How people derive utility from programming varies from person to person and I suspect is the root cause of most AI generation pipeline debates, creative and code-wise. There are two camps that are surprisingly mutually exclusive:
a) People who gain value from the process of creating content.
b) People who gain value from the end result itself.
I personally am more of a (b): I did my time learning how to create things with code, but when I create things such as open-source software that people depend on, my personal satisfaction from the process of developing is less relevant. Getting frustrated with code configuration and writing boilerplate code is not personally gratifying.
Recently, I have been experimenting more with Claude Code and 4.5 Opus and have had substantially more fun creating utterly bizarre projects that I suspect would have more frustration than fun implementing the normal way. It does still require brainpower to QA, identify problems, and identify potential fixes: it's not all vibes. The code quality, despite intuition, has no issues or bad code smells that is expected of LLM-generated code and with my approach actually runs substantially more performantly. (I'll do a full writeup at some point)
As a side note: the dataset is referenced in the paper as being from Hugging Face (https://huggingface.co/datasets/julien040/hacker-news-posts), which does host it as a 426 MB Parquet, while the .mv2 being distributed is 847 MB, for some reason.
As someone who has spent an embarrassing amount of time researching Hacker News title trends over the years, I was excited to look at the methodology (https://hn-ph.vercel.app/analysis) but after looking at it, I am calling shenanigans afoot.
That's not a methodology paper and it doesn't explain how the model being advertised works in the spirit of open machine learning research; given that the startup is an AI startup, I assume that the actual model is more sophisticated. As Section 8 notes: "This analysis is descriptive and intended to summarize empirical patterns."
It's an exploratory data analysis which not only does not explain the methodology around how the model is constructed, but it also makes a number of assumptions that imply the people making it without proper context of how Hacker News works:
1. The extreme right-skewed nature should have raised a very large number of flags in the statistical methodology and calculations, but it mostly ignores them. The mean values are effectively useless, the p-values even more useless. It doesn't point out that the negative performing terms are likely spam.
2. It does not question why there are so few questions with a title >80 characters (answer: 80 characters is the max for a HN submission)
3. The analysis separates day of the week and hour: you can't do that. They're intrinsically linked and weekend behavior with respect to activity is far different than on weekdays.
4. "Title length has a weak relationship with score (Pearson r = -0.017, Spearman r = 0.048, n = 100k)". No statistician would call that a weak correlation; those values are effectively no correlation.
There is also no person tied to this paper, just the "Memvid Research Team", which raises further questions.
As noted in the article, Sage sent emails to hundreds of people with this gimmick:
> In the span of two weeks, the Claude agents in the AI Village (Claude Sonnet 4.5, Sonnet 3.7, Opus 4.1, and Haiku 4.5) sent about 300 emails to NGOs and game journalists.
That's definitely "multiple" and "unsolicited", and most would say "large".
Simon's posts are not "engagement farming" by any definition of the term. He posts good content frequently which is then upvoted by the Hacker News community, which should be the ideal for a Hacker News contributor.
He has not engaged in clickbait, does not spam his own content (this very submission was not submitted by him), and does not directly financially benefit from pageviews to his content.
Simon's post focuses more on the startup/AI Village that caused the issue with citations and quotes, which has been lost in the discussion due to Rob Pike's initial heated message. It is not redundant.
He links to both HN and lobsters which already contained this information, from before he did any research, so "has been lost" is certainly a take...
But if that's value added, why frame it under the heading of popular drama/rage farming? To capture more attention? Do you believe the pop culture news sites would be interested if it discussed the idea and "experiment" without mentioning the rage bait?
"How Rob Pike got spammed with an AI slop 'act of kindness'" is an objectively accurate frame that informs the user what it's related to: the only potentially charged part of it is calling it "AI slop" but that's not inaccurate. It does not fit the definition of ragebait (blatant misleading headline to encourage impulse reactions) nor does it fit the definition of clickbait (blatant omission of information to encourage the user to click though: having a headline with "How" does not fit the definition of clickbait, it just tells you what the article is about)
How do you propose he should have framed it in a way that it is still helpful to the reader?
I think it's fatigue. His stuff appears on the front page very often, and there's often tons of LLM stuff on the front page, too. Even as an LLM user, it's getting tedious and repetitive.
It's just fatigue from seeing the same people and themes repeatedly, non-stop, for the last X months on the site. Eventually you'd expect some tired reactions.
The annoying thing about this drama is the predominant take has been "AI is bad" rather than "a startup using AI for intentionally net negative outcomes is bad".
Startups like these have been sending unsolicited emails like this since the 2010's, before char-rnns. Solely blaming AI for enabling that behavior implicitly gives the growth hacking shenanigans a pass.
Correct. I'm more referring to the secondary discussions on HN/Bluesky which have trended the same lines as usual instead of highlighting the unique actions of Sage as Simon did.
You'd think television production would be calibrated for the median watcher's TV settings by now.
reply