Probabilistically Generating HN Post Titles
Languages and libraries
text generation
SHOW HN2013
Legend
Probabilistically Generating HN Post Titles
Probabilistically Generating HN Post Titles
grantslatton.com683points
260comments
#8of 1,739 in 2013
Top 1%of 2013 by points
How HN took it
LovedMost comments that take a side are positive
167 positive, 52 neutral or asking, 0 mixed, 5 critical, out of 224 comments about the project. Thread score +0.69.
Idea: 58 for, 1 againstExecution: 33 for, 4 against
The thread, comment by comment
Each row is a top-level comment and its replies, in HN's order. Select a cell to open that comment on HN.
positive neutral or mixed critical about another commenter or something else the maker
What HN said
What people liked
Very funny output
From 3 comments
That is literally hilarious. "Nobody’s going to download your bullshit app" , "HN will be retired, transition to Google Wallet Anti-Competitively" , "The best interface is no…
jiggy2011 on HNTurn it into a game. "Which of these is real? [A] vs [B]" give people ten pairs of headlines and score them. > MongoDB Gotchas and How To Avoid Getting Fired From Your Own Company
DanBC on HNThis is the greatest thing I have seen on Hacker News in a very long time, and I mean that. This is genius, the titles that it generates paint a pretty clear trend of topics HN…
DigitalSea on HN
Objections
Some generated titles are real or near-duplicates of real posts
From 2 comments
Suggestion: filter the titles that are identical to real ones. Or maybe it was part of the plan, demonstrating how they are indistinguishable. ``` My deadly disease was just a…
BoppreH on HN``` Design Books Every Hacker Should Read (At Least Twice) ``` I could swear I've read that post (at least twice).
Groxx on HN
Requests
Turn it into a guessing game of real vs fake titles
From 1 comment
Turn it into a game. "Which of these is real? [A] vs [B]" give people ten pairs of headlines and score them. > MongoDB Gotchas and How To Avoid Getting Fired From Your Own Company
DanBC on HN
Better duplicate filtering
From 1 comment
If you look at the code, I filter all titles that are substrings (including identical) of real ones. Sometimes the Markov chain will simply add a single word to it, though, so it…
gamegoblin on HN
Questions
How to improve the filter against near-copies
From 1 comment
If you look at the code, I filter all titles that are substrings (including identical) of real ones. Sometimes the Markov chain will simply add a single word to it, though, so it…
gamegoblin on HN