No Sloptober challenges engineers to abstain from LLM tools for a month

The event is a personal challenge rather than a corporate mandate, positioning the abstention as a mental fast to develop individual nuance regarding LLM utility. It explicitly rejects the idea that AI use is a moral failing, instead framing it as a skill maintenance exercise. The core premise is that relying on LLMs can obscure an engineer's actual capabilities and create a dependency that hinders learning.

The material cites specific technical risks, including the tendency of LLMs to generate excessive code cruft, over-engineer solutions, and make dangerous security choices. It references the 'Shai Hulud' incident as an example of agents installing malware by accident. The proposed 'Onarheim's Law' suggests that while agents maintain or increase entropy, humans are uniquely capable of decreasing it, implying that manual intervention is necessary for system stability.

For the working engineer, the cost of participation is a temporary reduction in productivity and the loss of AI-assisted convenience. The challenge requires active effort to replace AI workflows with manual processes, such as writing blog posts about the experience or performing cost-risk analyses on LLM usage. It also demands a shift in mindset from delegating tasks to understanding the underlying gaps in one's own knowledge.

The initiative stops working if the engineer is under corporate duress to use AI to maintain employment, a reality the text acknowledges with a footnote about 'AI under duress.' It also makes exceptions for language translation tools, recognizing their unique value in global participation. The challenge is not a permanent ban but a diagnostic period to determine which tasks truly benefit from automation and which require human judgment.

The framing differs from typical AI adoption news by focusing on the negative externalities of over-reliance, such as 'AI psychosis' and the 'Meat Proxy' phenomenon where users passively accept AI output without review. It encourages engineers to measure their team's velocity, incident rate, and cost with reduced LLM usage to find potential savings or risk mitigations. This approach treats the month as an experiment in software development's 'iron triangle,' forcing a choice between good, cheap, or fast outcomes.