Week #67#

It was a slow and steady week. New habits are emerging. It was a pleasant experience, I felt good, I kept myself out of the social media and content spiral doom-scrolling. Why? Because something changed the perspective of action and effect, the piece I was missing was intention.

Action, Intention, and Effect#

I heard or read somewhere, that we feel anxious or we waste our time, not because of motivation, rather due to the lack of intentionality of the actions. If you are going to the web with a clear focus and a objective, no force or urge can stop you. Motivation helps but doesn’t carry you all the way. Its not one emotion or quality thing, rather all-hands together approach. You solely can’t rely on focus either, you need your inner fire, you need purpose, you need energy, you need the skills to do it. Nothing comes on its own with a single cause-effect, its a multi-dimensional world of action and effect.

I allocated time, to write, to read. And I did complete it, not fully, not perfect, but it got me moving, I knew A and D, it helped me get to B and understand that C exists. Sometimes, you don’t know the actual goal, you assume it is something, but it reveals as you go and take action. I wrote around 9k words on my project in 7 days. I am a satisfied person. What happens if NaNoWriMo isn’t there this year, I am not after writing 30k words in 30 days, I am on mission to complete the project. Maybe I know the goal, but the actual goal is appearing as I write my way through.

I am a software developer, trying to be a writer too. Its hard, but there is no harm in giving a shot, you never know what bug it might reveal, or it could be a feature not a bug!

Quote of the week#

“What comes easy won’t last, and what lasts won’t come easy.”

– Ntsiki KaCaleniA perfect quote for the perfect time. AI generated slop. It comes easy. I don’t think it will last. By last we can relate to getting the love, getting the reward, or feeling accomplished. None of it is easy to get. If it was easy, it won’t be worth looking for, right?

Some things in life come with struggle, and that for a reason. If we really want to appreciate somethings, we really need to understand its value, its worthiness. We can’t get something easily and be satisfied forever with it. AI generated anything is not lasting long. Code, Images, text, nothing. I delete the code that I accepted last week this Monday. If not, eventually next week when something breaks. Because I didn’t toil hard for producing it, that’s the thing, if it has become easier to produce art or code, does it mean art is not worth it? No, AI-generated art is not worth it, human generated, or natural art is what drives me here to write this exact post, and has carried humanity thus far in the world.

Read#

Learning loops and LLMs- This is a true analogy. This is highlighting about the world relating software development with a assembly line.

In software, design emerges through implementation. We often need to write code before we can even understand the right design. The feedback from code is our primary guide. Much of this cannot be done in isolation.

  • I like this mindset. I agree to this 100%. Building software is about coming to a design, not just producing it. The why, the what are the questions the software development solves. How is the question that the code solves, its just a means to an end. I have learnt it the hard way.

  • There are no shortcuts to learning. This is another one that sticks to me. You can produce code, but that’s not the only responsibility of a developer. The code needs to solve something that you know, not LLM. You need to architect it based on the problem we are trying to solve.

  • Using LLMs to increase the things that we can experiment with is a good tool and direction. But then you need to plan, think, fail and iterate. It cannot be just prompt and done. You need to sit and read the things it generates, think about them, make changes, remove and write it yourself if needed. It can write code, but cannot build software solution. Yet!

You should write an agent- A good one. You need to experience the thing in order to have an opinion. That’s quite obvious. But people coming on conclusions about AI and LLMs don’t quite think that way. They do one thing and generalize their opinion. Generalization of experiences is not good.

  • I have used Cursor, cursor-agent cli, gemini cli, and amp cli free version. I can have opinions about them. But I cannot about Claude Code, Codex, and the bazillions of those CLI agentic models.

Turns out: context engineering is a straightforwardly legible programming problem. You’re allotted a fixed number of tokens in any context window. Each input you feed in, each output you save, each tool you describe, and each tool output eats tokens.

  • This is well put. I agree to that. People and marketing blogs really buzz about context window, but its just a list of strings, it becomes quite obvious why it can bloat pretty quickly if you keep on adding more functions (tools, its description, the parameters, the docstrings, and what not for each call).

  • It also simplified the buzzword of context engineering, its just what effectively you can put without repeating and keeping in only precisely the ingredients for producing the meal.

Is it really good enough- This is a banger post. It has so many relatable things and points to discuss. But the below quote hits it.

Most software is garbage. AI-generated garbage isn’t notably worse.

  • A precise description of AI-generated garbage, its spitted like anything. I can add one more quote to this, which would be the quote of the week for this week. If a code can be produced easily, its not worth it, the thing that takes time is the worth.

Lessons from vibe coding- I would have loved this post, but the AI-generated image just threw me off a bad impression. But it was chess, how can I resist this.

  • I read this, and felt good. A human admitting his mistake. Going all in AI and failing hard, then going again but from the lessons learned from the fall.

AI coding tools are force multipliers for developers who know what they’re doing. They’re force randomizers for those who don’t.

  • A banger of a line. If you know what you are doing, then you already have done your part, that is to think.

  • And this one too.

Thinking AI eliminates complexity is like thinking cars eliminate the need to know where you’re going. You’ll move fast, but it’ll probably be in the wrong direction

  • Speed is useless without direction. We all can agree to that. Having a dumb engineer with claude code is -10x efficient then having gemini cli to a real soy dev.

AI is Dunning-Kruger as a Service- Agreed. This is the opposite of imposter syndrome. It hits like ego. You think, that you have done the job without thinking. We are at such a conjecture that we are trying to feel like done something without doing it. AI-art for instance, if you put it on your thumbnail or post it, what are you really doing, pretending that you made it? Having the feel of creating something without creating it?

  • I stopped creating ai generated thumbnails on this newsletter after a couple of attempts as it felt boring. I didn’t get anything from that. No satisfaction nor skill.

Mr. TIFF- It was just a pleasant read. No controversy, no fluff. Just two humble and honest humans doing their job. This world would be so beautiful if each human did its job correctly, just correctly, not exceeding the expectation or performing a all-nighter. Just completing his or her job correctly is so under-rated.

Watched#

I haven’t watched anything this week that is worth sharing and technical enough. I have replaced my youtube watching time with 1 hour writing sprints. For the past whole week, I wrote around 9k words. I had a goal and I am moving towards it. I don’t know if youtube is worth watching anymore. I’ll enjoy other things while I can.

Learnt#

Flask has a after_request decorator/hook, that we can use to run at the end of a request.- I added it to log a request, created a state and appended/put to the state after a even happening at any part of the API. Then using that state, I dumped it in a log, that gave a reliable and a singular data point to gather and analysis data. A good pattern to remember and learn from.

Zellar’s congruence algorithm to calculate the weekday from a given date- This was a problem of the day in the freecodecamp daily challenge. I love it, it helped me learn this algorithm. Its so trivial to implement yet feels so useful to get weekday of any given date. Leave the timezone mess aside for a moment.

  • It as simple as this

`year, month, day = 2025, 11, 05 if month < 3: month = month + 12 year = year - 1 k = year % 100 j = year // 100 day_digit = (day + (13 * (month+1)) // 5 + k + k // 4 + j // 4 + 5 * j) % 7

0 is Saturday, 1 is Sunday and so on.`## Tech News#

Moonshot.ai releases Kimi K2 Thinking modelFor more news, follow the Hackernewsletter (#770th edition), and for software development/coding articles, join daily.dev.It was a fun start of the penultimate month of 2025. Hopefully, the rest of the month go just like this and I will have most of my goals of the year completed.