its weaknesses without losing too many of its strengths.

> If you want to build trust it has to work on the agents producing the content...

LinkLonk's algorithm works that way. It builds trust in sources of information (including users who rate content). And it does not and will not try to understand the individual pieces of content.

Unlike the social media feeds that are powered by black box neural networks, LinkLonk's algorithm is transparent. You know how your interactions with it will be interpreted. I hope that this transparency will help build trust in the system and in the sources of information you are connecting to.

Yes, bad actors will try to game any system to gain the attention that they don't deserve. I'm not claiming that LinkLonk is game-proof, but I think it has better feedback loops and incentives than other systems such as popularity based ranking (please don't take it as a challenge).

I used to hate canned tuna in the past but I love it now!! Your mechanism doesn't account for this!

Your connections to feeds and other users on LinkLonk evolve over time as you rate content. If you start upvoting tuna-related content then LinkLonk will connect you to other tuna lovers and your connections will gradually "forget" your past hate for it.

Canned tuna is just about the only food I can't bring myself to like. What changed for you?

Neat project, thanks for sharing.

We need a standard API to subscribe to blogs on other sites, that will appear on our blog feed as news feed -- like how facebook or tweeter feeds work. RSS is too personal, too much outside the flow.

RSS is a standard API to subscribe to blogs. Maybe what you're suggesting is that more tools are needed to make RSS more accessible to lay people?

RSS is too personal, too much outside the flow.

I have no idea about "too personal" mean, but the only reason you would consider RSS "outside of the flow" is due to the concerted effort by Google, Twitter, Facebook and Apple to reduce support. Even Mozilla(!) has been involved in removing support.

RSS is a perfectly good, tested and usable mechanism. Coming up with yet another syndication mechanism would be a huge waste of time and effort, most likely resulting in insignificance.

I don't understand how RSS is too personal. An RSS feed of a blog is usually the same for all readers of the blog.

I didn't get that comment either. I've doubled-down on RSS for this purpose in my side-project Haven[1]. Write your own private blog, share it with people who can then subscribe with personal http-basic-auth RSS links (or view it on the web), and I've recently built in (still a WIP) a feed reader so you can create your own facebook-style news feed of anything on the web or things your friends write privately on their own Haven.

[1]: https://havenweb.org

Maybe they mean that since there are many different feed readers a person could use, a blog can't have a "Click here to subscribe via RSS" link? Most feed readers will have a bookmarklet for 1-click subscribing, but the blog owner doesn't have the ability to make a prominent "call to action"-style button.

Sure they can, well they can have a link that brings up the feed anyway. What the browser does with it is a different story, and that depends on whether RSS preview extensions are installed. Copying and pasting that URL wherever it needs to go isn't all that complicated.

Like OPML? http://opml.org/

OPML could be part of a solution for authoring blogs that then can publish using RSS. It is an interchange format for outline with attributed text.

Back in the old Web 2.0 days we used OPML files to publish and exchange blog rolls, Dave Winer even had a site you could log into and share, in a sort of social-networking way, your blog/RSS subscriptions via OPML.

Dave has been working on Drummer [1], a browser-hosted outliner. I haven't used it, but I am glad he is still at it.

[1] http://drummer.scripting.com/