lonk11

15 ms·

91.
by lonk11 5y ago
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 p
92.
by lonk11 5y ago
Yes, an echo chamber is a concern. But unlike other systems, you are in total control of who you get your content from. You are responsible for what you see. Which I think is the best place to have this responsibility. This does not work fo
93.
by lonk11 5y ago
How about this way to automatically address the Eternal September problem: When you upvote an item - you connect stronger to people who upvoted that same item. When you downvote an item - you connection to those who upvoted it becomes weake
94.
by lonk11 5y ago
They are similar. Both are a type of collaborative filtering ( http://en.wikipedia.org/wiki/Collaborative_filtering ). Though I don't know any details about Last.fm's algorithm. Most likely it is based on neura
95.
by lonk11 5y ago
When you upvote a link, LinkLonk tries to extract the feed url from the metadata of that page. So most of the time you only need to submit a link to the content you found useful and LinkLonk will start tracking the source feed and will brin
96.
by lonk11 5y ago
I think the way LinkLonk works will support the scenarios you are talking about. If you read every issue of the Money Stuff newsletter and upvote 5 out of 10 that are most informative and then if I upvote 2 of those 5 that you upvoted then
97.
by lonk11 5y ago
The "Show HN" post is up: https://news.ycombinator.com/item?id=28405643
98.
by lonk11 5y ago
That's exactly how my project https://linklonk.com works. When you upvote an item, you connect to other people who upvoted that item. When you downvote - you disconnect from those who upvoted it. So those who upvote useful
99.
by lonk11 5y ago
I built https://linklonk.com - an information discovery system. It’s an experiment to try address the issues with the three ways we typically discover information: 1. We follow individual sources and it gives us a high signal-to
100.
Show HN: LinkLonk – information discovery with aligned incentives(linklonk.com)
24 points by lonk11 5y ago|6 comments
101.
by lonk11 5y ago
Yes, I managed to delete the "A" DNS record last night when I was adding records for mail hosting linklonk.com. Sorry! It is up now.
102.
by lonk11 5y ago
Do you have any specific suggestions on how to UI? Small tweaks would be most appreciated as I could implement them before the upcoming Show HN post. To answer your questions: 1. The list of all feeds used by LinkLonk is not publicly access
103.
by lonk11 5y ago
I share your concerns about user privacy. The information LinkLonk collects is what you explicitly provide (ratings, etc) and the regular server request logs (which include your ip address, user agent). I clarified this in https:/&#x2
104.
by lonk11 5y ago
Thank you! Indeed, the upvote button on those item-based recommendations didn't work. It is fixed now.
105.
by lonk11 5y ago
Thanks! It looks like 13 people signed up, which is really encouraging. I wrote about the performance tuning I did in preparation to the Shown HN post: https://linklonk.com/item/277645707356438528 If you have any feedb
106.
by lonk11 5y ago
I am building a project ( https://linklonk.com ) that does information discovery in a way that respects your attention. In short, when you upvote content you connect to other people upvoted that content and to feeds that posted t
107.
by lonk11 5y ago
Here is an example of how your likes affect what other see (or in this case, don't see): https://boffosocko.com/2017/07/11/the-facebook-algorithm-mom... If you know that the content you like will get mor
108.
by lonk11 5y ago
Most platforms use a black box ML model (a deep neural net) to optimize for business objectives such as time spent which correlates more with the number of ads viewed than if they optimized for "Likes". These systems have far fewe
109.
by lonk11 5y ago
One problem is the design of the “Like” button. The target audience - the recipient - of the action of pressing this button is the other people. They will see a higher number next to the thumb up. If this is the visible result - this is wha
110.
by lonk11 5y ago
The overall system behaviour is hard to predict. So yes we need to try it out and see. But as a user I can say that I do behave differently when it comes to upvoting content on LinkLonk. It is much less impulsive because my ratings have con
111.
by lonk11 5y ago
Correct. This also provides some protection from gaming of the system. You can't simply upvote a bunch of popular items to get other people who already upvoted it become connected to you. You have to be good at prediction what people w
112.
by lonk11 5y ago
You can think of every user as a neuron where upvotes on the same items strengthen their connections. Though there is one important bit of asymmetry: you connect to those who upvoted that item *before* you. That way people who recognize use
113.
by lonk11 5y ago
You are right, LinkLonk is a filter bubble. The difference from other systems (e.g., algorithmic feeds powered by machine learning that optimize for "engagement") that exhibit the filter bubble dynamics is that LinkLonk puts all
114.
by lonk11 5y ago
The purpose of the downvote is for you to say what content wasted your time. Those who brought that content to you deserve to lose your attention so they do not waste your time in the future, do they not? That's why LinkLonk decreases
115.
by lonk11 5y ago
> In this paper, we test the hypothesis that adaptive influence networks may be central to collective human intelligence with two preconditions: feedback and network plasticity In the paper they had the participants of the experiment to
116.
by lonk11 5y ago
The newsletter/sample page has a broken link: "I hope you find the following stories interesting. If you don’t, you can [unsubscribe]({{ unsubscribe_url }}) and I won’t pollute your inbox again." Is supposed to be substituted
117.
by lonk11 5y ago
You can subscribe to articles posted by a user as an RSS feed: https://hnrss.org/submitted?id=bengtan The problem though is that it is hard to find users you want to subscribe to. You need to inspect their past submissions
118.
by lonk11 5y ago
My hobby project https://linklonk.com (use invitation code 'hn') is a content recommendation system that is built around these principles: - Control: what you see is based on your explicit upvotes & downvotes, not