Feedle Reached the Front Page of Hacker News!

I was enjoying my lazy Saturday afternoon when I got an email from one of the alerts I’d set up for our side project, Feedle. I began to half-cross my fingers hoping that the service wasn’t down for some odd reason, but my surprise was twice as big when I saw the reason. Our service load was multiples higher than usual, so I immediately opened Plausible, and it looked like an entire football stadium was currently checking Feedle out, everyone seemingly coming from the same place - Hacker News. Opening Hacker News was the bigger shocker because, believe it or not, there we stood proudly on spot #2 on the Hacker News front page?

While we’d always wanted to engage with the HN community, I’ve been putting aside the thought of promoting the site bluntly because it could often die down in vain rather than become popular. But there it was, posted by I have no idea who, and it was killing it!

While the euphoria eventually settled down, the HN thread kept getting bigger and getting more comments and questions. I tried answering a few right in the thread, but I felt like it would make more sense to just wait to get people’s concerns and address them in one take on my site. So, here I am, trying to address as much as possible while keeping an open door to some of the unknowns.

1. Submit Link Functionality: Several users reported that the “Submit” link for adding blogs or podcasts redirects to an unrelated page or is broken. This is a known issue with the form provider we use (Tally). Fixing it is our highest priority, but what’s making it difficult is that it is difficult to reproduce, as it seems to occur for only a very small portion of our visitors. While we are still digging for the root cause, anyone who wants to add their blog or podcast to Feedle could do this using the form directly.
2. Content Discovery Challenges: Users expressed a desire for more sophisticated discovery features, such as categorization, opposing viewpoints on topics, and consensus finding. With our team’s background in search engine technologies and our recent interest in (carefully and sensefully) adding AI to the list of skills, we can certainly do much of that. At the very core of Feedle, we use a state-of-the-art search engine technology (Apache Lucene), which has been powering private and public Internet services for decades and is capable of way more than what we are currently squeezing out of it. While we can always add more and more interconnected information to every search, our main concern is making it valuable and seamless. We don’t want to complicate the user experience or over-crowd the screen space with information clutter. We want people to enjoy coming back to the site, so please bear with us while we figure out the best ways to improve content discovery.
3. SEO Spam and Quality Control: How will Feedle defend against SEO spam and ensure that only high-quality, genuine content is indexed, especially considering the ease of creating reasonable-looking blogs with AI? I would say, the only sane way of doing that is by keeping up doing the same things we’ve been doing until now - manual curation. We look at every submission. Literally, every submission, which is why the list keeps growing, and sometimes it feels like it is taking forever to get a site in there. But currently, that’s the best way we know of.
4. Multi-language Support: There were concerns about the lack of support for non-English languages. Non-English languages are actually supported out of the box by our search engine technology of choice. The problem is in the reliability of the said tech when it comes to identifying languages of blog posts with extremely minimal metadata. Our identifier currently marks languages correctly in 90% of the cases, which could be a problem. We don’t want to show people content in languages they don’t understand, and we don’t want to falsely filter out blog posts based on their language. That said, multi-language support is coming; we just need to tweak a few settings.
5. Dead Links and Search Noise: Users suggested features for flagging dead links and removing them to reduce search result clutter. This is a good one—I added it to the list.
6. Open Source Development: Have you considered making Feedle open source to benefit from community contributions and improvements? Yes, but it’s too early to have a say on what exactly and when that would be.
7. Monetization Strategy: Without compromising user trust or data privacy, what monetization strategies are you considering for Feedle? We are a small team, and being able to pay our bills with the things we love is the best kind of compliment we can achieve. At the moment, we are bridging the gap between client work and other projects, but we would love to be able to move Feedle forward sustainably. We don’t want to hamper the current experience by putting a paywall on it, so most likely, if a paid plan were to arrive, it would step up on that and provide more functionalities to selected groups of users willing to pay for those. We have also discussed the idea of giving content authors a way to control their Feedle presentation or reach out to visitors in exchange for a small monthly contribution. All those things are open for discussion - if you want to reach out to us with ideas, you are always more than welcome to do so.
8. User Engagement and Audience Building for Bloggers: Some people in the thread highlighted bloggers’ struggles with building an audience. Obviously, our contribution here is that others get to learn about those content creators unintentionally by subscribing to a topic of interest rather than multiple people’s feeds. The more popular our site becomes, the bigger the discovery effect will get. At the moment, the best way someone can contribute to our project is to simply write about us, link back to us, talk about us on a podcast, or mention us on social media. The positive effect of that is that they will return back to their own communities, blogs, communities, etc. Of course, besides the pure network effect, we are actively looking at ways of identifying when some content deserves to be shown more often - be it based on the pure value of the content or some popularity metric. However, we want those to be subtle nudges.
9. Indexing and Search Accuracy: How do you ensure the accuracy and relevance of search results, especially for nuanced or complex queries? Our search engine tech supports an entire array of special keyboards, Boolean operators, and other syntax mechanisms that can help filter content out to the exact set of wanted results and nothing more. Basically, any advanced search feature Google can do, we can do, too. We just have to find ways to educate our visitors about those.
10. Where are the podcasts in “blogs and podcasts?: With a few exceptions, we still primarily focus on showing blog post results on Feedle. This is not to say that we are not indexing podcasts. We are, but with time, we have decided to do so in a separate index on the backend. Feedle users will still be able to get both types of content in their search results. The semantic distinction would better target podcasts - i.e., provide a built-in audio player, fine-tune podcast RSS feeds for direct usage in podcast apps, etc.
11. BONUS - Add an “I feel lucky” button for serendipitous content discovery. I love this one, and am taking it straight to the rest of the team. I think that having a constantly refreshing list of random search results would add a playful discovery element to the overall user experience of Feedle.

There you have it - I hope that I was able to address people’s feedback and concerns about our Feedle. In a future blog post, I will focus a little bit more on the technical side of the service and how we were able to handle the HN load with zero downtime and barely registering CPU and memory consumption (while being a Java app, imagine that). Stay tuned, and in the meantime, feel free to send us more feedback in the comments below or on our email.