From memory, you can probably name your favourite books, albums, and movies. Maybe even your favourite play, or favourite painting. But can you recall the best things you’ve read online?
While I haven’t forgotten Paul Ford’s 38,000-word magnum opus in Bloomberg (nor the BuzzFeed post about the wedding dress with the changing colours) there’s not much else from 2015 that sticks out in my mind. I could easily recite the dozens of books, albums, and movies I enjoyed this year, but when it comes to internet writing, my mind draws a blank. So is my memory especially bad, or is internet writing inherently forgettable?
That question inspired me to conduct a little holiday experiment: looking back on the year (and beyond…), which pieces of writing were able to stand the test of time? I decided to use HackerNews as my test subject, since it’s one of the few social networks that privilege written content over other forms of media (like videos, cat gifs and listicles).
Here’s what I found (make sure to skim the list, the interesting part comes after):
The 50 Best* HackerNews Posts (EVER)
- A Brief, Incomplete, and Mostly Wrong History of Programming Languages by James Iry (submitted independently 6 times, 1243 points total)
- How to be a Programmer by Robert L. Read (6 times, 492 points)
- Advanced Programming Languages by Matt Might (5 times, 497 points total)
- Statistical Data Mining Tutorials by Andrew Moore (5 times, 433 points)
- Ten Rules for Web Startups by Ev Williams (5 times, 293 points)
- The Nature of Lisp by Slava Akhmechet (5 times, 261 points)
- Bit Twiddling Hacks by Sean Eron Anderson (5 times, 234 points)
- Computing Your Skill by Jeff Moser (5 times, 226 points)
- The TTY demystified by Linus Akesson (4 times, 645 points)
- The Impossible Music of Black MIDI by Michael Connor (4 times, 499 points)
- How I explained REST to my wife by (4 times, 386 points)
- A regular expression to check for prime numbers by Avinash Meetoo (4 times, 344 points)
- Gamma error in picture scaling by Eric Brasseur (4 times, 343 points)
- What does one trillion dollars look like? by Joe Barta (4 times, 324 points)
- The Absolute Minimum Every Software Developer Must Know About Unicode and Character Sets by Joel Spolsky(4 times, 319 points)
- The secret guild of Silicon Valley by Michael E. Driscoll (4 times, 302 points)
- The Graphing Calculator Story by Ron Avitzur (4 times, 296 points)
- What every computer science major should know by Matt Might (4 times, 266 points)
- A Whirlwind Tutorial on Creating Really Teensy ELF Executables for Linux by Brian Raiter (4 times, 250 points)
- Modern art was CIA weapon by Frances Stonor Saunders for The Independent (4 times, 226 points)
- Odd Comments and Strange Doings in Unix by Dennis M. Ritchie (4 times, 212 points)
- Interview with John McCarthy, Father of Lisp by Guy Steele (4 times, 206 points)
- The Evolution of a Haskell Programmer by Fritz Ruehr (4 times, 199 points)
- OCaml for the Masses by Yaron Minsky and Jane Street (4 times, 194 points)
- Why bitcoin will fail by Avery Pennarun (4 times, 129 points)
- Groupware Bad (2005) by Jamie Zawinski (4 times, 119 points)
- Detaining my partner was a failed attempt at intimidation (2013) by Glenn Greenwald for The Guardian (3 times, 1474 points)
- There’s no speed limit by Derek Sivers (3 times, 1191 points)
- Why GNU grep is fast by Mike Haertel (3 times, 1158 points)
- Don’t Call Yourself A Programmer by Patrick McKenzie (3 times, 1110 points)
- The First Few Milliseconds of an HTTPS Connection by Jeff Moser (3 times, 871 points)
- Girls and Software by Susan Sons for Linux Journal (3 times, 798 points)
- Brown Out by Snopes (3 times, 777 points)
- The Brutal Ageism of Tech by Noam Scheiber for New Republic (3 times, 645 points)
- The Sierpinski Triangle by Antonio Marquez-Raygoza (3 times, 629 points)
- The Socratic Method by Rick Garlikov (3 times, 569 points)
- 10 Technical Papers Every Programmer Should Read (At Least Twice) by Michael Fogus (3 times, 568 points)
- Re: [git pull] drm-next by Linus Torvalds (3 times, 515 points)
- Polygonal Map Generation for Games by Amit Patel (3 times, 507 points)
- Essential Math for Games Programmers by James Van Verth and Lars Bishop
- How Not To Sort By Average Rating by Evan Miller (3 times, 490 points)
- Lisp as the Maxwell’s equations of software by Michael Nielsen (3 times, 471 points)
- Great Works in Programming Languages Collected by Benjamin C. Pierce (3 times, 471 points)
- The $5000 Compression Challenge by Patrick Craig (3 times, 464 points)
- The illustrated guide to a Ph.D. by Matt Might (3 times, 463 points)
- Books Programmers Don’t Really Read by Bill the Lizard (3 times, 460 points)
- Let’s Build a Compiler by Jack Crenshaw (3 times, 459 points)
- Your body wasn’t built to last by Brian Skinner (3 times, 454 points)
- In Praise of Idleness by Bertrand Russell (3 times, 431 points)
- How to Fold a Julia Fractal by Steven Wittens (3 times, 418 points)
Yeah, so about being the best*
Check the front page of HackerNews any day of the year, and you’ll never find a collection of links as wonderful as this. (I say this with absolute certainty, despite not knowing a lick of code myself!) Entertaining, didactic, outrageous — every submission you see here is certifiably timeless. And you don’t have to take my word for it: these links have been submitted to HackerNews on multiple occasions, garnering at least 10 points per submission (which is roughly the amount of points required to reach the front page).⁰
It’s rare that a story gets submitted to HackerNews more than once. Rarer still is a story so compelling that it hits the front page multiple times.
Which blogs are the best?
Dig a bit deeper into the data, and you’ll find even more surprises. For instance, who are the authors of those rare, timeless pieces?
These are 50 publishers whose stories have been submitted to HackerNews more than once.
You’ll notice that billion dollar media outfits — BBC (9 stories), Bloomberg (9), The Washington Post (7), The Verge (7), The Wall Street Journal (6)— have written fewer timeless stories than mortals like Paul Graham (19), Matt Might (13), James Hague (10), and Ben Horowitz (9).
What’s so special about Matt Might (13) that makes him able to write more “timeless” stories than the entire editorial team at TechCrunch (11)? There are a few reasons. First, writers at major media hubs are expected to publish multiple stories a day, whereas a part-time blogger like Matt (who doesn’t need to blog to pay the bills) can take his time writing a piece until it meets his standard of quality. It’s not as easy to produce something of lasting value when you work for a content farm.
Second, Matt is writing directly for the people who read HackerNews, whereas mainstream media shoots for a broader audience. Nevertheless, it’s reasonable to expect that with their money and resources, mainstream media could find the time to write more than a dozen stories with enduring appeal to hackers.
Those two factors can explain the triumph of small bloggers over large media corporations (within the microcosm of HackerNews), but they can also explain the rise of Medium itself.
By reducing the expectation of prolificacy (since contributors can publish as infrequently as they’d like), Medium encourages its writers to wait until they have something worth writing about before posting. There’s no blog to maintain, no content editor to appease — Medium lets you take your time and decide which stories you want to put your name behind.
What a big surprise, therefore, that those are the type of stories people like to read, again and again!!! In its three years of existence, Medium has published just as many “timeless” stories as The New York Times (33), although granted, Medium publishes more stories per day.
Altogether, I think this little experiment brings up some pretty interesting questions about the longevity of writing on the web. It’s clearly possible to write things that remain culturally relevant for years— but how can we ensure that timeless pieces aren’t forgotten about? Will it ever be possible for timeless content to compete on an equal footing with shiny new content on Twitter and Facebook? And what strategies will publishers like Medium develop so that timeless stories can continue to find new readers over time?
Joe MacNeil is a writer based in Montreal, Canada.
If you really really liked what you read (or want to collaborate on future experiments) hit me up on Twitter.
PS: Since this was just a fun holiday side-project, there are some limitations to the analysis I carried out, which I address in the HackerNews comments here.
PPS: If you’d like to download the full data set, help yourself.
PPPS: Huge thanks to Michal Pawlowski for crunching the numbers for this analysis.