* The current generation could not invent the internet. They don't have the attention span or the culture for it.
* People who spurn well thought out technologies like SQL or relational databases, are doomed to re-invent them, badly.
Or you could just not make broad generalizations about different generations. Which is preferable.
I don't blame anyone, every step down this path has seemed like a good idea at the time, eh?
You vastly underestimate how much thought went into the internet. Look at:
- TCP delayed acks + Nagle’s algorithm
- IPv4 address size (Vincent said 32-bits was roughly as likely as 24 or 48).
- the 7 layer OSI
- class-based (A, B, C, …) routing
- most of the options in IPv4 headers
The Internet was as much hacks and a playground as other layers of the stack are today. The good stuff is still in use but there is a vast graveyard of bad ideas that didn’t pan out because they were security nightmares, couldn’t scale, etc.
The same applies to core protocols like BGP. BGP hijacking is still possible today.
.. in the same sense as http hijacking is possible today, i.e. without httpS. For BGP we have RPKI ROA for a quite a while, and ASPA is coming.
> and ASPA is coming
Yeah, but it’s not here and requires verification at each AS append, which means we’re a good 5 years before it’s really enforceable on arbitrary prefixes across transit networks.
> You're vastly underestimating how hard the old thing was
Sounds like you're agreeing with him, if anything. Where is gerbilly's 'underestimation'?
* 99.99% of modern software startups could be built, maintained, and scaled in production by a single dedicated programmer and served on a single machine. Yes, even with tens of millions of users.
* Most tech companies hire for the prestige and to not feel lonely, not because they need to hire. (And to have pretty young interns walking around the office).
* YCombinator has deviated far from the original vision of Paul Graham and his essays. Nowadays all of the startups coming out of YC batches are just trash, and the YC panel has suffered "death by committee", i.e. bland risk-averse pop conformism riding the trends with no deep understanding of anything.
i don't understand how this would work
Even production Twitter in 2023 is estimated to be able to fit on a very beefy machine: https://thume.ca/2023/01/02/one-machine-twitter/
Of course we would use a CDN in 2023, which is cheating, but from a business point of view it is juse one machine you manage.
This simply reads to me like a "kids these days" type of old-man argument, which has, as we know, existed since time immemorial.
And this counterargument has existed since time immemorial + 1. It proves nothing.
But I stand behind what I claim. A large collaborative, and disinterested project like DARPANET could not get off the ground today.
Are you seriously arguing that the generation who used pocket protectors and HP calculators didn't have an advantage in concentration over the one who's been shown to distractedly check their phones every 15sec in case a cat might be doing something interesting?
But really the argument behind your argument seems to be that there is a monotone increasing function to 'progress'. I'm arguing that it's not and that sometimes we regress.
We can't even agree on what is real anymore, isn't that a serious disadvantage?
Where did I ever say that? Your claim was about the Internet, not "progress" in general.
> Are you seriously arguing that the generation who used pocket protectors and HP calculators didn't have an advantage in concentration over the one who's been shown to distractedly check their phones every 15sec in case a cat might be doing something interesting?
Yes, because again you're generalizing to an entire population based on a stereotype. Do you think everyone back 50 years ago was similarly concentrating? No, because certain people can simply concentrate more than others, and those are the people building the Internet back then, and those are the same kinds of people who can build an Internet today. 99% of people were not doing so, we are talking about the differences of 0.0...1% of the population who built the Internet, so let's compare the same amount of people today too.
I have worked at a lot of companies and I find the kind of people who can or are willing and able to do that is decreasing. It's nearly zero today.
If the Internet was invented today, it would work only for the happy path and fail with a million different error modes, each handled differently depending on the specific implementation.
And you know what? People would accept it's unreliability probably, the way we accepted the 'all computers crash' from the DOS era till about NT 40. Everybody knew that computers just crash like that. :shrug:
Today ChatGPT makes up facts and gets things wrong. But it's OK cos it's newer than Google, or printed books.
Again you're focusing on coworkers and teams here and there at various companies, not in the best of the best who are the ones who built the internet and who would build it today. They are not average developers like one might normally work with, as I said, they are 0.0...1%.
This is exactly how the internet works today and you just don’t see it. Try using source routing in your IP packets and see how far it gets. BGP implementations had to carry a “work with Cisco’s bug” flag for decades. Send some >1500 MTU packets across the world with varying paths and see how fragmentation and assembly is inconsistent across operators and implementations.
RFCs are loaded with errata, they skimp on critical details, and have bullshit “design by committee” features that never panned out. They aren’t special.
You seem to be ascribing some moral failure to the younger generations whereas there are also factors at play that go beyond individuals. It might be more sensible to say that the internet could not be invented in the current economic/social climate.
It wasn’t a giant coordinated thing that needed decades of patience. It was the epitome of an iterative project that involved stitching university systems together as the opportunity arose.
Meanwhile $GENERATION is inventing general AI. Your so-called unpopular opinion is basically "old man yells at younger people", but with even less foundation.
I've had it confidently invent URLs when challenged to show the sources for an obviously false claim. It invented a URL which has never existed.
I mean it's impressive, but it has a long way to go to become intelligent.
You claimed the younger generation doesn't have the attention span to invent the internet. Someone else mentioned the clear and ongoing progress in machine learning, which involves many things not named ChatGPT, and has become one of the biggest economic engines of the world.
You're just making an inane generalization. You will rightly be criticized for this.
And you're completely missing out on the inevitable fact that those zoomers could teach you a lot of things you don't know too.
They're not inventing anything, they're just taking the concept of machine learning that has been around for ages and throwing more data and hardware at it. Underneath it all, it's still just an algorithm that finds and repeats patterns and nothing more, the patterns are just getting more complex. It's not in any way comparable to the invention of the internet in terms of innovation.
Not saying current generative ai will not have a huge impact. It will.
Do we lack the vision for such big projects? Absolutely. Is it only the current generation's fault? Absolutely not. This is very widespread.
> They don't have the attention span or the culture for it.
I'd argue there is almost _no_ technical culture whatsoever. It is basically "trending on HN" and comment section discourse. This is a massive problem and contributes to the shoddy state of a lot of software nowadays.