sunir

Tomato tomato. It’s sufficiently cybernetic. We can treat it therefore as something with agency.

There is a mind; the model + text + tool inputs is the full entity that can remember, take in sensory information, set objectives, decide, learn. The Observe, Orient, Decide, Act loop.

Maybe. Courts aren’t magic machines that do the right thing.

Consider: you don’t give a warrant to a wiretap subject. That itself is not that big a loophole. And therefore is unlikely to provoke change.

That came later than the beginning. Workhouses came before the loom. You can see this in the progression of quality of things like dinner plates over time.

Making clay pottery can be simple. But to make “fine china” with increasingly sophisticated ornamentation and strength became more complex over time. Now you can go to ikea and buy plates that would be considered expensive luxuries hundreds of years ago.

My chairs resemble each other. Have you tried ikea?

If you are talking about code which isn’t what I said, then we aren’t there yet.

Simplify? It’s like saying a factory made chair building… what?

It’s not simpler. It’s faster and cheaper and more consistent in quality. But way more complex.

I'm building a zork-like dungeon explorer for vibe coded projects. Ok, the zork interface is not that important, but it adds an extra layer of fun, and does reflect the reality of how I dig through a codebase to understand it. You start at the entry point and start exploring each code path to build a map of what is going on, taking notes as you go, and using tools if you're lucky to get a sense of the overall structure. You can also go up and down a level of abstraction like going up and down a dungeon.

It incorporates also complaints from a static analyzer for Python and Javascript that detects 90+ vibe slop anti-patterns using mostly ASTs, and in some cases AST + small language models. The complaints give the local class and methods a sense of how much pain they are in, so I give the code a sense of its own emotional state.

I also build data flow schematics of the entire system so I can visualize the project as a wire diagram, which is very helpful to quickly see what is going on.

Is clunker some new slang that's different than clanker? I'm asking for a friend of my friend Roku.

p.s. thanks for making this; timely as I am playing whackamole with sandboxing right now.

Do you feel like you have no mouth and you must scream?

I don't feel like the abstraction away from assembly language resulted in fewer software engineering jobs. Nor do I feel like Java's virtual machine resulted in fewer systems engineering jobs. Somehow I don't feel that writing in English rather than pure logic will result in fewer engineering problems either. A lot more actually. But at least we'll get the requirements out of users into something concrete faster.

What is definitely going to be abundantly clear is just how much better machines can get at creating correct code and how bad each of us truly is at this. That's an ego hit.

The loving effort an artisan puts into a perfect pot still has wabi sabi from the human error; whereas a factory produced pot is way more perfect and possesses both a Quality from closeness to Idealism and an eerieness from its unnaturalness.

However, the demand for artisan pottery has niched out compared to Ikea bowls, so that's just how it is.

That’s a strong point. I interpret that: Covid created needs that weren’t permanent and also caused overinvestment to address digitalization and remote life needs to support lockdown as well as predictions for a permanent change in the future. Either way a lot of the work to be done in software is done for a while.

Thank you so much.

In my case I have set up the agent is the repo. The repo texts compose the agent’s memory. Changes to the repo require the agent to approve.

Repos also message each other and coordinate plans and changes with each other and make feature requests which the repo agent then manages.

So I keep the agents’ semantically compressed memories as part of the repo as well as the original transcripts because often they lose coherence and reviewing every user submitted prompt realigns the specs and stories and requirements.

The only place MCP made sense for me, so I thought, was to give my Claude Web agents access to the same cli tools my Claude Code agents had. In this context, agents don't have shells.

However, then I discovered MCP servers on Claude Web are forced onto my laptop for Claude Code, which is very confusing. I don't know if there is a way to stop that, but it has messed up my Claude Code agents.

Is this experience common, and is there a known way to stop this?

I can't repair my car, which used to be the hallmark of technical masculine skill in the era of Grease the musical, because mechanical maintenance is not primary in my lifetime. Nor do I have any idea how to manage a farm. I think the kids will be fine. On the other hand, I fear I will not survive once my Internet connection goes out.

Not every career path starts at a software first company. Not every software first company works on the most intense codebase.

And therefore in my experience not every senior engineer would hack it as a senior engineer at a more intense company myself included.

This isn’t a software unique experience. It’s life.

Agreed. We are still in a capital crunch so overhiring is out of fashion. People don’t remember the early 90s or the dot.bust when the same things were said.

Kraft 1977 Programmers and Managers talked about this if I recall. Still the best alternate take on our industry I have ever read.

Why would there be a lack of original ideas? People who are born to code so to speak will do it. Information wants to be free as the saying goes. It only takes one time for an innovation for it to be to copied everywhere.

We don’t need the same volume of developers to have the same or faster speed of innovation.

And conversely if there is stagnation there is a capital opportunity to out compete it and so there will be a human desire to do the work.

Tl;Dr. People like doing stuff and achieving. They will continue to do stuff.

ps it’s too much to claim other people don’t experience creative ideas using AI. You don’t really know that’s true. It hasn’t been my experience as I have had the capability and capacity to complete ideas on my back burner for decades and move onto the next thing.

Until problems are a solved problem, I feel I'm ok.

We lost the Internet to AI. Just accept it. It's bots talking to bots about bots.

After the dot.com, there was the O-pocalypse that terrified me as a recent grad.

- Open source - Outsourcing - Offshoring

It was driving the labour cost of an engineer to zero I felt as a young man.

Then time passed, and I learnt that engineers aren't paid to code. Engineers are paid to solve problems for a business.

If you recall, the dot.com bust and 9/11 crashed finances for a few years. When the money printing gun went whir because "Deficits don't matter" Washington, then engineers were in demand again.

Right now we are in a weird situation where money is being printed and it is also tight. Most of it is going to the hardware and infrastructure layer, like the fiber optic bubble in the dot.com. Software will have its time in the sun again.

Reminds me fondly of advogato.

It’s hardly basic social skill. This is an executive management skill set. That’s the advanced game.

The basic social skill is to avoid conflict and seek acceptance. Go along to get along.

One wouldn’t rewrite the app on one’s on recognizance without peer approval first if this is your vibe.

I find these throws of passionate despondency similar to the 1980s personal computing revolution. Oh dear. Giving mere mortals the power of computing?! How many people would abandon their computers or phones.

It’s not like it changes our industry’s overall flavour.

How many SaaS apps are excel spreadsheets made production grade?

It’s like every engineer forgets that humans have been building a Tower of Babel for 300000 years. And somehow there is always work to do.

People like vibe coding and will do more of it. Then make money fixing the problems the world will still have when you wake up in the morning.

Or drop the price to $20 a year instead of $20 a month and and focus on small software updated infrequently. Software as a service has a dirty secret that it was more service than software. The companies became larded with payroll and most never had great gross margins.

It’s a simple math problem. And it is also Conway’s law that says all software design follows the organization that built it—that is all software design is political.

A framework calls you. You call a library.

A framework constrains the program. A library expands the program.

It’s easier to write a library that is future proofed because it just needs to satisfy its contract.

It’s harder to write a framework because it imposes a contract on everything that depends on it.

Just like it is hard to write tort law without a lot of jurisprudence to build out experience and test cases, it is hard to write a framework from only one use case.

No one likes lawyers because they block you from doing what you want. This is the problem with frameworks.

However the government likes laws because they block you from doing what you want. Same with whomever is directing engineering that wants all other programmers to work in a consistent way.

Thank you. I thought I was going crazy reading the article which doesn’t connect open and close parenthesis :: higher and lower precedence :: indent and outdent :: +1 and -1 and just flip it around to get the opposing polarity.

A real Wesley Crusher moment.

Not necessarily controlling stakes.

You’re right and wrong at the same time. A quantum superposition of validity.

The word thinking is going too much work in your argument, but arguably “assume it’s thinking” is not doing enough work.

The models do compute and can reduce entropy; however, they don’t match the way we presume things do this because we assume every intelligence is human or more accurately the same as our own mind.

To see the algorithm for what it is, you can make it work through a logical set of steps from input to output but it requires multiple passes. The models use a heuristic pattern matching approach to reasoning instead of a computational one like symbolic logic.

While the algorithms are computed, the virtual space the input is transformed to the output is not computational.

The models remain incredible and remarkable but they are incomplete.

Further there is a huge garbage in garbage out problem as often the input to the model lacks enough information to decide on the next transformation to the code base. That’s part of the illusion of conversationality that tricks us into thinking the algorithm is like a human.

AI has always had human reactions like this. Eliza was surprisingly effective, right?

It may be that average humans are not capable of interacting with an AI reliably because the illusion is overwhelming for instinctive reasons.

As engineers we should try to accurately assess and measure what is actually happening so we can predict and reason about how the models fit into systems.