3 ms·
Everyone seems to be attacking this as a statement about the obviousness of technological limits. That is exactly what the article is not. The article seems (
by jvanderbot 2mo ago
Everyone seems to be attacking this as a statement about the obviousness of technological limits.
That is exactly what the article is not.
The article seems (at least to me) to be about the uses of technology, and how deeply that interacts with both a) what people want and b) what people are willing to tolerate to get what they want.
The examples here are not architectures or benchmarks, they are stories about the slow change of real life, even if the rate of technological change blurs and accelerates to some degree. It's about the flattening curve of demand and deployment (markets), not of capability.
Yes, the internet, cpu speeds, etc etc, and now LLMs, but the water heater example is an analogy for how "We only want so much of this stuff, and eventually we'll figure out how much is enough". Steve Jobs quotes pin this to consumer tech even more: You don't want the prettiest thing, you want what solves people's real problems in a way that works for them.
TFA is trying to show that you can make a better widget, but its impact might fade into the background, countering the "technoptimist" "singularity" "exponentionaly blah blah", or whatever narrative.
- michael0church 2mo ago[dead]
- sublinear 2mo agoIf the future of software development is more collaborative with the end user, that sounds like the beginning not the end. Nobody liked the steel tycoon era of railroads either. We were already almost there decades ago with plugin architecture. LLMs slot very well into that kind of customization. It doesn't change the social order, nor kill software though.
- to11mtm 1mo agoI tend to agree, it's one of the things I like about this era. My first 'Programming' Job was actually doing automation for CAD drafters. They would ask about a feature, I'd have to estimate the effort to implement versus how much time it would save over 1 or 2 years. If I could implement it quicker enough over the estimated time savings, I got the green light to code it instead of doing CAD work. Which, frankly was good experience for me; I was translating semi-to-barely-technical people's requests into requirements. Skip to today, I can have a complaint about what someone has to do manually, and can easily turn it into what's necessary to get Copilot/Claude to spit out a usable, simple tool to automate the needs of many teams. Sure, it's not always super fancy stuff, but 30-60 minutes of prompting/reviewing/testing the results can save 5-10x that time if not more for teams.
- jvanderbot 2mo agoAnd now it's flagged. Someone got mad about the silicon valley tropes brought up, when this was actually quite a thoughtful article.