Rafe Colburn’s Post

My favorite thing about LLMs is that they attempt to answer the specific questions they are asked. They don't answer your question with a question. They don't try to replace your question with one they're more confident they can answer. They don't try to figure out why you're asking the question in the first place and give you a self-serving response. They just take the question at face value and try to give you what you asked for. (That's not to say they're unbiased, but I find the biases acceptable.) In the end, I find LLMs almost refreshing in this regard. LLMs aren't always right (neither are people), but the fact that they address questions directly makes their answers easier to evaluate. I will also add that find this pretty inspirational in that I also want to give straight answers to questions, and my own experiences remind me that most people really crave straight answers as well.

Is anyone else finding that Slack performance has gotten really bad? I think the issue is the time for the client to get fully in sync with the server when you wake up your laptop. It takes a long time to receive all the messages you need to download and for the server to process newly sent messages as well (you can see how long it takes for your messages to be editable after uploading them). At this point I think it's a pretty big usability issue. The latency between getting a push notification about a new message on your phone and getting the message in the Slack client on your laptop is routinely shocking.

This is one I use a lot when managers seem wishy-washy about someone’s performance — if they said they were leaving, would you try to talk them out of it?

Whenever people suggest adopting practices from Google I ask them how useful they find vacation recommendations from their richest friend.

I loved this interview with Richard “Good Strategy, Bad Strategy” Rumelt. In some cases I think the interview is enough but I encourage folks to read this book. What’s interesting to me is that the book is recommended frequently, but I never see the approach applied. GK Chesterton is said to have said that the Christian ideal was not tried and found wanting, but rather found difficult and left untried, and I suspect that this is the case with Rumelt’s approach to strategy as well. The discipline demanded is hard to cultivate.

It's compensation and promotion season, and there's one heuristic that I stand by above all others. I have repeated it so many times that recently someone told me that when they used it, the other person said, "Sounds like you've been talking to Rafe." The heuristic is that we should level and compensate people based on what it would take to replace them, either in terms of responsibility or output. A Senior Manager you'd have to hire a Director to replace should be promoted. Same thing with a mid-level engineer who produces the output of a senior engineer. You should promote them. By the same token, if you have a really strong senior manager who you could replace with a less skilled senior manager based on the scope and complexity of their role, maybe you should defer the promotion and find a way to expand their responsibilities instead. From the perspective of a person who's advocating for themselves, arguing about what it would take to replace you is more effective than arguing based on what you could command in terms of salary or level on the open market. Anybody who interviews well can parlay their current job into a higher paying job. Better to argue based on what it would take to replace you. In the meantime, when you think about your work day to day, you should also be honest with yourself about what it would take to replace you. If you're a senior engineer and you're always working on tickets a junior engineer could do, you should talk to your manager about finding more impactful work. From a compensation perspective, this heuristic also works well within the bounds of any company's overall compensation strategy, because it avoids comparisons to other companies that may pay more (or less). It's also a helpful reminder that if you don't pay or level people fairly, you very well might wind up hiring their replacement.