Hacker News Reader

Comment by trulyme | Hacker News Reader

Well it goes both ways. Young people are cocky and dismiss older peoples' experience as irrelevant, and more senior people think they've seen it all already and that there is nothing new that hasn't been around since 60s.

Neural networks are a good example. If you skipped the last decade (/two) and you think you know about ANN because you mastered them in 60s... boy do I have news for you. :) Another paradigm shift for me was React (declarative web frontend development), and let's not even go into the whole Rust thingthing

One of the biggest pitfalls of an experienced person can be lack of curiosity, and it is really easy to fall for it, because let's face it, most of the new things are crap and will be forgotten in a few years. However, there are nuggets to be found in the mud, one just needs to keep looking.

View on HN

>One of the biggest pitfalls of an experienced person can be lack of curiosity, and it is really easy to fall for it, because let's face it, most of the new things are crap and will be forgotten in a few years. However, there are nuggets to be found in the mud, one just needs to keep looking.

It's easy to reflexively dismiss reimaginings of things that have been tried and failed half a dozen times over the years. But sometimes the concept has been tweaked enough, the environment is different enough, the technology underpinnings are better enough that it actually works this time. Virtualization (z/VM--or whatever it was called at the time) was mostly a curiosity on IBM mainframes for years. Then VMware came along (and Linux on Z was pretty successful on IBM mainframes as well).

NNs sure, React and Rust ... uhm no.

React was a pretty big step for frontend-development; people have done good and bad things in React just as they've done in any other framework, but the important thing it (or other frameworks like it) did was establish a better way of tying state to a series of rendered elements that was better than doing CSS-selector-fu and trying to update things that way.

I'm talking about the difficulty of acquiring these skills given experience. It took me about a week to learn Rust

And it takes me about 2 hours to spin up an overfitted NN in TensorFlow. Learning and mastering are quite different things.

React was doing the right thing for sure, and perhaps in a more principled way than Qt in terms of state propagating downwards and events bubbling up. (CSS still makes layouts unreasonably complex imho.)

But also, React is doing things that desktop GUIs had figured out years ago. It did a really good job at distilling the essence into something useful. Angular started the trend of finally moving towards state-driven views. Kudos to all of these developers for making web development less sucky.

To label it as a new thing that nobody else had done before though, that's going a bit far. New on the web, sure. By the time these got introduced, I had been waiting for years for a web framework that incorporates some of the non-web lessons learned. Let's give it lots of credit, but not more than it deserves.

> To label it as a new thing that nobody else had done before though, that's going a bit far... Let's give it lots of credit, but not more than it deserves.

I never claimed it was "first" and I even called out "other frameworks like it." The iPhone was also not the first smartphone.

> CSS still makes layouts unreasonably complex imho.

This we can agree on. Though it could be a lot worse, and some of the newer CSS definitely gets rid of some of the warts. A lot of the bad CSS is mostly the backwards-compatible stuff.

Not only did React not start that, it wasn’t even in the first generation of frameworks that does it.

And the iPhone wasn‘t the first touchscreen phone, either. Doesn‘t mean it didn‘t have a big impact.

Reply on news.ycombinator.com