# 65% of Hacker News Posts Have Negative Sentiment, and They Outperform **Author:** Philipp D. Dubach | **Published:** January 7, 2026 | **Updated:** March 15, 2026 **Categories:** AI **Keywords:** Hacker News sentiment analysis, negativity bias social media engagement, BERT RoBERTa sentiment comparison, attention inequality power law social media, Hacker News data analysis engagement, DistilBERT sentiment classification, transformer model NLP comparison, social media negativity bias research, Gini coefficient content distribution, tech community sentiment research, Hacker News score distribution, multi-model sentiment comparison, negative content engagement premium, power law online communities, NLP sentiment classification pipeline ## Key Takeaways - Across 32,000 Hacker News posts, nearly 65% register as negative sentiment, a pattern that holds across six different models including DistilBERT, RoBERTa, and Llama 3.1 8B - Negative posts average 35.6 points versus 28 overall, a 27% engagement premium for negativity, though most HN negativity is substantive critique rather than toxicity - Score distribution follows a power law with high Gini coefficients, meaning a small fraction of posts capture most attention while the majority get almost none --- ## Negativity Bias and Engagement on Hacker News This Hacker News sentiment analysis began with a simple observation: posts with negative sentiment average 35.6 points on [Hacker News](https://news.ycombinator.com). The overall average is 28 points. That's a 27% performance premium for negativity. Hacker News sentiment analysis distribution across 32,000 posts showing negative skew This finding comes from an empirical study I've been running on HN attention dynamics, covering decay curves, preferential attachment, survival probability, and early-engagement prediction. The preprint is [available on SSRN](https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5910263). I already had a gut feeling. Across 32,000 posts and 340,000 comments, nearly 65% register as negative. This might be a feature of my classifier being miscalibrated toward negativity; yet the pattern holds across six different models. ## Six-Model Sentiment Comparison: Transformers vs LLMs Sentiment classification comparison across six NLP models: DistilBERT, BERT Multi, RoBERTa, Llama 3.1 8B, Mistral 3.1 24B, and Gemma 3 12B I tested three transformer-based classifiers (DistilBERT, BERT Multi, RoBERTa) and three LLMs (Llama 3.1 8B, Mistral 3.1 24B, Gemma 3 12B). The distributions vary, but the negative skew persists across all of them (inverted scale for 2-6). The results I use in my dashboard are from DistilBERT because it runs efficiently in my Cloudflare-based pipeline. What counts as "negative" here? Criticism of technology, skepticism toward announcements, complaints about industry practices, frustration with APIs. The usual. It's worth noting that technical critique reads differently than personal attacks; most HN negativity is substantive rather than toxic. But, does negativity cause engagement, or does controversial content attract both negative framing and attention? Probably some of both. ----- ## HackerBook Dataset: Cross-Validation With 22GB of Hacker News Data Related to this, I also saw [this Show HN](https://news.ycombinator.com/item?id=46435308): 22GB of Hacker News in SQLite, served via WASM shards. Downloaded the [HackerBook](https://github.com/DOSAYGO-STUDIO/HackerBook) export and ran a subset of my paper's analytics on it. _Caveat: HackerBook is a single static snapshot (no time-series data). Therefore I could not analyze lifecycle analysis, early-velocity prediction, or decay fitting. What can be computed: distributional statistics, inequality metrics, circadian patterns._ Summary statistics table for HackerBook Hacker News data sample ### Score Distribution and Power-Law Fit Hacker News score distribution CCDF with power-law fit showing heavy-tailed engagement ### Attention Inequality: Lorenz Curve and Gini Coefficient Lorenz curve of Hacker News story scores measuring attention inequality with Gini coefficient ### Circadian Posting Patterns Hacker News circadian posting patterns in UTC showing volume versus mean score by hour ### Score vs Comment Engagement Hacker News score versus direct comments log-log scatter plot Direct comments distribution CCDF on Hacker News showing power-law tail Mean score versus direct comments on Hacker News binned in log-spaced buckets --- ## Frequently Asked Questions ### What percentage of Hacker News posts have negative sentiment? Analysis of 32,000 Hacker News posts found that nearly 65% register as negative. This pattern persists across six different sentiment models including DistilBERT, BERT Multi, RoBERTa, Llama 3.1 8B, Mistral 3.1 24B, and Gemma 3 12B, suggesting it reflects genuine community behavior rather than classifier bias. ### Do negative posts get more engagement on Hacker News? Yes, negative posts average 35.6 points compared to the overall average of 28 points, representing a 27% performance premium for negativity. This aligns with broader research on negativity bias in social media, where negative content consistently generates higher engagement. ### How do BERT and RoBERTa compare for sentiment analysis? Both produce similar negative-skewed distributions when applied to Hacker News data. The negative pattern holds across all six models tested, suggesting the finding is robust to model choice. DistilBERT was used for production due to its efficiency in Cloudflare-based pipelines. ### What is attention inequality in social media? Attention inequality measures how unevenly engagement is distributed across content. On Hacker News, score distribution follows a power-law with high Gini coefficients, meaning a small number of posts receive disproportionate attention while most receive minimal engagement. This mirrors patterns found on Twitter where Gini coefficients exceed 0.9. ### Why is Hacker News so negative compared to other platforms? Hacker News negativity is predominantly substantive technical critique rather than toxicity. The community values rigorous skepticism toward technology announcements, industry practices, and API design. This culture of critical evaluation produces negative sentiment scores even when the discussion quality is high, distinguishing HN from platforms where negativity tends to be more personal or inflammatory. ### How does negativity bias affect engagement in tech communities? Negativity bias, the human tendency to weight negative information more heavily, translates to measurably higher engagement on tech platforms. On Hacker News, negative posts earn 27% more points on average. Research across platforms shows negative content is shared 34-61% more often, with each additional negative word in a headline increasing click-through rates by 2.3%. ### What is the power-law distribution in Hacker News scores? Hacker News scores follow a power-law distribution where a small fraction of posts capture the vast majority of attention. The complementary cumulative distribution function (CCDF) shows a heavy tail, and Gini coefficients are high, indicating extreme attention inequality similar to patterns observed on Twitter and YouTube. --- *Philipp D. Dubach — [http://philippdubach.com/](http://philippdubach.com/) — 2026*