Prompting delimiters and JSON style prompting

I had a chat with Gemini (and ChatGPT today) about the topic of delimiters and json style prompting. I’m not a fan of <prompt>putting stuff between</prompt> xml tags. It’s just too much typing for me. And since LLMs are supposed to be all about natural language interactions, it feels like it’s defeating the purpose to have to dress up a prompt like that.

So I asked Gemini and ChatGPT about it. Gemini’s response was a little more thorough, so here is that chat.

Continue reading “Prompting delimiters and JSON style prompting”

More on journal injuries

The simplest way to think of “journal injuries” is to think of them as injuries you give yourself through journaling about unmet expectations while ignoring your actual accomplishments.

You sit down to work and you journal about the process of working not the work itself. If you journaled about the work, you would see clearly that you’ve accomplished more than you’ll give yourself credit for later when you look back at all the expectations you set but didn’t meet.

Challenge yourself to write 3 hours a day every day but fail after the first 2 days? You’ll see that in the journal entries. In loving detail, you’ll discuss the things that got in your way, or how you’ll stick it out this time, for sure, or how you took a break for coffee and how in the world did that take thirty minutes…? You won’t see that you finished two chapters, made a cover, set up a page on your website for your upcoming book… Do you see where that’s going?

Journal injuries, at least mine, have come from a mindset where I value the process more than the work. If I spent my time journaling about the work instead of the process, I’d have a lovely record of all my accomplishments. Instead, I have a treasure trove of journal injuries waiting to get claws into me when I revisit them.

So, any future logs, or journals, have to come from me with the mindset of journal the work, not the process.

It’s worth repeating. Journal the work, not the process.