Short-form insights you can read in under 5 minutes.
An effective technical team leader learns what they need to know in ways that help the team get their essential work done.
Use what you learn from both to provide the transparency and predictive planning you need for the larger organization.
As tech workers, when we tell ourselves, “I have to do this.” We are dodging accountability. Even when the downside risk of a decision is dire, it is still ours to make. The societal and environmental consequences — both intended an unintended — are on us. That’s the job.
Working through issues with windows Mark of the Web (MOTW). Evocative naming from Microsoft. Quite rare from the people who gave us “ActiveX”, WPF and Bob.
For those “elder” statespeople still building software. To thrive in this immature, ageist industry — and assuming you haven’t done so on other peoples’ backs — you’ve developed some singular strengths that make you capable of moving faster, driving change, and bringing people along with you.
You recognize patterns, combine them in novel but practical ways, and communicate them. This is a prize age and experience have given you.
You embrace new tools and ways of doing things while holding onto your craft (contrary to stereotype). You’ve done it many times before. To have done it well you have learned humility and take joy in the challenge.
In short, your wisdom and learning mindset are a gift. If you share them generously, a superpower.
One of my recurring Anthropic Claude prompts: “First, lets add test cases that will fail until I make this change.” That’s after using the model to reason through the problem.
In the bank validation is a beautiful thing. Turns out technical leaders on a legacy rewrite don’t want AI magic. They need to reduce the cost and risk of knowing what the old code actually does and then recommendations for how to reimplement anchored in their chosen architectural approach — so their teams can build a better solution — the work they were hired for and love.
I developed a static analysis tool to navigate the complexities of legacy .NET code. Our open-source C# demo app embraces modern architectural patterns, presenting exciting challenges like identifying interface implementations, mapping code to database schemas, and managing schema alterations. This approach enhances context for LLMs, leading to improved quality and reduced compute costs—an essential step in responsible engineering.
People will use Generative AI to produce poor code at loss of usability, reliability, privacy, and safety for other people.
This can also be true of code written by humans — and for the same reasons — the culture, priorities, and values of organizations as reinforced by leaders.
Well there you go. I thought the general claude chat UI was drifting from my use case of code analysis, reasoning partner, and code suggestion tool.
This should get sonnet 3.5 back on track as my go to productivity enhancer.
Legacy code has incidental complication that constantly impresses me. Form submit actions that touch 100+ database tables. Applications consisting of 1000 compiled libraries, 10,000’s of code source files.
I believe in curating what I revere as my traditions. As with values, what matters to me is whether i follow through on them. How they make me and the people I care about feel. How my actions effect others.
I guess, in an attempt to manage highly subsidized infrastructure the frontier model providers set up their agents to provide “…you can fill out the rest…” responses.
Which feels like, “I, bot, do thinking, you, human, do rote typing.”
This is, of course, a— backwards.