I've reached a point in my career where i can essentially call the shots as far as engineering and even product direction is concerned.

But recently had to do some consulting as an IC due to circumstances (worst case scenario deal closing delay), and i can tell you that there are places where non-technical people make tech decisions without consulting anyone. Also in my already quite distant qconsultant past (wow, time flies) while it has always been common for me to sort out requirements, and help with business analysis to align them with engineering, even when you dismiss flying unicorns, the class of core requirement evolutions that can make you regret choosing a tech stack is still very large. I've seen a couple of folks fired literally due to this.

Even now that i call the shots and sell products rather than time, the roadmap uncertainty is still high (largely driven by the need to make products user-friendly enough based on feedback), so i default to .NET + angular as it keeps my game open.

B2B2C at the cross-roads of ecommerce, fintech, blockchains. So you need to deal with vendor integration requirements while making the integration user-friendly from a UX perspective + and keeping your own core UX premium, especially for the B2C segment, and being competitive. Requirements change fast, especially at the beginning, so you don't want your engineering to constrain you or make things more difficult than necessary.

> Different tools for different jobs. (Me? I honestly love React, just a more minimal stack. Zustand and TanStack and Vite. No Redux or Next.JS)

My concern with is that i build line of business applications for a living, and literally, every project i work on ends up growing in scope such that even the simplest ones, reach a point where the customer asks for items that are just much simpler to do with a front-end framework. And these are often cross cutting concerns, such that it is not as easy as simply adding some js to one page.

Example: "i want to be able to reorder all the columns of all the tables of the website, and hide some columns, with my preferences remembered for each table". Customers pull that kind of thing off their hat every time, such that even "basic websites" can move out of the trivial territory any time. And this is just the most trivial example i can think of.

I want the tech stacks i use to be future proof, as in not having to rewrite everything when requirements become more complex (as it is also typically a time where we aren't given much time budget, as they need it for yesterday).

For something for which i fully control the roadmap, i may consider htmx (but even then, even my own products grow in ambition rapidly) however for customer projects i find this choice way too risky.

I need to be able to capitalize on the tech infrastructure i've built as requirements get complex, both due to customer time constraints and because this is where we get some of the highest profit margin from our work. Having to rework most of what i've done every time a complex requirement comes up, would be inefficient with this regard.

To me it's a case of knowing your tools in depth, to make meet most requirements (including from a performance perspective). Maybe erlang or go would have been more suited for a project from a philosophical perspective but since i specialize in .NET + angular, unless there is hard blocker, i'll use this tech stack, and everything is fine from the perspective of the customer. But i gain that the project is future proof and i won't need to reengineer it when it moves of its initial sweetpot.

This is not a criticism but I suspect that the people enjoying HTMX are either people preferring old-school server-side rendering or react users.

Just sharing my experience, because HTMX is the absolute darling of hacker news.

Since i mainly develop real SPA, i found that is more complex than just using typscrpit if you build something non-trivial as managing state on the server is not fun at all.

In my opinion if you are happy using angular, you'll find that HTMX is a step backward in terms for dev experience. Can't speak about react, but since it is not a battery included stack unlike Angular, i can understand why many people find they'd be better off moving things server-side instead of messing around with 15 third-party libraries. My 2 cents

What i see today is the opposite of what you see : product owners not knowing a thing about software engineering but being able to vibe code prototypes handed over to the dev team are rock stars.

They are closely followed by senior software developers having more of an architecture & design background than a low-level computer science background. Most businesses are looking for builders these days.

Where what you say may converge with my observation is that to be able to do to things such as proper database query optimization, even using AI assistance, you need to be able to understand the concepts of working memory set, cache misses etc...

I've found huge problems, like database servers being grossly underprovisioned (like, 60% cache hit, 4gb RAM server for a 700gb dataset with an 50gb circa hot data set). SSD were used and only latency was measured, so no one realized how problematic the situation was (including a consulting shop they hired to help them manage their DBs - backup, maintenance etc...).

However, having a high affinity with hardware is not a driver / computer science of hiring decisions from what i can see in the enterprise software world. But it would make sense for it to become the case within 10 years. I suspect that you work in a niche where performance optimization matters a lot.

Hum, this reminds me something... "O: open to extension, closed to modifications". Old things are new again.

From context efficiency with approaches such ad DDD and clean architectures, all the way to items such as this one, AI is not creating new tradeoff, it just acts as a multiplier, multiplying productivity for teams doing things right, and multiplying debt for teams having a low quality bar as far as design and architecture are concerned.

And beyond VCs, which are like massive subsidies funded by printed dollars to which no other country has access, even in industries like electric vehicles, Chinese total direct subsidies to their EV companies are like $5bn per year, while the the ones provided by the US to their auto manufacturers are in the range of $50bn per year.

I don't think the US are cheaters or are doing something bad. But i do think that this propaganda about China flooding the market through "overcapacity" and subsidies is very dishonest and needs to stop.

Many people use the term harness to refer to the agent coding software (eg. Opencode, Claude Code...), i use this term more broadly to refer to the environment (set of skills, system prompts, constraints, memory, hooks etc...). What the OP is referring to is not just one giant skill. It's usually a comprehensive ecosystem of skills, bespoke tools to make certain agent tasks deterministic (eg localization), and so on.

I've seen someone post Github repos in this thread, these can be very useful especially if you use the same tech stack, but you won't reach the level of productivity reported by successful teams unless you invest substantial time to build your own harness. But the way to do so is to do it progressively : start with something simple to address the need you have on day 1 . And then, turn recurring prompts into skills, turn recurring coding patterns and coding style recommendations into guidelines, turn repetivive tasks for which the LLM tends to build a python script that it occasionally gets wrong into a deterministic tool documented in a skill etc...

And after a couple of days, weeks, and months, you'll have a very dependable harness giving you optimal productivity, without needing to invest weeks of work upfront or take the fun out of agent-assisted coding.

Hope this helps.

Thank you for sharing this and for having the intellectual courage to hold to a sound reasoning that may be unpopular initially.

For example, if i ask a question regarding an implementation decision while it is implementing a plan, it answers (or not) and immediately proceeds to make changes it assumes i want. Other models switch to chat mode, or ask for the best course of action.

Once this is said, i am not blaming Anthropic For that one, because IMHO the OP has taken a lot of risks and failed to design a proper backup and recovery strategy. I wish them to recover from this though, this must be a very stressful situation for them.

I do not think that anyone who read my comment understood it differently. But I grant you this point, this is just my opinion based on my personal experience not the result of a scientific study.

Once this is said, i wasn't submitting a scientific paper for preprint, just posting my opinion on an internet forum.

Not sure why you are making such a big deal out of it, especially for something for which people can decide within minutes if it works for them or not. And I haven't seen you nitpick on other people saying that all Chinese models are garbage incapable of doing even the most basic task, without quoting any study. This kind of scrutiny tends to be one-sided.

Edit: and regarding what the z.ai team is saying about their models, just check their Discord and the articles they link there. They themselves say that their latest models have leading performance on a number of aspects. It is misleading to suggest that the authors of the model are not proudly saying that their models have best in class performance.

My only goal is to encourage people to try it out so they can see if it moves the needle for them, because there are fair chances that it will. I am not trying to start a flamewar or something.

I also use Claude premium daily for another client, and i use Codex. and i can tell you that GLM5 is at this point much more capable than Claude and Codex for complex backend end work, complex feature planning, and long horizon tasks. One thing i've noticed is that it is particularly good at following instructions and guidelines, even deep into the execution of a plan.

To me the only problem is that z.ai have had trouble with inference : the performance of their API has been pretty poor at times. It looks like this is an hardware issue related to the Huawei chips they use rather than an issue with the model itself. The situation has been substantially improving over the past few weeks.

GLM5.1, GLM5-Turbo and GLM5v are at this point better than Opus, Codex, Gemini and other claude source models. We have reached a major turning point. To me, the only closed source model still in the game is codex as it is much faster at executing simple tasks and implementing already created plans.

Try GLM5v for your PDF work, it's their last generation vision model that has been released a couple of days ago.

As adoption increases, liquidity (which may be what you allude to when you talk about coordination) will be even better.

I'd also note that facilitating liquity can indeed be a complex problem, but there are people specialized in making money out of it and who therefore have an incentive to continually keep the market liquid, read about "market makers".

We are over 8 billion on earth, even if as little as 1% of the population rely on it directly or indirectly due to some form of oppression real or perceived due to political opinions, a long investment thesis on bitcoin can make sense.

I am not a bitcoin holder once this is said. Arrived at the party too late IMHO, and i have better investment options for now from a risk-adjusted perspective.

However if i had more cash to invest than i know what to do with, i'd definitely take a long position on bitcoin.