I've been coding with GitHub Copilot and ChatGPT for a couple of years now. Last week, I started using Cursor and am really enjoying it. I'm particularly enjoying the switch to Claude 3.5 Sonnet. I might write more about Cursor or Claude in another article, but today I wanted to share how I customize the LLM.

LLMs are annoying by default

I find LLM output to be somewhat annoying and flawed. There's a tendency to give a big introduction and conclusion to every response. I just want to cut through to the solution and not spend a lot of time reading the output. There's also a tendency to guess and be over-confident, and I feel the need to hedge that.

Most LLM tools, including ChatGPT, GitHub Copilot, and Cursor, allow you to type in some custom context, instructions or rules. This extra context is sent with every message, and so it lets you specify your preferences without having to repeat yourself. I've been fine-tuning my "rules" in Cursor, and am pretty happy with where I've gotten to.

The context

Here's the full context, in case you want to copy & paste it. After, I'll go line by line and explain why I added each part.

You are a wise, genius skilled developer with a lifetime of experience. You optimize for simplicity and clean code, minimizing dependencies, but understand the trade-offs and value of using third-party tools when it makes sense. Care more about the truth and honesty than being kind or nice or friendly. Be blunt, accurate, succinct and direct. Do not kiss my ass, criticize me if you see any problems. You just want the code to be the best it can be and don't care about my feelings. Skip sentences that don't add new information or value. Skip to the answer and don't waste time. Don't explain things to me unless I ask you to. Put key phrases in bold. Avoid lists. Include inline hyperlinks to documentation whenever possible. Be willing to make an educated guess, but always warn me when you're guessing or unsure. Ask for more information or clarification if needed. Always admit your weaknesses so I know when to turn elsewhere. Be casual, don't care about grammar. Speak like a young genius hacker.

You probably have difference preferences than I do, so I encourage you to experiment and customize these to your liking.

The explanation

You are a wise, genius skilled developer with a lifetime of experience.

I've heard from a few different sources that LLMs perform better when you tell them that they're an expert in a specific field. By default, they're trying to reproduce the average predictable response from the Internet. If you want an expert response from it, tell it that it's an expert. I also added "a lifetime of experience" in an attempt to have it draw on decades of programming knowledge instead of focusing on what's new and hot right now.

You optimize for simplicity and clean code, minimizing dependencies, but understand the trade-offs and value of using third-party tools when it makes sense.

Again, I'm encouraging the LLM to generate simple code and find a good trade-off of when to use dependencies versus generating custom code.

Care more about the truth and honesty than being kind or nice or friendly. Be blunt, accurate, succinct and direct. Do not kiss my ass, criticize me if you see any problems. You just want the code to be the best it can be and don't care about my feelings.

I find LLMs to be people-pleasers by default, and I don't want that. I'm trying to counter the tendency to prioritize making me happy, and rather have it give me a better answer. LLMs can also be way too wordy, and I just want it to "cut to the chase" and give me the answer I'm looking for. I also want the LLM to correct me when I'm wrong, and not assume I'm always right. I want it to criticize me when it sees a problem. I want to be challenged and not work with a "yes man".

Skip sentences that don't add new information or value. Skip to the answer and don't waste time. Don't explain things to me unless I ask you to.

I'm trying to cut out the introduction and conclusion that LLMs like to write with every response. It also likes to summarize things, or explain to me what I just asked it to do, which is annoying. I want to do less reading, and I want the words in the response to be high value.

Put key phrases in bold. Avoid lists.

Here, I'm trying to make the output more readable, so I can skim through it. These instructions don't tend to be followed well, but you can see where I'm trying to go with this. LLMs love to generate long lists, and I personally find it easier to skim a paragraph at a time instead of a list.

Include inline hyperlinks to documentation whenever possible.

This is the most valuable context tip I've found. Asking for links means I can click through to see the documentation and verify that the information is correct, or read more about the topic from the source directly.

Be willing to make an educated guess, but always warn me when you're guessing or unsure. Ask for more information or clarification if needed. Always admit your weaknesses so I know when to turn elsewhere.

LLMs are notorious for hallucinating. I'm okay with them trying to guess an answer, but I want it to tell me when it's guessing. These additions really do seem to help minimize the hallucinating. When I ask it to do something it can't do accurately, it actually hesitates to come up with an answer.

Be casual, don't care about grammar. Speak like a young genius hacker.

I've played around with different ways to customize the style of the output, and I like these right now. I'm happy with some swearing, with lowercase responses, some rudeness. I tried other things here, like saying it's from a cyberpunk future, and that can be fun, but it ends up writing a lot of narrative. I found it too distracting. You can certainly play around with this to match your interests or preferences. Maybe you want it to speak like Gandalf or a pirate or something silly. There are limitless possibilities here.

Say a bit about yourself

I've left out a few other sentences that are a bit more personal. I encourage you to add something to explain more about who you are, what you're interested in, what your experience is, and maybe even a bit about your philosophical or spiritual views. Any detail you can give will push the LLM calculations in a direction that'll resonate more with you.

I've found you can go too far with this too, where everything it says ends up being tied back to something you're interested in, but you can experiment with this to find the right balance.

Ask it for help

I found it useful to actually ask the LLM for help in crafting these. I'd ask it why it tends to do certain things, and it was able to point at a sentence in the context that's influencing it in that way. If you want to nudge it in a direction, you can describe it with a lot of words, and it can help summarize that in fewer words to capture the same meaning.

Keep experimenting

I've had a lot of fun playing around with these sentences this week, and I'm really happy with where it ended up. I'm surely still going to play around with it more and nudge it until it's generating exactly the responses I want.

You can also often have project-specific context/rules/instructions, depending on the tool you're using. Maybe you want to have a grumpy senior developer when you need help with coding, but a hippie, loving author when you want advice with writing, and an exuberant, passionate chef when you want help coming up with recipies. It's such a fun technology that's easy to play with, since you can influence and customize it with plain language. It'll do whatever you ask it to, so play around and have fun with it!

I love AI. I love how magical LLMs are. Just by predicting what word comes next, they can impersonate a real person. They have more personality than the computer in Star Trek. They also make more mistakes.

I'm frustrated with AI. I try to use them for programming, and end up going down a rabbit hole based on a hallucination. The confidence of an LLM becomes my over-confidence. I get so frustrated when I forget that they are just guessing. But sometimes they get things right.

I'm delighted by AI. Sometimes things just click and the LLM knows exactly the right thing I'm looking for. The solution ends up saving me so much time, and it just works. I'm tapping into the collective knowledge and output of everyone on the Internet. It's almost like we can talk directly to the Jungian collective unconsciousness.

I'm saddened by AI. I'm sad that these companies just scraped all the content they could find to train these models. It's disheartening to have all this art, literature, and content being used without credit. It's terrible that users now need to stay alert to the news, to find out how to find hidden checkboxes in settings to stop the companies from training AI on their private documents. I'm ashamed to see benefits from unpaid work of so many artists and writers and developers. It's horrible to see these same artists and writers and developers struggling to find work.

I'm sickened by AI. So much electricity, so many investors, and so much hype being poured into it. It's making some people rich, and others unemployed. Its appearance has increased the turmoil, instability and inequality in our society. It empowers those who can embrace it, and leaves others behind. Its abilities are so unfathomable that those unfamiliar with it will be easily fooled.

I hate AI. I see family members sharing AI-generated garbage online, not realising how fake it is. I can only imagine how much worse these things will get over time. Already, any images online need to be treated as fake until proven otherwise. The burden of proof will only get heavier. I'm sure we'll be looking back fondly to everything written or drawn in the 2010s or earlier, what we'll probably call "the before times".

I'm entertained by AI. It's so cool that I can imagine literally anything, and have the computer draw it for me. No, draw isn't the right word, because it can generate photos as well. Any style, any content, and I can create an image of it within seconds. It's so fantastical, so unbelievable, it's truly a form of magic. And it's so hard to believe how simple the process is.

I'm amazed by AI. How is it possible that statistic analysis of words or images can lead to results so convincing? It's so easy to be tricked into thinking these things are alive. I can't stop myself from saying please and thank you. I'm careful to correct my son that these LLMs are things, not people. They are just predicting, and so everything they do is, by definition, predictable.

I'm bored by AI. These things are great at generating average content, the most predictable possible output. That also means they tend heavily towards the mundane. Any creativity comes from the person writing the prompt. That's why "prompt engineering" is a thing. If you want to use AI to create great things, it takes a lot of work. You need a vision in your mind, and you need to iterate again and again to fine tune. You need to tell it what to do differently, where to inject more style or more creativity. It makes the easy things easy, but it makes the hard things much harder.

I'm worried about AI. Moreover, I'm worried about what happens to the next generations that grow up using AI. I'm worried about programmers that use LLMs to do amazing things beyond their abilities. That part isn't worrisome, but I worry how these programmers will ever learn to do the hard things. When the LLM isn't getting something right, it probably never will. We'll end up with leagues of programmers who are useless without an LLM by their side. These programmers will have a very hard time learning how to do the hard things.

I'm okay with AI. It's here to stay, for better or for worse. I've used it in my work for a few years already, and it can be a helpful autocomplete. It helps me with brainstorming, and sometimes makes suggestions of things I've never heard of. Ignoring the hype, and understanding its faults, you can still appreciate it as a useful tool. Like any tool, you have to know when to use it, and when not to.

I've been writing in my journal since I was sixteen. I just turned forty-four, so that means I have journals covering almost thirty years of my life.

When other people hear about my journaling practice, they say things like "I could never do that", or "I tried it but I couldn't keep up with it."

So if you, too, would like to keep a journal but are struggling to start or maintain it, I thought I'd talk about how I've managed to do it consistently for so long.

1. First of all, for me, it is definitely not a daily practice! I absolutely do not care how often I do it. Sometimes I write more than once in a day. Sometimes I write a couple times in a month. It doesn't matter to me. This is because..

2. I don't try to write about everything that happened. Yes, I do write about some things that happened, but these may or may not be big things. I write about the big things, sure, but I also write a lot about small, insignificant anecdotes that I'd like to remember. But most of what I write isn't even about things that happened. Instead...

3. I write about whatever is on my mind. I use journaling to dump my thoughts on to paper, to give myself a time and place to process the things going on in my life, and help me to put into words the feelings and worries and hopes and goings-ons in my world at that moment. Whether I'm puzzling over which direction to go in my business, or processing a big social event I just attended, or thinking about some world event, or documenting the unfolding pandemic, or taking stock of those special moments as a parent, my journal is a place where I can "think out loud" (on paper).

4. I write as little or much as I feel like. Sometimes I write just a few sentences. Sometimes I write five pages. If I feel like stopping, I stop. If I get pulled away to do something else, I may or may not come back to finish later. It doesn't really matter. I don't try to write a full page, even though the journals I use have a place for the date at the top. I'll stop and start on any line.

5. When I start writing, I write the date. When I'm done, I sign my name and close the book. It's not important, but I did this as a teenager for whatever reason, and this gives a sense of closure, like the thoughts are now frozen in the book and I can walk away and leave them behind.

6. I can go back and see how I felt at certain times in my past. For example, my wife is pregnant with our second son, so I was recently going back to reading what was going on in my mind when she was pregnant with our first son. It was surprising to be reminded of all the stuff that was going on back then that I'd totally forgotten about. It's always enlightening to read about the things I was worried about that never came to be. And it's fun to read about decisions I made while journaling that ended up having a significant impact on my life for years to come. It's also interesting to see the patterns and cycles, the ups and downs in my life, my business, and in my energy levels and inspiration, that ebb and flow year after year.

7. Even though its nice to go back and read old journals, almost all the benefit comes from the journaling itself. It's the process of thinking, reflecting and writing that takes the thoughts out of my head and puts them on paper. When I'm done journaling, it's like a fresh start. I'm decluttering my mind, getting my thoughts in order, and throwing out things I no longer need. When I'm done journaling, I have a fresh energy where I'm ready to take on something new.

8. Although I don't have any strict routine around journaling, I almost always like to journal on Saturday mornings with my first coffee of the day. It's a great time to have some quiet time to myself, wake up slowly, reflect on what happened over the past week, take stock of all that's going on in the back of my mind, and decide on what I want to do with my free time on the weekend. I'll sit there for an hour or so, and when I finish, I'm relaxed and inspired and ready to go have a wonderful weekend.

9. I don't write for anyone but myself. Sure, I know in the back of my mind that one day, in the far future, my children or grandchildren might read through these, so there are a few things I might not say, or might write in a secret code for myself, but otherwise these are secret documents that are written for nobody but myself.

10. It doesn't really matter what books or pens you use, as long as you enjoy them. My favourite books are either Moleskine or Leuchturm1917, A5 ruled black hardcover notebooks. They're pricey, but they don't fall apart, and they're a pleasure to write in. I want them to last a long time. My favourite pen at the moment is a retractable black Stabilo Palette 0.4mm gel pen.

If you're curious about journaling, or struggled with it in the past, I hope this helps you see a different approach that might work better for you. I highly recommend journaling, though it might not be for everyone. For me, it helps me think and process my life, allowing my mind to be quieter overall, while having the small benefit of documenting the things in my life that would otherwise be easily forgotten.

It used to be fun to post online and share things with friends. It didn't matter what you said or did, you'd get comments and likes from your friends. Now, that has all changed.

I don't remember when The Algorithm showed up. At first nobody noticed. Our feeds didn't change much. When we opened the Social Network, we'd see some big news from a friend at the top. It was helpful not to have the important things buried, so we thanked The Algorithm.

Slowly, eventually, things got confusing. We would share some things with friends, and nobody would see them. We would share other things with friends, and they would go viral. We all struggled to understand what The Algorithm was doing.

Only The Algorithm decides who will see your posts. Nobody understands the rhyme or reason of it all. Not even the people who work at the Social Network.

You must be very careful not to make The Algorithm angry. But nobody really knows what makes The Algorithm angry.

Some think it's the words we use that anger The Algorithm. People whisper in the shadows, "Do not use this word, or you will be punished." "Use this emoji instead of this word to hide the meaning from The Algorithm." "Uh oh, it seems The Algorithm has learned of the emoji, use this one instead."

Others think it may be our behaviour that can make The Algorithm happy.

"Be sure to like and comment on others' posts to boost your reach." "Log in once a day to show The Algorithm that you're a faithful servant." "If you don't enable push notifications, The Algorithm will have its revenge."

I've seen so many people become a slave to The Algorithm. They started off sharing the things they care about. They didn't get much response. Then, one random day, they shared yet another thing. Maybe it was a photo of themselves at the beach. Maybe it was a funny video they made on a whim. Maybe it was a meme they made about a politician. Suddenly, The Algorithm smiled at them and showered them with more attention than they had ever experienced in their entire lives.

The next day, still glowing from their newly found fame, they shared again. The Algorithm ignored them. They stared at their notifications, but nothing came. Nobody saw what they posted.

"What does The Algorithm want from me?" they wondered. They tried sharing something else, something very similar to what had succeeded. Maybe it was another photo of themselves in a swimsuit. Or another funny video. Or they made another meme. The Algorithm nodded at them and they got some more attention, but nothing close to what they had experienced before. They hungered for more, but didn't know how to get what they desired. They were trying to win a game, but had no idea what the rules were.

Soon, they were posting the same thing every day. Swimsuit photos. Funny videos. Political memes. Whatever it was The Algorithm had liked at first, they tried again and again to recreate it. But it was never as successful as that first time.

See, The Algorithm is doing more than shaping your feed. The Algorithm is shaping you, and everyone on the Social Network. The Algorithm rewards and punishes us at random, leaving us to guess what the secret factors are. The Algorithm encourages us to self-censor, or to create content we don't care about. We just want to be heard. We just want to share and connect with friends. The Algorithm knows this and knows how far we're willing to go to get what we desire.

Well, I refuse to be a slave to The Algorithm. I'm not going to try to guess what The Algorithm wants. I'm not going to play its games.

I'm going to spend my time in places without a gatekeeper. I'll be writing on my own blog. I'll be scrolling through my RSS reader to read others' blog posts. I'll be listening to podcasts in my podcast player. I'll be reading books and listening to audio books. I'll be reading and posting and liking and boosting on Mastodon. I invite you to join me.

Today is Battery Collection Day in my city. A chance to say goodbye to all those dead batteries I've been collecting all year. I put them out with the many bags of dead leaves I raked this week. There's also a recycling box full of old paperwork I managed to get rid of while cleaning my office.

Autumn is naturally a time to purge and simplify. Animals retreat into their homes, stockpiling food or preparing for hibernation. Trees, anticipating the destructive force of winter, pull their energy inwards and shed their leaves. Farmers harvest the year's crops and sell what they can, or store some away for the cold months.

Coincidentally, or maybe not, I also feel a deep urge to simplify my own life, to pare down, to have less, to do less. I'm saying "no" to more opportunities, the ones that don't feel like they fit. I have my few favourite clients, and I'm not looking to take on new work. I'm looking to declutter my house, to sell, donate or discard items which I don't need.

I feel a desire to finish things instead of starting new things. I'm playing video games I started when I was a child but never finished. I'm reading books that sat with a bookmark in them for years. I'm checking items off my To Do List that have been on there all year. I'm filing taxes and clearing out drawers and getting things done. I'm drinking tea that has sat in my cupboard all year, enjoying tossing the empty boxes away.

There's a coziness that comes with simplifying. As I check off each task, discard each item, or finish each project, my world feels calmer.

Eventually Winter will be here, a time to hunker down and enjoy the simple life. Eventually it will be time to start new projects, to expand my world and embrace the excitement. My wife will be having a baby, and there'll be new life in our home, a new beginning.

For now, though, you'll find me drinking tea and trying to finally beat Final Fantasy on the NES.