Apurva Shukla/writing

08-09-25: Internet goodies I read last week

  1. In User Research, Don’t Stop at “Yes” or “No”

    • "The purpose of research is to inform decisions. That means the insights coming out of research must be actionable and inspire next steps. A binary answer accomplishes the opposite: it creates the impression of finality. Researchers asked to perform binary validation will gather binary data: quotes and measurements that back one or the other result. But without insight into why, both “yes” (we succeeded) and “no” (we failed) present a dead end. This is the reason why teams engaged in validation are pressured to come back with a “yes”: a flat-out “no” means that all the work up to that point was a waste. But a hypothesis will rarely be completely right or completely wrong, and the data gathered through research needs to be able to capture that nuance. Teams that can learn how their idea succeeded, and simultaneously how it failed, gain much more value out of their research, because it creates a complete picture of the product’s strengths and weaknesses. Any “no” is far easier for stakeholders to swallow, because it comes with a caveat (“not in this context”) or a path towards remediation (“not without this feature”). A product leader using research solely as a tool for validation will end up either getting a far too low return on investment (ROI) or throwing away the promising parts of a solution together with the mistaken assumptions. But a nuanced understanding of the solution’s strengths and weaknesses creates an iteration mindset that will lead the team towards much greater value."

  2. Overkill Objects for Everyday Life

    • Delightful article on an equally delighting article. Maybe sometimes, going overboard with gear meant for pros or the military is actually smart. I don't know how to describe it well, but I see cohesion between this website aesthetic (clean, minimal, like a NASA manual from the 70s) , the content of this article (timeliness, professional, tested, functional products), the work this person does (clean futuristic typography) and the professional life they have had (space/industrial engineer). Goals

  3. TX-02 Berkeley Mono™ Typeface

    • Berkeley Mono™ is a sleek, no-nonsense monospace typeface designed specifically for coding, blending the charm of vintage computing with modern precision. It’s all about clear, distraction-free readability, making it perfect for developers and anyone dealing with detailed text. Plus, it comes with flexible licensing and customization options, so businesses can easily adapt it to their needs without fuss. The key takeaway? A good font isn’t just about looks—it's about making your work flow smoother and your projects stand out subtly but confidently.

  4. Mitchell’s New Role at HashiCorp

    • Discovered this press note that details that the co-founder of HashiCorp took an IC role after getting it past IPO. Wonder how that experience was like. Would love to hear a podcast/read about this experience.

  5. How to Sell if Your User is not the Buyer

    • Interesting article on some GTM dynamics, some which I have experienced personally. In large enterprises, the user of your tool is often not the buyer. How do you make the sale them? This article says that you need to understand actually holds the power—whether it’s the dev using the tool or the CTO signing the check. The key takeaway? A lot of times, the power is with the dev when its a technical tool. In that case, you need to empower the user with clear, real data to make the case internally, because they’re the real champions who push the purchase forward or at least the leadership looks upto them. The article says that you should focus on making the users look good, but I don't think its that simple. It's a real challenge though.

  6. Gene Pool Engineering for Entrepreneurs

    • Fav find of the week. Thanks Maruthi. Interesting learnings from Vinod Khosla on hiring.

  7. No, AI is not Making Engineers 10x as Productive

    • The hype around AI making engineers 10x more productive? It’s mostly just that—hype. After diving deep into various AI tools myself, I found they’re great for boilerplate code or one-off scripts but still struggle with complex, real-world coding and context. The big takeaway? AI can help, but it won’t magically speed up all parts of engineering work or squash the human processes that really matter.

  8. How we will be measured

    • Nate Soares is a big figure in the AI Safety space and his book 'Replacing Guilt' was recommended by the Anthropic co-founder in a podcast. The way to think about your life is that its an "act to determine the shape of our actual universe-history". I have the book with me, and will post a summary here in a bit.