> It's just not possible (and maybe not economical viable) to build enough software to make everything not suck.

Honestly, it's been my experience that there's no motivation to do this, either. Many of the people that buy the software are more interested in a shiny, new button than they are in making sure all the existing buttons do what they want. And they each want a _different_ shiny, new button... and too many (barely functional) features just makes a product worse.

> not economical viable

I think that's part of the key. Nobody wants to pay for great software

>Honestly, it's been my experience that there's no motivation to do this, either. Many of the people that buy the software are more interested in a shiny, new button than they are in making sure all the existing buttons do what they want. And they each want a _different_ shiny, new button... and too many (barely functional) features just makes a product worse.

I was working at a particular organisation 13 years ago, and we were tired. Everything was a half completed project. Everything needed work. One of the file servers was busted. We had cobbled together enough to make the customer experience ok, but the guts were on the deck.

The organisation expanded, hired a new CTO, moved the old Pseudo CTO to an architect role. New CTO sat down with everyone in the team for a 1 on 1 chat.

He asked what the biggest issue was, I said we needed time to fix everything and make it work. He said everyone on the team told him the same thing. That we have a solid environment and it just needs to be completed.

Next day he announces a shift to the cloud. We had all our priorities suspended as we forced 365 and Azure into everything. I bailed like 3 weeks later.

I think it is economically viable but we as devs have to realize our true worth here beyond just a paycheck.

Do you think that LLMs will do much to help to alleviate this?

I just did a major refactor of a project to move it many versions up on a framework and whole process was effectively vibe coded. I'd estimate I did in a couple of days what would have taken a couple of weeks.

That's good and expect that could be shaved down even more. I was spending most of time just waiting for it do the work.

But I don't know if that fundamentally changes the situation or not. We've had steady improvements in developer technology for decades. Even pre-LLM, I'm building significantly more complicated applications now in less time than ever before. But as quickly as our developer technology improved, the demands on applications we build has gone up. I'm not sure even LLMs can outpace the demand for software.

Thanks for your honest opinion. Not sure why you were downvoted.

Likely because some people don't like to hear about positive experiences with LLMs.

I don’t get it either. I feel people are increasingly knee-jerk about downvoting, triggered by certain phrases, perhaps „vibe-coding“ in this case.

Link to blog post? Didn’t see it on quick look at your site.