Hacker News Reader

Comment by jofzar | Hacker News Reader

I feel like no one has said the elephant in the room.

QA teams were fired/never hired in the first place (put onto the Devs/support/customer to report and test)

Management want features and selling not Lovability and polish. We are just hitting an all time of make make make.

This is the correct take. QA was under attack the entire Agile(TM) cycle we just exited. Now with agentic coding, the only thing keeping it on the rails is TDD. Either you write the tests or you write the spec.

You are also responsible for the output. Welcome to the New Age. Management is just as clueless as they have ever been (some more than others) and yet most of them lack the intelligence to know exactly how it all works. Hell, there’s still plenty of engineers that don’t know how it all works.

Eventually, when you come to understanding or you reach that “enlightenment” stage, no corporate BS will penetrate you and you’ll forever see past their shenanigans. At this stage though you’ll be a grey beard and be unemployable. So they cut out anyone who knows the BS to bring in folks who believe the BS so they can continue shipping BS.

I agree with most of what you said, but this...

> At this stage though you’ll be a grey beard and be unemployable.

isn't as true as people think. I'm a graybeard and remain very much in demand, as do the majority of the graybeards I know.

YMMV but the vast majority of greybeards I know are unemployed or are doing other things than software engineering now.

What I have noticed is that graybeards who work for SV-style companies are treated as disposable, incompetent, etc. But outside of that, things aren't as bad as people think, and most dev jobs aren't in those companies.

True but for backend i just started building through TDD. this has helped a lot

I don't hold a lot of hope for TDD if the tests are AI-generated from the AI-generated code... which I see a lot.

There's a much larger elephant in the room

Ooh ohh Ohh I think I know what it is!

While I don't want to ruin this thread with coming out, I can't find a better place - sorry.

While I have a guess on what elephant everyone is thinking about, I wonder if it's just the stock market - at a higher level capitalism itself. I have felt Android get less stable with pretty much every release for many years now, well before the current boom. I think this is all a long term trend devaluing craftsmanship, with a root cause of why care if you can still make tons of money anyways.

I tend to agree. This does seem attitudinal and around for a long time. Plus “Hurry up and break things” etc. has added to it. Furthermore, support & support metrics which could have been a great signal for such things is, I suspect, measured on something else (e.g speed to resolution of a ticket ) I am surprised that quality does not seem to have caught up yet as a moat - atleast amongst those who can afford it and atleast for big ticket items.

At this point there's a room in the elephant.

No room anymore, just a massive elephant

It's elephants all the way down.

The infinite tower of elephants is ringed by dancing worshippers wearing elephant masks and tusks, bowing with arms raised, then raising their faces up to the sky in ecstasy. "All hail, the Great Elephant of all existence! No room can contain your glory!"

This has been happening for a long time - more than 10 years ago there was a movement away from having QA.

Partly it was about continuous delivery - you can’t push out to production instantly if you have people doing QA.

Partly it was - developers should be responsible for the quality of what they produce.

And it was also trendy to say - I’ve just got rid of our QA team - we don’t need them.

QA is more likely needed for humans, but in the LLM era, when agents write consistent code, there is no need for it anymore in my opinion. A bad example, but still: I recently developed a TG bot without a single test, and yet, it works.

Another elephant in the room is the widespread impact of neverending COVID re-infections of people who don't wear N95s in public (most people). Vaccines don't prevent transmission, which leaves everyone open to acquiring long COVID. Long COVID is very likely underdiagnosed due to widespread ignorance, and not helping matters is that 40% of infections are asymptomatic during the acute phase.

Long COVID can include issues with memory and risk taking.

https://www.camh.ca/en/camh-news-and-stories/rsch-new-study-...

I don't think software is the only field impacted by this, but it's undoubtedly one of them considering how few people take proper precautions via regular N95 usage.

COVID in general is also heavily politicized (as seen in the silent downvotes), and the back-to-normal agenda was forced by capital interests, including tactics like mass disinformation campaigns.

How does this relate to TFA or GP's reply?

How does impaired human memory and increased risk taking relate to software stability?

Respectfully, that's a bit of a stretch. Impaired human memory and increased risk taking are coefficients that would impact far more industries than software. I think the notion is that software, in particular, seems buggier. IF these side effects of repeated COVID infection have anything to do with software bugs, I would think it's contribution would be dramatically smaller than say LLM assisted coding, rapid merging of code, lack of human review.

Just a couple of examples:

- Linux kernel ai commits (https://lunduke.substack.com/p/ai-submissions-to-linux-hits-...)

- Omarchy 4 30k lines of ai generated code (https://x.com/dhh/status/2057907663967543618)

False dichotomy. There's no reason it couldn't impact software as well as other things. Also, it's obviously not the only acting force.

Reply on news.ycombinator.com