This blog post argues that speed is a "magical" and transformative, yet often overlooked, feature in software. The author posits that fast software changes user behavior, eliminates cognitive friction, and signals product discipline and simplicity. By stripping away non-essential features and forcing technical honesty, speed becomes an expression of respect for the user. Examples like Superhuman's sub-100ms response time and Instagram's optimistic photo uploads are cited as instances where speed creates a superior, almost magical experience. The author extends this idea to the current era of AI, noting that while today's LLM workflows are faster than manual human processes, they are still clunky. The post concludes that a future focus on optimizing the speed and experience of AI tools will unlock new, unforeseen capabilities.
The Importance of Speed in Software: Key Arguments and Discussions
The Core Premise: Fast is a Transformative Feature: The article's central argument is that speed is a critical, though often unrequested, feature. Fast software eliminates cognitive friction, changes how users interact with tools, and can feel "magical." Examples include Raycast for quick application launching and Mercury for instant bank settlements.
Fast Signals Discipline and Simplicity: To make software fast, developers must often strip away non-essential features and handle complexity in the background. This brutal honesty forces discipline and signals a deep focus on what truly matters to the user. A key example cited is how Instagram would optimistically begin uploading a user's photo while they were still writing the caption to make the process feel instant.
The LLM Speed Paradox: While the author suggests LLM workflows are a massive speed improvement over manual work, many commenters disagree. A prominent theme in the discussion is that LLM-assisted tasks are often slow, taking minutes to return results that may be convoluted or incorrect, ultimately requiring more time to fix than if the task were done manually.
Developer Experience and Iteration Speed: The concept of "fast" extends to the development process itself. Commenters highlight that rapid compilation and deployment cycles are crucial. The debate between Go (known for fast compiles) and Rust (known for pulling many dependencies and slower compiles) is mentioned as an example where development speed can be more critical than raw performance for maintaining flow.
Users Assume Speed, But Notice Its Absence: A key takeaway from the discussion is that while users don't explicitly ask for "fast," they assume it as a baseline and are easily frustrated by slow interfaces. "Fast" becomes a powerful marketing tool when a competitor demonstrates that a slow product could have been much faster. The simple, fast interface of Hacker News itself is frequently praised as a prime example of this principle in action.
The Business Value of Speed: Several commenters share anecdotes highlighting the direct financial impact of performance. In one notable story, a VP states, "every saved second here adds $1M to our yearly profit," making a clear case for prioritizing performance optimization.
When Fast Isn't the Only Goal: The discussion also provides nuance, noting that speed is not always the primary virtue. For example, a slow, phased deployment is safer and more reliable than a millisecond-fast one that risks system-wide failure. Similarly, some systems (like TurboTax) have found that being "too fast" can make users distrust the results, necessitating artificial delays to build user confidence.