This post is an adaptation of one of the first HN posts that I've read (it was 2013 at that time) and had bookmarked ever since. I loved the approach and the results were very funny, which is why I made an account and kept coming back to HN almost daily, for years now, looking for cool programs like this.
The link is sadly down for a long time now and only available through the wayback machine but since it was only trained on titles before 2014 I thought it would be interesting to train the titles on different years and see what comes out on the other end. Grant, I know you're on HN, if you don't want this online let me know and I remove it :)
Meanwhile, you can try out it out over here:
Disclaimer: I have no control over the generated titles, they are generated according to an algorithm (might be NSFW here and there).
The updated code (Python 3) along with the data to run this locally are located here: https://gist.github.com/KarimJedda/42359052d4a166791b035ae15e7b4a75
My favorites (so far)
- Life in the browser
- I Have Resigned from the internet as we know it
- Docker operations slowing down makes life better
- ‘YouTube recommendations are toxic,’ says dev who worked on more?
- Buttplug: An open-source VR headset that you can also use as RAM is getting language translation
- Mozilla Welcomes the Rust compiler one last time
- Amazon Braket – Get things from one building to another with zero downtime
- Farmers Deserve the Right to Read a Paper (2016) [pdf]
- Thank HN: My personal website is offline
- What Happens When You Aren't a Technical Paper [pdf]
The project here is an aparté to another project that is roughly using the same codebase that I'll be also posting soon. You know how it is, one project leads to the other and on and on...
And I'm on holiday for once, taking it easy with my bro