LLMs are more miss than hit for me in the coding work I do. Yet I still try to use them as much as possible because they are a helpful accelerator when it works. It’s easy to review the code and either try again with some more prompting or just write it myself.

Currently I start writing myself while Claude is thinking through the result, like we’re both racing.

There’s a lot of hand-wringing about juniors shipping code they don’t understand, but that doesn’t logically make sense for why experienced people are avoiding it. It’s another tool in the toolbox. Use it like one. If your coworkers are shipping code they don’t understand then deal with that problem regardless of where it came from.

Tobi's been really weird lately.

He's been posting straight out conspiracy theories. For example, just yesterday, he re-posted (re-tweeted) this unfounded rumour claiming election interference by unions across the whole Canadian government:

https://x.com/mario4thenorth/status/1908652063808270398

There is no source but "trust me bro" on this - thus it can not be refuted - and it is portrayed as being done in an untraceable fashion. So yeah, it is just spreading distrust.

This is harmful to Canada to be spreading distrust in this fashion, although it is in keeping with the fashion of a lot of posts on X these days.

I feel like all CEO edicts, by nature of the fact that they are a blunt instrument have weird side effects. The general sentiment here is the right one, but it will also lead to strange behavior, goodhart’s law, and maybe in certain corners have the opposite effect intended (I think we’re going to start talking about “AI-think”, a similar effect to “group-think” as teams start to coalesce around an AI’s way of thinking.