Hey everyone! I love how we often talk about older classic articles here (posts with the year in the title like [0]). I haven't seen that practice on other sites, which is a shame. For me these posts often lead to the most valuable discussions.

So I built this small website [1] to browse and search across all timeless HN posts [2]. The idea is to make use of the immense library we're all curating by posting and commenting here. What you can do with the site:

• Browse the best links of all time, according to HN: https://hn.lindylearn.io/best

• Search for resources for example on "career advice" (prioritising old content that's still reposted frequently): https://hn.lindylearn.io/search?q=career%20advice

• See the best links per publishing year. For example, which articles from 1995 do we still talk about? https://hn.lindylearn.io/years/1995

• What are the best posts on any website? https://hn.lindylearn.io/from?site=paulgraham.com

It was a quick project to put together, but I hope it's somewhat useful not just for myself. Halfway through I realised there's some prior art for it [3], but I think this site adds enough to make it worthwhile. I'm curious to hear what you think.

[0] https://news.ycombinator.com/item?id=29590667

[1] Named "Lindy Hacker News" after the Lindy Effect https://en.wikipedia.org/wiki/Lindy_effect

[2] It indexes all HN posts with a year in the title that have more than 40 upvotes, where the article publish date is >= 2 years before the post date (results in about 10k posts). Articles are ranked by the summed score across all posts for the url.

[3] https://news.ycombinator.com/item?id=16442888

This is awesome but this is also my worst nightmare. I now have even more links to read and threads to click on… this is like bringing back the whole bag of chips instead of putting a reasonable portion into a bowl

Also the website design is really elegant and well done.

Haha I am sorry. The idea is to come back to the site when you're curious about something, and be ok with not knowing everything (like with a library). That's different to how we consume a lot of our news, so just an experiment.

At least the recent "timeless" posts view on hn.lindylearn.io/ brings down the daily HN posts to ~4.

One request would be if possible to link to the web archive of the link rather than the original. I just ran across one that no longer works because their wordpress site was made private, for example.

I have thought about this, but not am not sure how to automatically detect if a site is dead. Sometimes it still returns HTML but with permission errors, redirects etc. I don't want to redirect every link to archive.org.

Do you have an idea? What's the link that is broken for you, so I can manually fix it?

Not sure beyond sending them all to the archive. Unless you check periodically or something?

This is the one: https://rumorsontheinternets.org/2010/10/14/magnasanti-the-l...

Mhh, maybe I'll add an "archive" button you can click to see the cached page easily - though the challenge with that is explaining the feature...

I fixed the simcity link, thanks for reporting!

Something similar

http://jsomers.net/hn/

https://news.ycombinator.com/item?id=16442888

A few people must have had this idea over the years. HN is the only website I know where old articles get regularly reposted.

I actually discovered http://jsomers.net/hn/ while building this site, as it turns up when you search for "hacker news". It's a bit different so I think both tools are useful.

> HN is the only website I know where old articles get regularly reposted.

I digress, but have you opened your Facebook feed recently? :-)

I have not, but I would imagine it to be full of news? That's not the case for you?

My point was that I often see "the algorithm" recommending me many-years old videos out of nowhere. Anyway!

Ah, I see what you mean (at least in the case of YouTube). I'd actually like that behaviour weren't it so addictive.

A computer scientist might find it confusing, 1321-2021 is not the date range on HN :)

https://hn.lindylearn.io/years/1321 for anyone else hunting

for some reason it doesnt show up in the decade view https://hn.lindylearn.io/years/1320

Oh, https://hn.lindylearn.io/years/1320 only shows things written exactly 1320, there is no "decade view" right now. That's an interesting idea though, maybe I can put it at https://hn.lindylearn.io/years/132X.

Edit: Done now, for example visit https://hn.lindylearn.io/years/198x for the best links of 1980 - 1989. Only works with one X for performance.

ah thank you! i misunderstood the 'go back a decade' and 'go forward a decade' links to mean the destination lists the decade's links :) this is very welcome!

https://hn.lindylearn.io/years/194x is terrifying! and makes me feel better about living in 202x...

Yes, to be precise the year refers to the original date the text was written. Hopefully it has not totally changed its meaning since then.

Interesting project. 1864 seems to be the first year that has two completely different items meaningfully discussed.

Oh god. What did you do?

Now I have 500 more HN stories to read through tonight. And I was just about to go to bed.

Obligatory Ian Malcolm: https://www.youtube.com/watch?v=g3j9muCo4o0

Well, you don't have to read through all stories now - just check back later when you're curious again ;) But I know what you mean.

You’re missing the Last Trial By Combat in France, 1386

https://news.ycombinator.com/item?id=22709166

Mhh, it looks like that article was written in 2020 not 1386 though? The site only includes posts that were at least supposedly written further back in the past.

Then I’ll have to find some posts from the lawyers and articles leading up to the trial in 1386

That would be interesting, as an unbiased window into life back then (if any text still exist).

If the post gets 10 upvotes on HN it will automatically turn up on hn.lindylearn.io/years. It's like our collective library.

This is beautiful. Nice idea and well executed. Congrats!

Hey! This is really well made.

May I ask what the tech stack is?

Thank you! Most of the data can be queried and aggregated through the HN BigQuery dataset [0], and I run a periodic Node.js function to poll more recent posts from the API. The filtered posts are hosted in Firestore for most of the pages + Relevance.ai for the search. The frontend uses Next.js and the original HN CSS + small tweaks like the year display.

[0] https://console.cloud.google.com/marketplace/product/y-combi...

Aren't these really "timed"?

What do you mean? Yes the articles itself are old, but the discussions around them are new. Which is the interesting part -- "timeless" content that's relevant not just the week it is published.

I was just making the joke that these "timeless" articles all have times.

That is true ;)