I'm gonna paste below what I managed to churn out the last time I tried writing an article about this, and then add some fresh commentary after the pasted section:
---
The bookmark itself is tied to a URL, and anything else related to the bookmark, such as the title, the scraped content (if the service scrapes on your behalf), highlights and annotations are stored as additional metadata linked to that URL.
There are some unfortunate restrictions that come with this data model.
Let's take the example of comments.
Comments by their nature are distributed; the same article can be shared on any number of websites for any number of different users and communities to discuss. Especially in the case of tightly focused communities, the commentary on an article is often just as valuable as the article itself.
When the data model is anchored around the URL, highlights become tightly coupled to that URL, and this tight coupling ignores the distributed reality of comments, leaving them with no real place to exist in the data model.
Ideally, highlights made on an article and comments saved about an article should be easy to connect and view together, because it is the article itself, and not the URL, that is the common denominator (the URL is an imperfect proxy for the article).
What if there is nothing of value in the article itself to highlight, but the discussion of the article contains the real information of value that you want to save? This is often the case when a commenter debunks bogus claims published in an article shared on link aggregation websites while simultaneously explaining their method.
If you think about this for long enough, you may also come to the conclusion that I have, that saving comments is also a form of highlighting, and both comments and highlights can be described in a common data model as "content".
---
Now this is definitely a bit more on the "knowledge management" side of things, but I think that people who curate digital gardens of knowledge are in a unique position to be curators of focused streams of knowledge in a post-firehouse digital world, and that this can take the form of RSS feeds.
Think of "firehoses" like RSS feeds of websites that throw any and everything at you, RSS feeds of subreddits, the HN front page etc. They are all impersonal, without context, and without any reason or explanation of _why_ you should spend your time consuming an item.
Our time is increasingly the most valuable thing that we have, and so if I am recommending that someone read something, I want to provide the reason why I think they should read it up-front (ie. one or more highlights or analytical comments in the body of the RSS item); if that clicks with the subscriber, they can go on to click through to the source and spend more time on it.
But just a title (which is unfortunately likely to just be clickbait these days) with no item body? Or with a huge wall of text? (I'm honestly not sure which is less respectful of an individual's time) - I can't consume from feeds like that anymore. And for that reason, I also refuse to produce feeds like that anymore.