On Linking Bookmarks

Published on to joshleeb's blog

In the last post, On Organizing Bookmarks, I wrote about an idea to improve tags for organizing bookmarks on social bookmarking sites such as Pinboard.

To continue this theme, I want to explore bookmark linking as a way to extend bookmarking tools to work better with the distributed nature of the web, specifically with how posts are shared and discussions happen in a variety of places, such as aggregators and forums.

Before getting into linking, I think it will be helpful to share some context on why I bookmark, and how I discover webpages to bookmark.

On the Compulsion to Bookmark

Personally, I get the most joy and fulfillment from the web with reading and writing blogposts. Or really, any long-form (i.e: longer than a 280 characters) text that is usually technical in nature. My favorite posts to read are where the writer is sharing something they’ve learned or discovered, or have some opinion they’ve formed from careful consideration and thought.

For this reason I hang around Lobsters a lot, looking through the top posts at least once a day while sipping on my morning coffee. And I use an RSS reader (Miniflux in my case). These are the two ways I mostly discover interesting sites and posts.

As much as I enjoy this long-form, information dense writing, when I find a post that looks interesting I can’t always engage with it fully. Perhaps there’s a time restriction, or I’m not in the right headspace to understand the details being discussed. And although I may not be able to look at the post in that moment, that doesn’t mean I want to lose it. I want to store these pages and come back to them when I’m ready.

When I create bookmarks that don’t come through Lobsters or Miniflux it’s almost always when I’m doing a deep dive into a particular topic. For example, the last one I posted about was on Content-Defined Chunking.

My deep dives begin with a discover phase, which involves collecting links and papers. And the last thing I want is to end up with a million tabs I have to keep open. These explorations sometimes take weeks, or months, and need to be done systematically.

And so bookmarking is my way of managing a reading list with the bonus that I can refer back to these bookmarks later and see what others are reading on sites like Pinboard. It helps me enjoy my time learning and exploring on the web, and prevents me from getting overwhelmed with the volume and depth of knowledge available.

Figure 1. A rough breakdown of how I discover webpages to bookmark.

Bookmark Linking

As I was thinking about how and why I bookmark, I also started imagining how I could improve my process. One of the ideas was to link bookmarks to other parts of the web. More specifically, linking bookmarks to

Linking with Aggregators

Lobsters is the aggregator I go to most. It’s a wonderful computing-focused community, full of smart people, interesting posts, and generally very insightful and encouraging discussions.

When I come across a submission I want to bookmark I also want to save the Lobsters page to stay up to date with the comments.

This is the idea with linking bookmarks to aggregator submissions. Along with the bookmark, there will be a link to where the webpage is posted on Lobsters, Hacker News, and any other aggregator or forum. Comments, upvotes and other activity can be pulled in, all in one place. There’s not much more to it.

Linking with Web Feeds

Since I discover ~25% of my bookmarks through a feed reader, naturally I was wondering if it would be useful to link bookmarks back to feed posts as well.

One idea is that this would be a way to create a stream of articles. Each bookmark could be used as an implicit subscription to that site’s web feeds, along with any explicit feed subscriptions, to help with discovering more posts that might be of interest.

There’s also a sense of completeness I like about linking back to web feeds. As the web feed comes from the site where the article is posted, it is the canonical source of data about these pages.

Figure 2. Graph visualization of linking web feed posts (P), aggregator submissions (A), and bookmarks (B).

Practically though, not every site has a web feed (and not all web feeds are valid). Which brings me to the final kind of bookmark linking.

Linking with Other Bookmarks

On a Pinboard-like social bookmarking site each user isn’t likely to have more than one bookmark of the same webpage. But across multiple users there are bound to be many bookmarks of the same page.

From the differing tags used across these bookmarks, we can provide tag suggestions for when another user adds a new bookmark of the same page.

And, if comments are also supported on the bookmarks themselves, then we can connect comments from users and aggregators when the site doesn’t have a web feed.

Wrapping Up

The web is a chaotic and inconsistent place with broken or redirecting links, and invalid or missing web feeds. These inevitabilities hurt the chance of linking bookmarks to aggregator submissions and web feed posts. And so linking will always be best-effort.

But still I think it’s better than no linking at all.

To test this, I collected a few entries from Lobsters to act as simulated bookmarks. For each one, I tried to discover the web feeds on the site and link the “bookmarks” to one of the posts in those feeds.

I’ll leave a detailed post of this method for another day, though the results were somewhat promising. Around 40-60% of bookmarks could be linked back to a web feed, which was honestly better than expected. Though that is probably because Lobsters is an unrepresentative sample of the whole web.

Anyway, let me know your thoughts on linking, where else but here on Lobsters.