I quite like Jonathan Blow. I think this talk is kind of poor though, I remember watching it a couple of years ago in its full context (linked in the youtube description: https://www.youtube.com/watch?v=ZSRHeXYDLko).

IIRC it had a bit of a "kids these days" / "they don't make 'em like they used to" feel to it. It's not untrue, but it doesn't really explain anything.

IMO, a MUCH BETTER version of a similar point was outlined by Bert Hubert in this article:

https://berthub.eu/articles/posts/how-tech-loses-out/ - How Tech Loses Out over at Companies, Countries and Continents (Video version: https://www.youtube.com/watch?v=PQccNdwm8Tw)

In short, this goes beyond software; it's in general about why the incentives around specialization and outsourcing/offshoring cause innovation and expertise to die out. Bert Hubert takes the example of toasters, but it's of course very generalizable:

> The problem is that this is not just a toaster problem. This is a continental problem. All over Europe, this is happening simultaneously, where we’re saying, look, we’re not that much into actually building things anymore.

> So we’re just getting everyone else to build stuff for us. [...] We’re just thinking about things and then telling some other people how to do their stuff.

> In the end, you cannot survive if all you create is intellectual property.

Oh fuck off, Jonathan.

Software is becoming more accessible and more popular. More people can write code than ever before. People are launching companies by cobbling together libraries from Github. Students are learning javascript by playing in the debug console instead of paying attention in class. All of this means that yes, we will see more bad software. That's okay! Some of these people may even learn to write "good" software.

Thank the gods that Jonathan fucking Blow is here to tell us how to do it correctly!

HN is full of interesting things. But not this one. Skip it. Save your time.

There's more software written today than in any other moment in history. Also software is only becoming more complex with more layers etc. It only seems logical the number of bugs would also increase.

Software isn't in a great state, granted, but I'm skeptical that it's a genuine civilization-ending crisis. It doesn't seem like knowledge is being lost, for one thing. For another, we're still exploring new techniques that help with program correctness, like borrow checking in Rust or adding gradual typing to already established dynamic languages.

To be clear, we shouldn't get complacent about the quality of software or ensuring knowledge gets transferred to the next generation. But I don't think these issues are putting us on track to a crisis.

I think the problem is we are utilizing software so much more and it is often so much more complex—perhaps more often than not needlessly so.

I do sometimes share his frustration. Every single day I feel I am working around the limitations of the tools I use and the bugs that just seemingly exist forever until someone decides to rewrite the whole thing and replace old bugs with new bugs. Corporations probably are not the best stewards of software for reliability purposes.

The original was longer, and had more context.[1] His example of the loss of general knowledge about writing video drivers was a particularly good one, in my opinion. We're building far more complex systems, we need tools that offer a better impedance match to the people who build them. Working with source code is like hand editing schematics.

  1 - https://www.youtube.com/watch?v=pW-SOdj4Kkk

Jonathan Blow is full of it. Seriously. The amount of time he spends self promoting and telling everybody how to do things is jaw dropping. He is a good puzzle game designer. That doesn't mean that he is good at anything else.