I saved that Hacker News story eight months ago and now it is gone forever

You remember the post. It was about distributed locks. Or maybe it was the one about Postgres advisory locks that solved the same problem in three lines. Or the comment thread under it where someone explained why Redis SETNX is a trap.

You favorited it. You are sure you favorited it. You are on news.ycombinator.com/favorites?id=yourname right now, scrolling. You have been scrolling for several minutes. The post is from sometime between March and September. Probably. You are not certain anymore.

This is the Hacker News save problem in a single moment. The list works perfectly the day you favorite something. It fails completely the day you try to recall it.

Why do Hacker News favorites fail you?

The favorites feature on Hacker News is intentionally minimal. It is a chronological list of every story or comment you have ever clicked the favorite link on, displayed in the same paginated format as the front page. There is no search box. There is no tag system. There is no full-text indexing of the story bodies or the comment threads underneath. There is no folder structure. There is no export button that does anything useful, only a URL you can scrape if you know how.

You remember roughly when you saved each one, and the chronological order matches the way your brain stores them at first. But that only holds for the recent saves. After a stretch of heavy reading, the early entries blur into one another. You scroll past titles you no longer recognize, knowing you saved each for a reason that has since vanished.

Many heavy HN readers describe the same pattern. The save action is a single click. The recall action is unbounded scrolling. The asymmetry guarantees the list grows faster than you can ever revisit it.

What does the HN favorites list actually store?

Favorites stores three things: the story title as it appeared the day you saved it, the link to the story (often pointing to a URL that is now dead, paywalled, or rewritten), and the link back to the comment thread. That is everything. It does not store the story body. It does not store the comments. It does not store the reason you saved it. It does not store the tags you would have added if there were tags.

This matters because the value of an HN story is rarely the headline. The value is in the top comment from the person who actually built the system being discussed, or the buried thread three levels deep where someone posts a benchmark that contradicts the article. None of that is captured by the favorites link. When the external URL eventually dies, which happens to many links over time, the title is all you have.

So your favorites list is, in practice, a list of vaguely remembered titles pointing at variably broken URLs and intact comment threads you have to scroll through manually to find the part that mattered.

Why does search not work on Hacker News saves?

The official site does not index your favorites for search. The third-party search tool, hn.algolia.com, indexes all of Hacker News, but it does not have a notion of "your" favorites. You can search for a title or a phrase you remember, but you get every match in the entire HN history, not the subset you saved.

In practice this means three things. First, if you remember an exact phrase from the title or the top comment, you can sometimes find a story through Algolia search and confirm whether it is the one you saved. Second, if you only remember the topic vaguely, Algolia returns many matches and you are no better off than scrolling your favorites list. Third, the comment search on Algolia is keyword-only, so if you remember the idea but not the words, you have no way in.

This is the gap that makes the HN save habit so frustrating for the people most likely to have it. The audience that favorites heavily on HN is the same audience that reads for ideas, not for headlines. The recall problem is a vocabulary mismatch problem, and keyword search cannot solve it.

What kinds of stories do HN readers actually save?

In our own conversations with heavy HN readers, the same shapes keep coming up. Engineering postmortems and architecture deep-dives. Tooling announcements with the comment thread where the trade-offs get hashed out. Career and hiring discussions, especially the ones with concrete salary or interview data. And the occasional essay that crosses over from a personal blog into wider relevance.

"The one about why their feature flag system blew up at scale." "The thread where the ex-Stripe engineer explained why they rebuilt their fraud system." "The post about the salary leveling at a company I am now interviewing at."

The favorites list, indexed by title and date, cannot serve any of these recall patterns. You are searching for a memory and getting back a chronology.

Why does the list keep growing anyway?

Because saving costs one click and recall is invisible. You favorite something while you are reading it, in the middle of a short commute or a coffee break, and the action feels useful. You will come back to it. Of course you will.

You do not come back to it. The favorites page is not in your browser tab rotation. It is not in your morning workflow. It only opens when you specifically remember that you saved something and want to find it, and by that point the list is too long to scan.

The loop tends to look the same in HN threads about saved stories: save aggressively for a year, lose track of what is in the list, give up on the list, start saving to a different tool, lose track of that one too. Some users move their saves into Instapaper, others into Notion databases or Obsidian vaults. The save habit is real. The problem is not motivation. It is that recall on a chronological list does not scale to a year of reading.

How does dEssence help?

dEssence is memory you don't have to maintain. You save it, forget it, ask for it later. There are no folders, no tags, no organizing. The three save surfaces are co-equal: Chrome extension, Telegram bot, or the web app at dessence.ai.

For Hacker News specifically: when a story is worth saving, click the dEssence icon in Chrome and add a single sentence about what made it worth saving. "The Postgres advisory locks thread that explained why SETNX is a trap." That sentence becomes the recall hook, stored alongside the link back to the HN comment thread. Months later, when the same problem comes up, you open the web app at dessence.ai and ask in your own words, the way you'd describe it to a friend: "that thing about distributed locks and SETNX." Your sentence and the thread link come back together. The comment itself still lives on Hacker News, and HN comment URLs almost never break, so the click-through is intact.

Honest about dEssence

dEssence is in beta. There is no native iOS or Android app yet, only the Chrome extension, Telegram bot, or the web app at dessence.ai. The paid tier is not finalized. There is also no team or shared list support yet, so it will not replace a group bookmark folder. It will absorb what you save from today forward, which is the part that compounds. To be exact about what gets stored: dEssence preserves the sentence you wrote about why a story mattered, plus the link back to the HN comment thread. It does not scrape or index the comment text itself, so the recall is your sentence and the link, not a searchable copy of the discussion.

Frequently asked questions

Where do my Hacker News saved stories live?

They live in a public URL at news.ycombinator.com/favorites?id=yourname. The list is chronological, paginated, and shows only the story titles and links. There is no private folder, no tag system, and no full-text search built into the page.

Can I search my Hacker News favorites?

Not directly. The official site has no search inside your favorites list. The third-party tool hn.algolia.com searches all of Hacker News by keyword, but it does not filter to only your saved stories, so you have to match by exact phrase and then confirm by hand whether it is one of yours.

How do I export my Hacker News favorites?

There is no official export button. The favorites page URL is scrapeable, and several open-source scripts on GitHub will pull your full favorites list into JSON or CSV. The export gives you titles, links, and dates. It does not capture the comment threads or the reason you saved each item.

What do most heavy HN readers do instead?

The common path is to stop trusting the favorites list as a recall tool and pair it with an external save layer. Some pipe favorites into Instapaper or a Notion database. Others use a memory layer like dEssence where the recall is a natural-language question instead of a scroll.