TL;DR: This piece delves into my personal journey through an information access crisis and my reflections on developing Sublink, a platform dedicated to link-sharing.
At my previous job, we had a company forum that allowed for the creation of discussion groups. I initiated a group called “Knew” with this introduction:
“Knew” is a platform designed for ‘brain-sharing’: every week, you can share the latest events and developments in your areas of interest. This facilitates knowledge sharing among all of us. You don’t need deep expertise in the topics you share; simply making others aware of their existence is significantly valuable.
I even selected an appropriate icon for it.
Weekly, I’d share intriguing finds like DALLE-2, Azuki, and vaccine efficacy data. My group quickly became one of the forum’s top three most popular, attracting many followers and likes.
However, the group quickly became a one-way street with only me sharing and others consuming, as no one else contributed. After several editions, I gave up, realizing the project didn’t align with my original intent of fostering mutual brain-sharing.
This experience led me to ponder how to facilitate easier, low-pressure information sharing among people. In September, I came up with the idea of Sublink:
a platform where you could curate a collection of anything you like and share it with friends. Any new addition to your collection would be instantly visible to them.
I shared this concept with two web developer friends in late August — They found the idea intriguing, but their busy schedules meant little progress was made until November.
Consequently, I took matters into my own hands, embarking on a self-taught journey in modern full-stack development using Next.js, Typescript, Prisma, and Supabase. With YouTube’s assistance, I mastered the basics of React State and Next.js App Router by the end of November.
Yesterday, after two weeks of development, I launched the site. It serves as a versatile platform for organizing, sharing, and subscribing to links, much like a shareable, collaborative bookmark. Ideal for curating collections of works, or simply sharing lists of movies, music, or articles, it also supports RSS for notifying subscribers about new additions.
But the question remains, what’s its practical use?
The Small Website Discoverability Crisis
I’m a fan of Radio Garden, where I can even listen to live radio from Pyongyang; it’s really a strange experience. I also love playing Neal’s password games. The internet hosts a myriad of shiny, intriguing, and one-of-a-kind sites, yet there’s no dedicated hub for preserving and displaying these gems. Whenever I come across a great blog, I tend to share it with friends and discuss it, but then it often slips my mind. Recently, a trending article on Hacker News, The Small Website Discoverability Crisis, resonated with me. It discusses how small websites struggle to get noticed by search engines.
The search engines driven by the Page Rank algorithm have led to the rise of content and link farms. Content farms produce vast amounts of irrelevant material using AI or scripts to improve their ranking in Google’s keyword searches. Link farms are about creating or purchasing backlinks across multiple websites to enhance a site’s Page Rank, resulting in a distorted online presence.
As a result, truly valuable websites become overshadowed in search engine results, rendering them virtually invisible.
Information Cocoon Rooms
We’re enveloped by recommendation algorithms. Yes, we can turn to GPT for fresh insights, but “you can’t ask about what you don’t know exists.” Therefore, even with the most knowledgeable teacher at our disposal, if we fail to pose questions that challenge our existing knowledge, gaining perspective-altering answers is challenging. Take, for instance, the Aronson Effect—are you familiar with it?
That’s why, in the age of Large Language Models (LLMs), there’s a heightened need for human-curated content feeds. These feeds bring us information in unexpected, raw, and sometimes unsettling ways, bypassing neural network filtering. While most people might not seek this, a select few, like me, find it indispensable.
What’s Possible with LLM?
Imagine this website fifteen years ago - it would have been just the same. But now, at the close of 2023, why not integrate the trending LLM into websites? I’ve envisioned two potential cases:
1. Automated Summaries
Originally, I planned to use LLM to extract webpage descriptions as brief summaries for each link. But I eventually decided against it: LLM charges based on usage, making it hard to control monthly expenses. For a website that clearly can’t turn a profit, the cost incurred from a surge in user-added links could be financially devastating.
The key to sustainability is keeping unpredictable expenses low.
2. Language Delivery Network (LDN)
I once came up with an idea, let’s call it a Language CDN (LDN): When I write a blog, my goal is for people worldwide, regardless of their country or language, to read it in their local language. Using browser-based translations or plugins is a common workaround, but this often results in repeated translations and poor control over the quality.
That’s where LDN comes in: Authors write a blog or create a website and use LDN to manage its multilingual versions. If a new language user visits and LDN doesn’t have a cached translation, it translates once and then saves this version for future visitors in the same language. This allows authors to maintain and fine-tune the translations, ensuring both quality and consistency.
Although this idea wasn’t embraced when I suggested it for the bearblog open-source project, I’m considering implementing it on Sublink, assuming there’s a demand for creating original content there.
I have modest expectations regarding this website’s profitability, but my goal is to sustain it long enough to significantly expand its collection of Links and Collections. Thus, finding a viable revenue source is crucial.
I’m faced with a dilemma: What truly defines Sublink? Is it about “subscribing to premier Collections” or “serving as a tool for organizing personal link compilations”? This discrepancy leads to two distinct models: the former would showcase other people’s Collections on the homepage, while the latter would feature a search box. Initially, the former is challenged by the scarcity of high-quality content.
This difference leads to distinct approaches in monetization. While it’s not impossible to charge for access to Collections, practically it’s almost unfeasible. If we adopt the latter model, similar to Notion’s approach, there’s a greater likelihood of successful monetization. However, this would likely divert us from my core mission of making exceptional links more visible to a wider audience.
This predicament has led to the current lack of a payment model. I am hopeful that future user engagement and feedback will provide greater insight and direction.
This sums up my reflections on Sublink. As both a user and the creator of Sublink, I’ve curated a list of articles I’ve enjoyed and found insightful. You can follow this list using any RSS reader; This way, whenever I bookmark new content, it’ll pop up for you.
Feel free to compile your own collections of cherished videos, movies, books, websites, or anything with a link, and share them with friends or keep them as a personal archive.
Perhaps, this is my small yet meaningful act of resistance in a world on the brink of being overwhelmed by content generated by LLM.