I have some coworkers that are similar in everything--education, work ethic, and intelligence--but some of the tick out ML ideas that work like clockwork, while others get hits rarely if ever. I cannot tell what makes it work for some and not others. Their ideas both sound equally good.
Sometimes a coworker will be an ML star for a year or two, but then suddenly run out of steam. It's brutal to watch.
I used to think most smart people had similar distributions of good ideas, and it was just that the hardest working tried out all 50 of their ideas to pick out the 2 good ones. But I've seen smart and hardworking people have a hit rate of 0.
C++ was a superb language for its time. There was nothing faster with as-powerful abstractions. It showed how far you can change a language too, with C++11 being a massively better language with shared_ptr and company.
It took in almost every idea, and the battlefield showed us which do work and which don't. We get to keep RAII, move vs copy, smart pointers, placement-new, and generics. We get to drop auto_ptr, copy-by-default, its specific exceptions implementation (fight me), multiple virtual inheritance, and templates as full code substitution.
In my opinion the battles have played out, and Rust is the best sum-up of what worked (it even inherited the compile times! Lucky us!)
AI code is often fine. The point of the code is for the computer to do a job. And code is supposed to be consistent and boring and straightforward in its structure. AI is pretty ok at this.
AI is terrible at writing. Its prose is awful and horrendous to read. Plus, the whole point of reading a written piece is to hear about the author's new ideas or experiences. If AI wrote your piece it's both bad and pointless. I can also prompt the AI to hear what it knows about.
I haven't tried opus 4.8 yet, but I hope the writing quality has returned to the Opus 4.5 level. Anthropic really lost something, where 4.5 had this really crisp writing style that flowed really nicely and 4.6 and 4.7 sound much more "chatgpt-like." It feels like they tuned it to be too much of a problem solver, and when you do that you get this terse, clipped textual output that's more difficult to read.
The major axis is urban+gritty to more suburby and spread out. It's a very personal preference where you want to be, but most people dislike the most gritty areas (tenderloin, most of soma). It's worth aiming for a neighborhood at the median as your first.
The best managers I've seen would turn this situation into a headcount request.
The problem is leadership has priorities 1-5. Your team works on 1-3, but the PM keeps getting hassled about 4 and 5, so they look for levers to get them to happen.
In this situation, the PM scrounged up headcount from elsewhere, but if you present the option of adding headcount to the existing team, then you create a more harmonious option of getting these lower priorities accomplished.
Of course, this guy was taken fully by surprise by the suggestion. It's much harder to present a better option after the fact, and I agree that letting leadership feel the consequences of its decisions is a reasonable thing to do in this case.
Is there something mass-produced that's flexible and consistent like gridfinity? As a non 3d printer owner, I've been looking for something I can just buy that would work.
Maybe it's time to do pair agentic engineering? Have two engineers at the screen, writing the prompts together, and deciding how to verify the results.
In MCP setups you do give the agent the full description of what the tool can do, but I don't see why you couldn't do the same for executables. Something like injecting `tool_exe --agent-usage` into the prompt at startup.
Great article otherwise. I've been wondering why people are so zealous about MCP vs executable tools, and it looks like it's just tradeoffs between implementation differences to me.