The hottest Errors Substack posts right now

Large language models like Sora often make up information, leading to errors like hallucinations in their output.
Systems like Sora, despite having immense computational power and being grounded in both text and images, still struggle with generating accurate and realistic content.
Sora's errors stem from its inability to comprehend global context, leading to flawed outputs even when individual details are correct.

Machine learning models can sometimes seem good but fail when applied to real-world data due to complexities that cause overfitting without being obvious
Issues with machine learning models are increasingly reported in scientific and popular media, impacting tasks like pandemic response or water quality assessments
Preventing mistakes in machine learning involves using tools like the REFORMS checklist for ML-based science to ensure reproducibility and accuracy

Research integrity issues are systemic, not just due to individual mistakes. Institutions must take responsibility for addressing these problems.
Tools like Proofig help catch errors in papers, emphasizing the importance of correcting mistakes promptly.
Reducing stigma around correcting papers, proactive responses from institutions, and encouraging a culture of self-correction are crucial for improving research integrity.

System stability is a steady state, not a goal.
Errors are unavoidable in software, but addressing them is crucial.
Reading and addressing errors based on impact and frequency is key to improving software quality.

Get a weekly roundup of the best Substack posts, by hacker news affinity:

The ABI SOLiD sequencer had significant limitations and was not competitive
Color space reads from the SOLiD were not easily converted to DNA sequences
The SOLiD sequencer had issues with error rates and limited utility in sequencing new genomes

Extensive corrections in scientific papers may lead to a retraction if confidence is lacking
Retracting a paper could be considered if there is an accumulation of errors, causing editors to lose confidence in the data integrity
Transparently correcting errors in scientific research is crucial to maintain the integrity of the self-correcting process of science

Trust in science is earned, not demanded.
Journals should prioritize publishing replications to correct errors and earn public trust.
Scientific institutions should adopt the Pottery Barn Rule for accountability and to encourage replications.