Re:Software is breaking down at a systemic level
by atomicfiredoll on Monday November 16, 2026 @09:16PM
(#50526558)
When I got to this thread, your comment seemed most aligned to some thoughts I've had for years. So I'll put these here on the article here. Sorry if it seems like I'm dumping.
> The seed was there since at least a decade
Yes, and the seed was always there, it's intrinsic to the way the economic and market systems are set up.
> Microsoft rushed Azure out of the gates under intense competitive pressure. Corners were cut... Instead of going back to the drawing board to tackle the growing technical debt, Microsoft relied on quick fixes
These aren't technology problems. They are what our broader systems incentivize.
> Because there were fewer details to take care of, tech professionals began to lose their specialisms.
Once it was possible, the market incentivized paying existing employees a little more and cutting out more expensive specialists.
My current working idea is that at a broader systemic level there are issues. So, I agree with parts of this article and your statement.
But the focus on NPM, when things like Ruby gems and other open source tech already existed, feels like picking an arbitrary starting point and makes the rest of the timeline feel a bit off to me. The issues existed earlier than NPM. It also feels like the article is somebody working their way toward understanding. The writer can see common threads, but when they try to sew all the different events together seamlessly, they haven't yet quite gotten to the actual nature of things, and can't quite do so.
> the stock market prioritises short term trends over long term engineering.
Very close--It comes very close to what I've been thinking, but doesn't quite totally line up.
Corner-cutting, cost-cutting, over-reliance on uncompensated work, poor organization, bad QA, people who focus on the short term for a few years then move to another job, etc. all exist in other business sectors. Our systems incentivize these behaviors. The security and stability issues (which aren't new) are the tech version of a car recall, or a Nebraskan town getting poisoned by an ethanol plant.
For tech businesses, "moving fast" and accepting the trade offs is/was a fairly logical conclusion within the rules of these systems. The people on the ground will potentially always feel like they don't have enough of some resource: time, talent, etc. because that's the beast.
With the advent of LLMs/generative AI, several trade offs have quickly changed. Things that were previously somewhat safe to marginalize are suddenly priorities, and obviously organizations are not always good at changing directions quickly. Additionally, new companies are being incentivized to enter the market with the idea thanks to "AI" they can do more with less, but they are often unaware of or actively trying to avoid the the full level of effort required.
Attempting to draw some conclusions, these are two of my next thoughts, which I'm not necessarily looking for direct answers or responses to here:
1. Does the software industry really care enough to change. Some right now are speculating that we're going through a painful phase while we effectively pay down the debit of what used to be in the margins. People are risk avoidant, and before uprooting their methodology, they may endure some pain while waiting to see how things play out. Maybe it's not worth changing and this is actually temporary, or on the larger scale this "crisis" is bit of a tempest in a teapot.
2. As long as approaches like cutting corners are effective, some actor will cut corners. If we decide to change, we probably need to consider changes to the system and/or incentives. This may be things like regulation or ensuring that people can stay focused on the long-term and foundational needs while staying healthy and keeping a roof over their heads. (Any regulation should be careful not to further entrench current winners or bad actors.)
On a personal note: I'm sick of tech companies trying to take control over devices I own and force me to use their deficient programs, broken interfaces, and bad products. It sucks and feels bad. I do not like the end result of these systems.
> The seed was there since at least a decade
Yes, and the seed was always there, it's intrinsic to the way the economic and market systems are set up.
> Microsoft rushed Azure out of the gates under intense competitive pressure. Corners were cut... Instead of going back to the drawing board to tackle the growing technical debt, Microsoft relied on quick fixes
These aren't technology problems. They are what our broader systems incentivize.
> Because there were fewer details to take care of, tech professionals began to lose their specialisms.
Once it was possible, the market incentivized paying existing employees a little more and cutting out more expensive specialists.
My current working idea is that at a broader systemic level there are issues. So, I agree with parts of this article and your statement.
But the focus on NPM, when things like Ruby gems and other open source tech already existed, feels like picking an arbitrary starting point and makes the rest of the timeline feel a bit off to me. The issues existed earlier than NPM. It also feels like the article is somebody working their way toward understanding. The writer can see common threads, but when they try to sew all the different events together seamlessly, they haven't yet quite gotten to the actual nature of things, and can't quite do so.
> the stock market prioritises short term trends over long term engineering.
Very close--It comes very close to what I've been thinking, but doesn't quite totally line up.
Corner-cutting, cost-cutting, over-reliance on uncompensated work, poor organization, bad QA, people who focus on the short term for a few years then move to another job, etc. all exist in other business sectors. Our systems incentivize these behaviors. The security and stability issues (which aren't new) are the tech version of a car recall, or a Nebraskan town getting poisoned by an ethanol plant.
For tech businesses, "moving fast" and accepting the trade offs is/was a fairly logical conclusion within the rules of these systems. The people on the ground will potentially always feel like they don't have enough of some resource: time, talent, etc. because that's the beast.
With the advent of LLMs/generative AI, several trade offs have quickly changed. Things that were previously somewhat safe to marginalize are suddenly priorities, and obviously organizations are not always good at changing directions quickly. Additionally, new companies are being incentivized to enter the market with the idea thanks to "AI" they can do more with less, but they are often unaware of or actively trying to avoid the the full level of effort required.
Attempting to draw some conclusions, these are two of my next thoughts, which I'm not necessarily looking for direct answers or responses to here:
1. Does the software industry really care enough to change. Some right now are speculating that we're going through a painful phase while we effectively pay down the debit of what used to be in the margins. People are risk avoidant, and before uprooting their methodology, they may endure some pain while waiting to see how things play out. Maybe it's not worth changing and this is actually temporary, or on the larger scale this "crisis" is bit of a tempest in a teapot.
2. As long as approaches like cutting corners are effective, some actor will cut corners. If we decide to change, we probably need to consider changes to the system and/or incentives. This may be things like regulation or ensuring that people can stay focused on the long-term and foundational needs while staying healthy and keeping a roof over their heads. (Any regulation should be careful not to further entrench current winners or bad actors.)
On a personal note: I'm sick of tech companies trying to take control over devices I own and force me to use their deficient programs, broken interfaces, and bad products. It sucks and feels bad. I do not like the end result of these systems.