Comment by nonameiguess

This seems like somewhat of a mischaracterization. He contrasted improvements in software with improvements in hardware, saying we'd never have something like a Moore's law for software where performance doubled or cost halved equally rapidly. Churning out software faster doesn't mean it's any better.

We also seem to fall into these ruts of not understanding what is meant by labor productivity. When an economist is presenting the common outputs / inputs measure, they don't mean raw quantity of output. They're talking about the value added by outputs divided by the value of inputs. Churning out software faster that doesn't earn anybody additional revenue is not making us more productive. It's disheartening that even c-suites with business education don't seem to understand this. That's not to say there is no productivity gain. Plenty of AI-adjacent hyperscalers are seeing ridiculous growth right now, but no non-startup is seeing revenue 10x what it was the year before, not even NVIDIA.

A lot of this is just basic diminishing marginal utility. There is only so much value to be added. Software is usually either a semi-automated controller or human decision making augmenter to some kind of physical manufacturing process, or entertainment, when we talk about ultimately delivering value. Everything else is an intermediate input. We can only be so entertained. For physical goods, we have food, space, clothing seemingly at a sufficient level to satisfy just about everyone, with the reasons for value not being maximized having to do with distribution. Unless your software manages to solve borders, bigotry, cultural incompatibility, poverty, mental illess, physical illness, violence, I'm not sure what the other big rocks are. Software can absolutely be a key part of infrastructure to facilitate distribution. That's exactly what the Internet is, along with all the backend business and logistics systems out there in existence. But without hitting the true big rocks, where is the 10x value supposed to come from? We're talking incremental gains simply because we're not in the dark ages any more and incremental gains are all that's there. Short of Star Trek style replicators and transporters, I'm not sure what could realistically multiply global value by 10.

Without the value, then sure, you may be churning out 10 times as many discrete projects used by at least one person, or 10 times as many lines of code, but that was never the point. Your personalized notes and grocery ordering apps you share with your wife might excite you for a few weeks, but I can assure you they aren't going to revolutionize your life.