I can't think of a person less qualified than him on this subject. He thinks so much of himself, and less of others. Of course he would think about the collapse of society around him.
-
Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
-
We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.
Jonathan Blow - Preventing the Collapse of Civilization
- Thread starter Soundchaser
- Start date
- Status
Man would it help to actually watch the video at least for a few minutes, huh?
Things are more complicated, but I don't believe it's for no reason. If nothing else, we're now slowly starting to take security halfway seriously. Which yes, means you can't just copy a file off the internet and run it directly anymore.
I've never liked arguments along the lines of "this thing is bloated for no good reason" that don't actually try to get at what reason existed for adding the bloat in the first place. You can't just sit outside the problem and say "I don't know why we need installers, thus we don't and they're stupid."
Anyway, on the whole Blow makes some good points, but I think he's exaggerating. It's true that a large percentage of programmers nowadays work at a high level and don't know the nitty gritty, but I would be surprised if the number of highly knowledgeable coders has decreased. We've successfully lowered the barrier to entry, growing the pie and letting devs of all skill levels contribute to our growing software ecosystem. But we've still got plenty of nerds pushing bits around.
Plus he's really selling modern developers short. Yeah, some guy can make a robust text-only editor in a week, but modern editors' capabilities are so far advanced that of course they'll take more time and have more bugs. The advancements we've made may not be cool to Blow, but smoother and more user-friendly interfaces are progress all the same.
and moving to "abstraction" and to what it leads:
It's getting all Warhammer 40K in here.
I agree wholeheartedly with the idea that many programmers nowadays don't bother to understand how their code actually functions on a hardware level and just trust managed development environments to do all the "hard" stuff for them. It annoys the heck out of me to see how many supposed programmers struggle with the most basic of low-level programming problems (protip: pointers are literally just address locations stored in a variable. They "point" to the address in question. That's it. This is not hard for goodness sake) because they were never taught or never bothered to learn. And it's not like you have to program in Assembly or anything like that, you just need to know what your high-level code evaluates to on the hardware level so that you can write better code and so when nastier bugs inevitably pop up you can have a good idea of why it might be happening instead of none at all.
Maybe I'm just a staunch traditionalist. I'd program everything in straight C if I could, even the most simple and straightforward OOP implementations introduce enough abstraction to give me a headache.
The idea that terrible software practices are being papered over by extreme and quick advances in computational power is a good one too. Why bother writing good code when smartphones and PCs are so powerful nowadays that inefficiences barely make a dent for most applications?
Maybe I'm just a staunch traditionalist. I'd program everything in straight C if I could, even the most simple and straightforward OOP implementations introduce enough abstraction to give me a headache.
The idea that terrible software practices are being papered over by extreme and quick advances in computational power is a good one too. Why bother writing good code when smartphones and PCs are so powerful nowadays that inefficiences barely make a dent for most applications?
His point also that programming becoming actually more complicated because of this. Too many unnecessary things.
Many things were much simpler even decade ago. Now it's overcomplicated for no reason.
Some of those complications are things we as people wanted. We can't have it both ways in "I want my computer to do this" but also "I don't want to have to do this to achieve it". And a lot of those "wants" are more security focused as a result of malicious actors proving what could be accomplished when the OS just lets you "do" stuff. People HATED UAC in Vista (and I don't blame them personally), but that was a response to the probably vast amount of examples Microsoft found of people having shit run on their machines without their knowledge or approval and the nightmare it was. It was an imperfect solution to what was becoming an increasingly larger issue.
I do not believe him, for a second, that generally we don't "want" what OSes have become, aside from things like Windows 10's extreme telemetry data gathering. And I say this as someone who still regularly uses machines from the 80s/90s and their respective OSes, when things were less "complicated". Because what we expected of systems then is not what we expect now.
And it's been right every time. Post-internet Earth might as well be a different planet than pre-internet Earth. Post-smartphone Earth is populated by an entirely different dominant species as pre-smartphone Earth.
The future is a bunch of bloated business Java code running inside a matryoshka-doll stack of virtualized machines and nobody will know how all of this work anymore...
I kind of understand his point and, to a small degree, it's something I have seen when confronting my experience in most of the IT industry versus how I use tech in my hobbies. Complexity is a cancer, but good luck changing the environmental causes.
I kind of understand his point and, to a small degree, it's something I have seen when confronting my experience in most of the IT industry versus how I use tech in my hobbies. Complexity is a cancer, but good luck changing the environmental causes.
Nah, it will never get to that point while we're alive.
The machines will just become self-aware and kill us all before bloated Java code takes over the world.
Calm down, I was just joking around. I like Blow and his games and he gave a good talk.
From a developer perspective, nobody wanted that. At least not what we have now.
Back then you had a task and you were writing a code, now – 90% of time you're basically maintaining and managing all those junk around it, even if you need to do just a simple fix. It's not productive.
why should anyone take what Jonathan blow says about anything seriously?
you should really understand how a typewriter works if you want to write a novel
says man who invested a lot of effort into building his own typewriter and is annoyed by the existence of word processors
says man who invested a lot of effort into building his own typewriter and is annoyed by the existence of word processors
because he's a great game developer, a smart guy, and he appears to be presenting a pretty insightful take on a nuanced issue that's been getting worse as of late.
knowing how a type writer works does nothing for a writer. knowing how a computer and low level software works can do wonders for the average developer. it's also important that we continue to have low level developers going forward.
Unfortunately, the goal isn't "productivity", the goal is safety. And I don't mean just from a security standpoint. The logical consequence of trying to make software development appealing and accessible to a wide range of people is that development tools must then hedge against the segment of that wide range that are passionate and motivated but have no idea what they're doing. This means that said tools now have to be sandwiched in-between layers and layers of abstractions, constraints, and sanity checks so that when aspiring software developers make dumb mistakes they aren't immediately greeted with instant memory corruption and blue screens.
I stress though, that these aren't bad things on the surface. Keeping expensive hardware from failing because someone mismanaged a pointer is absolutely a good thing, and as the complexity of our software increases so will the opportunities for serious problems to arise by even the most disciplined of developers, so measures to limit the effects of those mistakes are welcome. But the net being cast in this pursuit is so wide that inevitably there are checks and rechecks and all kinds of constraints handicapping the developer's ability to do even the simplest tasks because someone out there will screw up even the simple stuff. Memory management on a high-level isn't even a real thing anymore because these high-level programming tools can't trust the people using them to keep their pointers straight at all. Just let the GC do everything and don't worry about it. And all of this leads to bloated, messy code that has to jump through hoops to do even the simplest tasks outside the parameters of what the development tools in question consider "acceptable practice". It's irritating. Let me decide what "acceptable practices" are and if I screw up majorly, that's on me.
Many institutions that use software heavily in their backend infrastructure still maintain old, outdated and narrowly purposed embedded systems as part of their operation. Why is that? I suppose laziness and the costs and headaches associated with moving away from those systems is part of it, but I think the bigger reason is that even now there are still no better alternatives to many of those systems for the tasks they're designed for. And judging by how software development is progressing, there may never be.
The heck is that analogy?
This is more like if novel authors used ghostwriters to turn their ideas and narratives into finished novels without having actually written novels themselves.
Last edited:
unlike writing, software development is engineering. So you have to know all these stuff.
Imagine what would happen if architects will start designing and drawing up plans for buildings without knowing laws of physics.
I'll listen to him talking about tech. He has very similar opinions regarding image quality as me, for one thing.
Not about pretty much anything else though, that I agree.
If anyone else is tired of low-effort posts only designed only to derail discussion, I highly recommend hitting the "!Report" button on the lower left of these posts.
The advancements in machine learning and steamlined programming has led to people who know nothing about code to create some pretty compelling software, but I do understand the need for there to be people who know how everything works under the hood. I feel like there will always be that subset of programmers/developers who do learn the in depth aspects of a language, with others content with simply making content without worrying about why everything is working.
In relation to the most important thing in the world, video games, I feel like Blow's point is somewhat divorced from the business reality of the market in that games need to be put out consistently and adapt to completely different hardware structures. People are drawn to Unity and Unreal not simply because they are "easy" but because not every game has the time and money to use something else.
There are already very few tasks that need low-level understanding of hardware; and yet the internet is absolutely filled with people doing clever, amazing stuff that took profoundly deep knowledge — and then sharing and discussing that knowledge through articles, tweets, etc. I don't see that changing. I think it's much more exciting that there are so many areas of game-making that are being explored by people who don't have opcode cycle count tables memorized.
And then there's people who actually make hardware who are still around. We still exist :P
Yeah, Jon's right about a lot of this stuff. I think he's a developer in the middle of the frustration, like so many, but he is actually doing more than most to put his time where his mouth is: he's developing a compiler meant to simplify the issues of a language like C++ (and, perhaps, learning that it is not so easy to create simplicity in certain aspects), and he is building his newest game to run on that language. I don't think, realistically, that the end of that particular story ends up being very happy, but who knows, it could be a snowball of sorts. Something he hasn't discussed too much here, but has managed to discuss in other places, is how programming is just not very much fun anymore when it comes to dealing with certain types of problems. When you're trying to teach yourself how to do simple things, the amount of complexity in getting there is so daunting, and frustrating, and painful, that things like Unity or Unreal feel like the only viable option for everyone. And yeah, that's a mix between a lack of education, and the daunting reality of development today. You need to have a reasonable amount of technically-inclined friends to get at this, but if you are open and honest about communicating what you know and what you don't, it's surprising to see peoples' gaps. I got nearly the entirety through a CS program without knowing how to statically link code. Meanwhile, a friend makes 100,000+ as a programmer and doesn't know how to copy a string in a compiled language like C, or understand what a pointer is. Neither of us are inclined towards genuinely difficult, defining problem sets -- we're just trying to survive in a bloated, complicated field, that isn't getting easier or more fun to be a part of.
Modern languages ought to be better, it ought to be easier to do most of the things we do, and if we persist on the track we are on, we'll probably have a bad time.
Modern languages ought to be better, it ought to be easier to do most of the things we do, and if we persist on the track we are on, we'll probably have a bad time.
I've listened to him making similar points before but I found this one the most persuasive. The historical annecdotes really helped put things in perspective. Perhaps he could have ended on a more hopeful note but that would probably only have reinforced his point about our tendency to ever-so-optimistically sweep things under the rug.
Amazed by the humility expressed by Blow through the implication that he couldn't survive in a post-civilization world. Not being able to get off the Great Plateau in BotW without help must have humbled him.
This is basically redefining the singularity as "any technology that causes significant change in QoL". Which is the only way for the singularity to actually exist.
Indeed. The singularity is optimistic wish-mongering based on nothing. It's a science fiction scenario that anxious smart people have convinced themselves is an inevitability for some reason.
Heck, most religions have a more plausible and historically sound foundation than the singularity scenario ever will.
It's not about memorizing, it's about understanding how it works, and more important – why it works.
Plus "memorizing" is wrong way to learn. It solves nothing.
Richard P. Feynman:
The singularity will totally happen in 10 years. You just wait!
Of course it doesn't even matter because we're all in a simulation based on my probability exercise that has the basic assumption we have to be in a simulation!
When the need is great enough people will have to go get their hands dirty.
The torch will eventually need to be passed, but I think it's more important that they could learn those skills than it is that they have those skills when they don't need or use them. Like, every programmer should learn assembly at some point. But forcing them to use it all the time is cruel and unusual punishment.
It's not easy to mentor new engineers, but it's not hard either. What you're describing is not a real problem, unless your goal is to keep programmer wages low.
Progress will only slow down in the sense that gamedev companies might have to pay programmers real salaries and have them work sane hours. Oh no
None of these things are remotely unnecessary, fwiw.His point also that programming becoming actually more complicated because of this. Too many unnecessary things.
Many things were much simpler even decade ago. Now it's overcomplicated for no reason.
There is cruft in the software field (web dev is ground zero), but generally speaking stuff is the way it is for a good reason. Even in web dev (although that's an embarrassment). It's not to say things couldn't be simpler, but the reason there are always new languages, libraries, and frameworks is in the quest to make things better, and insofar as those things continue to fail to make things simpler "enough", it's a sign that maybe the complexities reflect the underlying realities.
Last edited:
I feel like there are always going to be people who think coding to bare metal or learning the ins and outs of assembly is the best approach to everything. I don't expect to see an Idiocracy-like scenario where suddenly no one knows how to make a game engine anymore because everyone's been dragging and dropping their logic in Unreal Engine or whatever. There have been and (I think) always will be people who love starting from first principles and building up.
There is no "dumb racists nonsense" being thrown at white people.
Learn what white privilege and whiteness is, it will do you some good.
Imagine criticizing someone's ego in a manner that betrays your own blatant arrogance considering you couldn't even take the time to verify whether your hot-take actually applies to the topic at hand.
Man, these examples for productivity are terrible, if not downright fallacious.
Veteran software developer not qualified to talk about about software development?
No, it really isn't. You can code an application without ever engineering any part of it. Tons of software developers only ever use engineering in their work when things start looking too rough and they have to science it out, and happily code as though it were text everywhere else. All kinds of people are using code to suit their needs with no actual knowledge or training in engineering.
Coding is essentially writing text.
Woah. I don't know how much I agree with Blow's analysis, but I would say spreading dogma like this could lead to the collapse of civilisations.
Coding is solving math and logical problems with text.
There are a lot of problems in software development that require 'engineering'. Engineering side of software development is not all about low level coding.
Of course using external libraries and frameworks help a lot and removes a lot of overhead, but how you use them and integrate with other system most of a time is an engineering problem that needs to be solved.
And of course not all engineering problems are equal.
The OP in this thread consists of a video with no context or a summary of its contents. This framing does not allow for good discussion, and thus the thread is closed. Please feel free to create a new thread on the same video with a more descriptive OP and title.
- Status