I think concrete examples of this are tricky because these "mini frameworks" only exist inside of a companies' proprietary codebase. My understanding of the concept is its an abstraction layer built on top of a more general abstraction, except the more general version is well documented and well understood by the company (for an internal framework) or even overall developer community (for something open source).
> Also, people lack the motivation to maintain existing stuff, because you don't get paid more or promoted doing this. Therefore, mini frameworks often die with the departure of the original authors, unless it has gained major adoption before that, which happens less likely than not.
I wonder how much of the problem stated in the article is actually a result of this resume-driven development style? The author says how the mini-framework was pushed by their engineering manager, I know it's cynical but I assume the real goal of the project was for the manager and engineers building the framework to have something fancy to show for their next promotion packet.
> I don't think that's true. A large backbone app has a lot of code that you'll have to trace through multiple files in different directories
This is exactly my experience with large scale Backbone apps (from 10+ years ago). Even with extras like Marionette it quickly became a complete nightmare to navigate or maintain. Zombie model and view objects leaking memory was almost inescapable.
I remember in 2013 I introduced Backbone to my current company, hoping to make sense out of our existing jQuery + ASP.Net MVC application. After ~3 months of code from junior and mid level developers (in house and off shore) I began to deeply regret my decision. There was just not enough patterns and utilities in the framework to keep things from going off the rails. We eventually shifted to Angular v1 and it was glorious, things just worked and even the ceremony I needed to add felt worth the trouble for the speed of development we gained.
Angular v1 was absolutely the game-changer for sane UI patterns. Although it could choke on large datasets and models were iffy, it allowed us to worry less about the framework and more about the work at hand.
At my last employer Elastic we definitely ran into these limits on the cloud SaaS team moving Elastocsearch node containers from our proprietary orchestration to k8s. I’m not sure how they eventually solved it but I believe the plan was essentially sharding ES clusters to different regional k8s clusters.
It’s hard for me to imagine how the current superpowers (US,China, Russia) governments will be good stewards for AI, or really anything important. I’m not saying big tech is better but with the world heading more in the direction of autocracy and fascism it scares the hell out of me for these people to control the direction of AI.
Can't answer about China nor Russia, as I'm not very familiar with how their governance works, but in countries with democracy, the government is supposed to be representative of the population (which may or may not actually be the case, it's the idea at least).
Contrast that with companies and corporations, where the basic idea is "Company makes money, our decisions should support that".
I sure know where I'd put my eggs on who could potentially run AGI best, considering the options. But I'm also a believer in democracy, and would ever prefer autocracy by a government than autocracy run by companies.
So assuming worst case (we're all be living under AGI that steers us into fascism and autocracy), I'd still rather that be under entities (supposedly) controlled by people, rather than entities under shareholders and CEOs.
> I'd still rather that be under entities (supposedly) controlled by people, rather than entities under shareholders and CEOs.
I think I felt that way most of my life until this current administration. I think I’d be more comfortable if AI nationalization (and scientific research in general) existed outside the control of the executive branch.
> I think I felt that way most of my life until this current administration.
Yeah, I understand for people who live in "faux democracies" might see this different, but luckily most of the western world has proper checks and balances, especially compared to the country where you (unfortunately) seem to live.
>but in countries with democracy, the government is supposed to be representative of the population (which may or may not actually be the case, it's the idea at least).
That you used 'supposed [to be representative]', i can only think of Switzerland where this may be true, or at the least, fairly close.
I imagine the folks in the article and others like it are not building libraries and foundational infrastructure but rather cranking out SaaS startup ideas and CRUD web apps. I find that kind of coding really can go quite fast using AI, particularly if you are building it from zero and not worrying about all the existing quirks of a large codebase or creating technical debt.
I know it’s cynical but it feels like people make these statements so they can feel like they are making some difference, while in actuality all they are doing is signaling their political belief system. Not a fan.
Doesn’t c++ also make a separate type for each distinct type parameter passed to a template? If that is still true maybe it adds up when you have a lot of different vector of T types? Been a while since I read up on c++ though maybe that’s no longer true.
10 years ago I was at an “innovation lab” of a big HR tech company that gave out free lunches and snacks. The best explanation I heard was a coworker originally from the big boring corporate HQ who said basically this is a software petting zoo for big investors to look at. “Good code monkey, here’s a banana!”
reply