AI does change the equation. It frees a solo developer to focus more on the big picture. BUT... with the current generation of AI agents, I think you are still right. You still need a second (or ideally more; I'd say 5) developer(s) to get enough perspective to have solid plans and roadmaps.
So, while I currently [mostly] agree, I think the / a next generation of agents might take that over a threshold and make solo development close enough to the equivalent of current 2-dev work to meaningfully change the equation. Furthermore I think that does not even need new models; I think current models with better "harness / tooling / system prompting / skills / etc." (whatever you may call the text files describing important procedures), might be able to fill most of that gap.
Obviously work that needs more than 2 devs planning might take even longer to fully solve with 1 dev + multiple agents, if ever.
My current mental model is that humans can very well think about and walk the boundary of problems, while [current] AI agents can fill the inside to some extent. If a problem has inherent "multi-dimensional boundaries", it might be hard for a person to imagine and walk it well to guide the agents. And I think most of the interesting problems fall in this category.
If you still wonder why there are sudden attacks on Mullvad, I "heard" there are Chinese (in addition to the others; dual- / triple- vendoring is key) LLM-based tools to check for swarm origins and campaigns.
Agreed, mostly. To me the effectiveness of window managers is a bellwether of the control aspect. So, IMO, if you compare them on control or on the quality of window managers, you'll get the same result. Linux has ended up with many window managers (effectively catering to various styles and needs), while macos, for example, has a one-size-fits-all approach with no REAL multiple-desktops (Spaces is a joke and a toy). As a result, I can easily manage 50-100 open terminal windows on my linux box, but on macos 5 is too many (and if I use Spaces to fit a few more, I can't get to the terminal I need in less than a second. In linux, even with 100 xterms, I can). So, a macos product manager might probably ask me why I would want 100 terminal windows, to find me some alternative. In linux no one asks me that question, and hence I can have what I want.
Honestly, at fundamental level, it is less of that need being dependent on third parties, and more of not having at least ONE set of consistent libraries (or one big library). To me the real value add of golang's std lib is not that it is developed by the language authored (that is maybe the third or fourth key point). The first one is the existence of one set (or 1.5, heh, if you consider golang.org/x slightly different) of libraries that are polished and hammered to be consistent together (primarily by spending an enormous effort on simplicity, but also on repeated polishes prior to the first notable public release and some thenafter). This is at the core of what I think rust needs right now to break through and penetrate into real C++ and golang circles.
The C++ std lib is no longer terrible. It is really at a usable level these days. Not fun, but totally bearable. The motivation for C++ has never been the quality of the language or the std lib anyways, so it can happily chug along in many places (including the browser I am typing this on).
I disagree about merging existing "done" libraries into a mega library. That can work to some extent, but that approach will not produce something lasting (in the sense that it will remain without the need for changes for a long time). The way to achieve a lasting mega library is by putting all the pieces you need, and constantly working to increase the consistency between them. Somewhat like turning a long winded rambling into a much denser article.
Going from a set of good and working libraries to a large CONSISTENT library would be substantially laborious. Hence the need for someone with deep pockets to take it on. (There are other ways for that to happen but those are rarer).
C++ (the language) and its extreme complexities (each perhaps added for a good reason at the time, and then held together with backward compatibility as glue) reduced itself to gcc and clang. Not the std lib. The std lib is, de facto, removing some of the complexity.
Recent C++, together with a couple of major libraries, and with a good style guide and a matching lint (that removes / restricts many footguns) is far better than it was, 15-20 years ago. While it is not exactly cool, it can be totally bearable.
My point is the current success can be "copied over" to rust by cloning (or otherwise obtaining) a good base std lib, and a few domain specific comprehensive libraries. That doesn't even need to affect the existing ecosystem. There can be several parallel ecosystems and each one can be relatively thriving (one doing nodejs style YOLO, one doing enterprise extreme best-practices, one doing efficient embedded or OS level work, etc.). However, apart from the nodejs style community, the rest will benefit [heavily] from a base library that is well designed and NOT subject to change. Even a travesty such as a vibe-cloned golang std lib would be an improvement over the current situation.
I believe by a wide margin the most significant long term problem with rust is lack of a good and comprehensive standard library, or a FEW set of crates that provides a really good set. We are in the age of supply chain risks and I personally know many great engineers who shun languages whose community / library providers don't take this problem with utmost care. At this point in time, rust is being treated like nodejs. That is not fair to rust, but that is the reality of the current situation.
Go, on the other hand took simplicity as a major goal, and while not the most efficient or expressive language out there, that single goal is paying massive dividents.
The good news is, all this means, is that rust can / will be a lot more serious when someone, possibly a big corporation with deep pockets to maintain this until a critical mass has formed, develops and maintains a golang-style standard library. It can be like C++'s in which the initial std lib was crap compared to what it is now. But people ended up using and eventually evolving the language and the std lib together. We are missing that kind of library at the moment.
Cloudtop to run builds, g4 commands, etc., and srcfs / srcfsn to actually write code. (caveat: I have never used neovim, so I don't know if that is different).
When using AI, I often find myself preferring either plain text (no markup whatsoever; just manual / text editor formatting of text blocks) or simple html to markdown, depending on the situation. To the point that I rarely see any point in using markdown for anything. If it is meant for to be a simple text mainly for human consumption, the markups often don't add much clarity (and often bring in an amateurish look, as if the author didn't know how to emphasize using English constructs), in which case plain text feels more pprofessional. If it is meant to be [lightly] processed before being presented to a human, or if it is meant to be processed by a tool / bot / LLM, then HTML is infinitely more straightforward.
Also I often call out my colleagues if they try to put a table in markdown. Markdown is not built for tabular data in most professional settings (i.e., one or two table cell could easily take a whole line of markdown to express). A basic <table><tr><td style="background: red">some number</td></tr></table> goes a long way.
I do get pleasure from typing out the code in some languages (and not in others; hello javascript, java!). Similarly, I love writing text with a calligraphy or fountain pen. However, I can't dedicate too of the much work / business time to whatever is more pleasurable.
So, I "doodle" some text / ideas / planning with a calligraphy pen, and type in some code, occasionally, both mainly for the fun aspects. There are side benefits to both, too. Writing some plans slowly and "beautifully" drags them out and I get to think longer on them, so the sporadic "nice looking plans" are often more well thought. And doing the coding all by myself stops my brain from losing the ability. I was initially in the 100% AI-writes-all-code camp for a while and noticed I am getting notably slow in some personal coding skills. It is too early to treat specs as the new code and old languages as assembly (but I admit we might get there some day).
In other words, I think AI doing 90-99% of the coding, depending on the language verbosity and AI accuracy for the code at hand, is quite reasonable.
Of course it can be used. But it is very buggy (as in missing or not well-though-out behaviors), which is unlike the typical polish Apple human interaction folks deliver. For example switching between Spaces and then between apps and windows and creating a new app window don't work as expected in some combination of steps and for some apps. There are several other "corner" cases that show the features were not laid out in a full design to exhaustively decide the desired behavior in each case. Which is very much like when someone bolts on a feature to a system without fully nail down its interaction with all other adjacent and relevant features.
This might have been the case a couple of years ago, but it is certainly not true any more, if you use AI [even occasionally] to manage some of your default.nix and flake.nix files. I learn by getting AI to edit it (default.nix for example), and then study what it did. It helps.
The quality of the managed / packages software, however, is still a bit subpar compared to Debian and Redhat.
Why would you use that feature? MacOS doesn't REALLY have multiple desktops (Spaces). That is merely a pre-release feature (for 10 years or so, I think). As evidenced by the many critical user journey bugs it has that don't get addressed.
I use both linux (with a decent tiling window manager; the tiling management being the least important part of it) and macos. And certain things are just not possible to do with macos. On linux I can have 300+ open terminal windows AND CAN find the one I need when I need to. On macos 20 (counting in Termianl tabs, which are implemented as windows, underneath) is about the high mark that it gets annoying to work on. On macos, you can't effectively work on multiple projects that use the same software (editor + terminal, for example). You can work with different Applications, though, and that is managed pretty well (better than most linux window managers that I have seen).
Every year or so I try adding a couple of Spaces, and always regret it a couple of hours later, switching back to a single Space (+ a few fullscreen apps).
Nix is not the same as nixos, and in this case the distinction matters. It has to step carefully around Apple's updates. This further highlights the fact Apple lacks the same quality package management as some linux distros. Nixpkgs (on macos), Ports, and Homebrew packages are toys compared to the EFFORT that goes into maintaining Debian and Redhat packages.
In terms of package management SOFTWARE, however, nix (and guix, lix, etc.) are state of the art and work fairly similar in both linux and macos. A deeper integration with the OS would have been nice.
The current market share shows how far you can go with just being a better Java.
If (or when? I haven't checked recently) a decent and well-thought-out LLVM backend emerges for it, ideally with some new underlying complexity seeping through, the market share might expand overnight.
And as for C++, while some complexity is certainly unavoidable, a rigorous complexity control is desperately needed. Ideally, the same way Bell Labs folks did when they initially conceived Go from Algol68 and C and similar (before or after joining Google; I couldn't tell), and Rich Hickey did when he initially designed Clojure. Some people are managing the complexity using style guides and clang-tidy checks. Which is great in that doing so doesn't need lengthy language committee decisions. But that approach hasn't been enough to make code _sufficiently_ safe; every now and then an enterprising engineer or team finds a way to abuse a feature in a way that produces unsafe or unpredictable results. Rust is a bit better and solves a few of the common problems, but sadly the list of potential issues (of using Rust in a codebase at scale; Engineers' faults, not Rust's) is long and growing. My verdict is we need both complex and simple LLVM languages, ideally co-designed to have no interop problems by design, while allowing expressing some logic in the simple parts and some logic in the complex parts. Or better, a 3 tier design would be nearly perfect: expressive config language, glue and research language, and core building blocks language. I think a clojure-style language can be designed to achieve all three.
"I Want to Wash My Car. The Car Wash Is 50 Meters Away. Should I Walk or Drive?"
I found it curious to ask the model to provide its reasoning in that turn, without changing the verdict. Here is one response:
To provide the full picture of the initial reasoning, we have to address the "geospatial" logic of your request. When you asked if you should walk or drive to a car wash 50 meters away, the reasoning relied on a fundamental distinction: The difference between the "Car Wash" (the business) and "Washing the Car" (the action).
Here is the expanded breakdown of that logic:
1. The Location Assumption
My reasoning assumed that the car is currently at your home and the professional car wash facility is 50 meters away.
If you drive: You are moving the "object to be cleaned" to the "cleaning tools."
If you walk: You are moving the "cleaner" (you) to the "cleaning tools" to prepare, or you are choosing to bring the "cleaning tools" to the car.
2. The "Walk to Wash" Strategy (Home-Based)
The most significant expansion of the "Walk" recommendation is the realization that 50 meters is well within the range of a standard garden hose. Most heavy-duty garden hoses are 15, 25, or 30 meters long. By connecting two hoses, you can reach the car without ever starting the engine. Walking to the car wash facility to buy supplies (or simply realizing how close it is) highlights that you don't actually need to move the car to achieve the goal.
Zero-Motion Cleaning: By walking back and forth with your buckets or hose, the car stays in its spot. You avoid "flash rusting" your brake rotors, which happens almost instantly when you wash a car and then let it sit after a short drive.
3. The "Scout and Prep" Logic
If you must use the equipment at the facility (like a high-pressure wand or an automated bay), walking there first is the superior "Logistics" move for the following reasons:
Queue Management: 50 meters is close enough to see the line, but walking allows you to talk to the attendant or check the bay's status. If you drive and there is a 20-minute wait, you are stuck idling your engine (wasting gas and creating heat) or turning it off and on (draining the battery). Walking allows you to timed your "driving" moment perfectly.
The "Dry Interior" Principle: If you walk to the wash to buy your tokens, prep the vacuum, or set up your towels, you are doing the "dirty work" while the car is safely tucked away. Once you finally drive the 50 meters, you do the wash and drive back. This minimizes the time the car spends in a high-moisture, high-chemical environment.
4. The Potential for "Mobile" Service
By walking 50 meters to the car wash, you might find that they offer mobile detailing or have "loaner" equipment. In a high-density area, a car wash 50 meters away is practically your neighbor. Walking over to establish a rapport can often lead to a "valet" situation where they come to you, which is the ultimate way to get the car washed without ever sitting in the driver's seat.
----
also noticed a typo and asked it about that:
> what does "timed" in `Walking allows you to timed your driving" moment perfectly.` mean?
That is a typo on my part—it should have been "time" (the verb).
That won't change until top-notch, simple (i.e., without 100 transient dependencies) org parsing and formatting libraries for a few key languages (go, javascript / typescript, and python; maybe also C++ and java) become available.
Which is sad because org-mode seems far more versatile than markdown, except for a couple of ergonomic features (e.g., ``` vs #+BEGIN_SRC, and * [x] task vs * DONE task).
Even libraries to parse and format a subset of org-mode would be good enablers.
I went to school in a poor country, and live in the US. The education budget was very low when / where I grew up, and it is pretty hefty where my kids go to school. I occasionally visit their school and volunteer to help. That has given me a good frame for comparison.
The quality of education my kids are getting is pure trash compared to what I receieved.
The problem is not the budget. It is the lack of real teachers, as well as a perpetually experimental curriculum. The "modern" methods that I have seen their teachers practice (which confuse the teachers, too, by the way; the teachers all have said that), are very visibly wrong. So wrong that even I can see all sorts of flaws, despite not having any background in education science. The curriculum is predictably set for failure.
I strongly believe technology, and AI in particular, can be a major enabler in improving education. However, for early education (first 5-6 grades), I think absolute lack of technology (except maybe a big e-ink class whiteboard, or some such) would be far more beneficial. Kids can learn to type very quickly when needed (ideally 6th / 7th grade). They can't learn thinking-while-writing, as quickly. They have to slowly build up that mental muscle. Let them have a few years of building structure and core understanding, then get exposed to tools for doing things faster.
So, while I currently [mostly] agree, I think the / a next generation of agents might take that over a threshold and make solo development close enough to the equivalent of current 2-dev work to meaningfully change the equation. Furthermore I think that does not even need new models; I think current models with better "harness / tooling / system prompting / skills / etc." (whatever you may call the text files describing important procedures), might be able to fill most of that gap.
Obviously work that needs more than 2 devs planning might take even longer to fully solve with 1 dev + multiple agents, if ever.
My current mental model is that humans can very well think about and walk the boundary of problems, while [current] AI agents can fill the inside to some extent. If a problem has inherent "multi-dimensional boundaries", it might be hard for a person to imagine and walk it well to guide the agents. And I think most of the interesting problems fall in this category.