I Prefer Thoughtful Engineering

Sat, 19 Sep 2026 20:12:00 +0100

I am proud to be one of those engineers who enjoyed writing code with the ambient light, a mechanical keyboard & Red Bulls beside me. I love(d) the whole shape of the program moving through my veins. The good ‘ol getting lost in the IDE and losing an evening to it without noticing. A problem would sit there, stubborn and half-formed, and then some connection would appear. You change one thing, run it, see what breaks, learn something, and keep going. Keep doing it until you are tired or have solved the problem.. It is beautiful.

Typing wasn’t necessarily the thing I enjoyed (although the mechanical keyboard does fell nice). I enjoyed the process of creation. There is a particular feeling in understanding a problem well enough to give it a shape in the world. It is extremely hard to beat.

Goes without saying that AI has made the typing part of this work much easier. It can take multiple passes, do repetitive edits, and eventually give me something to react to. I would be lying if I said I don’t miss the typing but deep down I am glad that I don’t have to worry about remembering the exact syntax anymore - I was never good at that. Remembering things is not my strong suite.

But I have thinking about something I came across recently - Shane Parrish shared from Tobi Lutke the idea of a “slop grenade.” It is a catchy name, which is probably why it stuck with me. It describes the work that is generated quickly while the thinking part of the work is outsourced to the reviewer. I feel sad that a lot of really good engineers have fallen into that trap to show how “productive” they are. It really stuck a chord with me and I have been wondering why.

I think the answer might be because I prefer thoughtful engineering. Thoughtful engineering is not really about choosing the right stack or perfect abstractions. It might just be opposite of that and starts much earlier. In my view, engineering starts much before I open claude. Thinking about “what are we actually trying to solve?”, “who will have to live with this decision?”, “why are we solving this problem?”, “is there something I can delete instead?”. (I do have a weird obsession with deleting things). The code or a decision to not write the code follows.

I love how OpenCode does these things. Dax has talked about how their team was moving too fast - shipping features that weren’t important and leaving ugly hacks around for later. I liked that the reaction wasn’t to punt these things in the name of velocity but to be delibrate about making changes and raising the bar.

He also talked about how writing code has become a smaller part of his day while thinking still takes the time it takes. That aligns well with how it feels to me. Code turning up faster has very little to do with the question of whether the thing is worth building or what the real value of doing it is.

I really like that AI gives me more time with the part I care about most. I can let it do the grunt work or explore a few paths while I spend more time thinking deeply about the real problem to solve. I get to decide what should exist and keep worrying at it until I think I have it right.

That is the part I want more of. I am still a fan of the low light and my mechanical keyboard. Red Bull, less so, because I getting older. But I still want the feeling that the code is moving through my veins and thoughtful engineering carries that feeling forward. It gives me more room to create, and more reason to create with care.

Craft on Akshay Katyal | MrDHat

I Prefer Thoughtful Engineering