It may console you: in some sense, the top 4900 are more valuable than the top-100.
Why? Everybody here knows Paul Graham. I know Krebs and Schneier, most of you will, too. In a long tail distribution like this, the top entries (left) are the obvious ones, the lowest frequented ones (right) might be noise (artifact of the methods e.g. bugs in the data cleaning), but the middle part is really where the value is: blogs we don't know but would like to know.
In search engine ranking, people needed a lot of time until the late Karen Spärck Jones finally discovered IDF (inverse document [collection] frequency) in 1972, the "Yang" to raw term frequency (TF), which had been the "Yin" that was missing a counterforce to retrieve truly relevant documents when balanced in the TFIDF formula.
So, plea to the OP: please release the rest of your list (101-100000).
+1 to this. I'd also argue that some on the list are unapologetic self-promoters like Simon Willison. Nothing wrong with it but it shows and I think it's much more impressive to be below that cohort but still only a reasonable distance away.
FYI, in my May 2023 survey of HN's archived front pages (under the "past" link in the HN titlebar) ... horse.sheep doesn't appear at all.
That's looking at just the top <=30 stories per day. Your high-water mark seems to have been 2022-12-19, with 88 points / 32 comments, appearing on the 3rd page of the daily archive, ranked #73:
Where does the "bio" field come from? Mine says "Developer and writer" and I suppose those are both things that I am, but not very close to what I'd have put there.
I'm guessing it's annotated by an LLM. Would be a lot of thankless work otherwise, so don't really blame the author, but it means you get the occasional nonsense summary.
They're about 95% accurate, so not completely incorrect. I think the value of the correct ones is higher than the few incorrect ones. I manually review and fix them, but for just a fun tool, it's not practical to write 5000 eight-word bios.
I started by writing them by hand, but it was taking forever to read enough of each author's blog to write a summary and topic list, so I used an LLM and then spot-checked.
I just updated yours, but let me know if you'd like something different.
I love that dynomight.net stands out with "existential angst" as a very unique category among the top blogs, as well as being written by an anonymous/pseudonymous author. I'm a big fan of their writing.
Also quite surprised to find my own site in the top 5000 for the past 5 years! It feels like Hacker News is simultaneously quite large but also a cozy community where you often recognize names from day to day.
This is something I wanted but I couldn't figure out a way to do it in a way that's meaningful. Authors like Simon Willison publish frequently, so even though he has a lot of high-scoring posts, he has a lot of low-to-no-scoring posts too. It feels unfair to penalize people who publish frequently just because not every post is a homerun.
Note that this is gonna be skewed pretty heavily toward domains that have existed for most of HN's history, at the expense of any newer domains that had fewer chances to rack up points.
If you look at any 2-4 year period, the ranking tends to be quite different. Well, Paul Graham is there pretty consistently, but everything else changes.
You can change the date ranges (e.g. just the YTD, or last 12 months, or set a custom range), and it gives an interesting overview of the evolution over time.
Like jvns.ca drops off the list entirely for 2025, but was consistently in the top 5 until last year.
I’ve been thinking of starting an anonymous blog with my thoughts just to record them and any projects I do. I want them to be visible on the internet and searchable instead of behind some facebook or instagram wall. What is a good blog service to use that will be around for decades? I don’t really want to run my own domain. Do things like Blogger and Blogspot still exist and will they continue to in the future?
I had to fiddle with the dates to find a couple examples of blogs that violate the single-author rule in the methodology (marginalrevolution, ribbonfarm) but it's probably better to have them included.
Even though a glance suggests the majority of high-scorers are self-hosted, I wonder if this dataset is valuable for predicting the strength of different blog hosts. Some fiddling did lead to a couple results that are hosted on Blogger or Ghost or Medium, so they are there.
That's weird that I'm excluded. Thanks for watching out for me :-)
Hypotheses: 1. it's an error. 2. I have powerful enemies. 3. Someone from the future is trying to stop me. 4. My blog triggered the FDIV bug and needed to be excluded.
There are a lot of odd exclusions on that list. Just spot-checking, I see blog.plover.com, the blog of Mark Jason Dominus, who by the way is looking for a job[1].
Also, dtrace.org is excluded, which hosts four individual blogs that surely should qualify.
I think what happened was that I was going through the list of domains and assumed that "righto.com" would be too valuable a domain name for a personal blog and excluded it without checking. Sorry about that!
paulg's blog shouldnt count as for its an extension of this and more of a long game sales pitch tailored for different purposes. Not a bad thing, but I just wouldnt consider it a blog.
I think that is wholly unfair. If Paul Graham is anything its a writer/creator first. Those articles have also been extremely influential to many entrepreneurs and people in the business world. I'm personally appreciative of them, even if I don't always agree with him.
Interesting that if you were to combine AstralCodexTen and SlateStarCodex it would be around top #20. Even with the traffic split he's in the top 50 twice.
136 comments
[ 4.3 ms ] story [ 231 ms ] threadWhy? Everybody here knows Paul Graham. I know Krebs and Schneier, most of you will, too. In a long tail distribution like this, the top entries (left) are the obvious ones, the lowest frequented ones (right) might be noise (artifact of the methods e.g. bugs in the data cleaning), but the middle part is really where the value is: blogs we don't know but would like to know.
In search engine ranking, people needed a lot of time until the late Karen Spärck Jones finally discovered IDF (inverse document [collection] frequency) in 1972, the "Yang" to raw term frequency (TF), which had been the "Yin" that was missing a counterforce to retrieve truly relevant documents when balanced in the TFIDF formula.
So, plea to the OP: please release the rest of your list (101-100000).
That's looking at just the top <=30 stories per day. Your high-water mark seems to have been 2022-12-19, with 88 points / 32 comments, appearing on the 3rd page of the daily archive, ranked #73:
<https://news.ycombinator.com/front?day=2022-12-19&p=3>
Edit: wait, that is actually how many points I got. Eh, I am still on the list.
interesting site.
https://github.com/mtlynch/hn-popularity-contest-data/pull/1...
Edit: yeah, the "methodology" page confirms this.
So just don't include them, rather no info than completely incorrect LLM hallucinations
They're about 95% accurate, so not completely incorrect. I think the value of the correct ones is higher than the few incorrect ones. I manually review and fix them, but for just a fun tool, it's not practical to write 5000 eight-word bios.
I just updated yours, but let me know if you'd like something different.
https://github.com/mtlynch/hn-popularity-contest-data/pull/2...
The combined score puts me at #71 all time or #31 since 2019 when I started writing. Very cool.
21 the last 12 months.
https://github.com/mtlynch/hn-popularity-contest-data/pull/1...
If there's a descriptor you'd prefer, I'm happy to update it. I'm just going by what I know of the blog and a review of the most popular posts.
[0] https://lapcatsoftware.com/articles/disclosure2.html
[1] https://lapcatsoftware.com/articles/disclosure.html
[2] https://lapcatsoftware.com/articles/2023/7/1.html
https://github.com/mtlynch/hn-popularity-contest-data/pull/4...
But privately, I think of you as a security researcher, and there's nothing you can do about it. : )
Thanks for your writing and your security findings!
Also quite surprised to find my own site in the top 5000 for the past 5 years! It feels like Hacker News is simultaneously quite large but also a cozy community where you often recognize names from day to day.
- the blog domains which I tend to comment on (relative to other people, not in absolute terms), or
- the people whose comments I most often reply to?
This is something I wanted but I couldn't figure out a way to do it in a way that's meaningful. Authors like Simon Willison publish frequently, so even though he has a lot of high-scoring posts, he has a lot of low-to-no-scoring posts too. It feels unfair to penalize people who publish frequently just because not every post is a homerun.
I'm open to suggestions!
I'm almost positive Paul Graham would be #1.
If you look at any 2-4 year period, the ranking tends to be quite different. Well, Paul Graham is there pretty consistently, but everything else changes.
Like jvns.ca drops off the list entirely for 2025, but was consistently in the top 5 until last year.
I've apparently put out some serious bangers to end up in such esteemed company.
https://refactoringenglish.com/tools/hn-popularity/?end=2025...
Ofc GitHub is will probably last longer, and so will running your own hosting/deployment
Even though a glance suggests the majority of high-scorers are self-hosted, I wonder if this dataset is valuable for predicting the strength of different blog hosts. Some fiddling did lead to a couple results that are hosted on Blogger or Ghost or Medium, so they are there.
Honestly HackerNews has been a great place to grow up. Started posting here back in college when I was 21 or so. Now here we still are at 37.
I credit hacker news with getting me from Slovenia to San Francisco. It's been a great journey so far. Some of which has made it to the front page <3
https://github.com/mtlynch/hn-popularity-contest-data/pull/2...
Sincerely, your biographer. : )
Looks like it's explicitly excluded[2].
[1] https://news.ycombinator.com/from?site=righto.com
[2] https://github.com/mtlynch/hn-popularity-contest-data/blob/d...
Hypotheses: 1. it's an error. 2. I have powerful enemies. 3. Someone from the future is trying to stop me. 4. My blog triggered the FDIV bug and needed to be excluded.
Also, dtrace.org is excluded, which hosts four individual blogs that surely should qualify.
[1]: https://mastodon.online/@mjd@mathstodon.xyz/1142231895042721...
Whoops, that was a mistake. Fixed now: https://github.com/mtlynch/hn-popularity-contest-data/pull/2...
>Also, dtrace.org is excluded, which hosts four individual blogs that surely should qualify.
I didn't realize the authors were on distinguishable URLs, so I've now added them back and canonicalized them to their new subdomain URLs.
But thanks for fixing.
This morning I am jolly and cheerful. Thanks again!
That was an error.
I think what happened was that I was going through the list of domains and assumed that "righto.com" would be too valuable a domain name for a personal blog and excluded it without checking. Sorry about that!
Ken is #8 of all-time now.
https://github.com/mtlynch/hn-popularity-contest-data/pull/2...
Sorry again!
HN has been a fantastic community for me.
Nobody should listen to me. I have no idea what I'm doing.
But my name is Lars Doucet, not Keith Burgun (that's KeithBurgen.net)
Happy to be included!
Fixed now: https://github.com/mtlynch/hn-popularity-contest-data/pull/1...