It seems like throughout human history, every piece of technology has had a complete chain of human knowledge for how it worked. No one person knows how to build an EV starting from dirt, but someone knows each step in isolation.
With LLMs, some people are claiming that the time for understanding code is over, and that leaves a gap in the chain of knowledge. I can't think of a time when such a gap has been adventageous, and can certainly think of times when it has come with risk or loss. Many of us have seen this manifest at small scale when the one person with obscure system knowledge leaves a company without adequate prior knowledge transfer.
At large scale, well, I think it will be a great time to be a dev who knows how to hand-code.
At every place I've ever worked that hired juniors, the juniors always have to leave by about 2 years, because the company will never (roughly) double their salary to what they can get on the open market.
That first year, the juniors are probably a net drain in productivity - for their work and the time other devs spend teaching them. Then the 2nd year, they're underpaid for as long as you can hold on to them. But the good ones will be out the door pretty quick.
The system always seemed kind of broken to me. But a shop without any junior devs is also kind of sad. If you hire the right person, it's rewarding to the other devs to teach them and watch them grow.
> I have been tracking junior software-engineering listings on LinkedIn and Indeed since late 2024.
I don't think such a short timeframe is indicative of anything at all. This data could be interesting if comparing to say before Covid or even before 2008, but as is, this is far too short of a timeframe to draw any actual conclusions. How would you know whether we're currently under hiring juniors or we were previously just over hiring and are now returning to the norm?
Good point to a degree, but it applies to all work, not just software engineering, and even before they get into the workforce. Young people are not going to have any depth of skills or knowledge if they literally just act as a go-between for AI to do everything.
Maybe they become experts in wiring up AI and robotics?
But 5 years out, we might be into a new compute-in-memory paradigm. And we may have exceeded the complexity of the human brain with models greater than 100 trillion parameters.
We already see performance that is superhuman in task specific ways with 1-10 trillion parameters models.
In under 5 years, most small and many medium business will be handled directly by AI agents. You won't even necessarily know what kind of software they are running. For small-medium business, it may have very little or no code and just be a semi-structured database that is neutrally rendered into any needed form on-the-fly.
I feel like this is very possible eventually, but under 5 years seems unrealistic to me. The compute power is already running into limits. And a few more major problems need to be solved before agents can have that kind of autonomy to create software from scratch and run entire businesses.
Given the intense pressure to find more efficiency, 5 years is probably enough to see deployment of radical new compute-in-memory paradigms such as nitrides-based ferroelectric devices that have a few orders of magnitude more efficiency.
I think the biggest problem is jagged intelligence which I would bet would be solved largely with bigger models, which comes back to rapidly evolving hardware paradigms.
Oh, this is talking about the junior to senior pipeline. I thought it was going to be about improvements in AI capabilities.
I think it's reasonable in 5 years that most of what I do as a senior dev (yes, even turning vague notions into concrete deliverables, giving high level ideas of tradeoffs, and so on) will also be automated away.
Does anyone have interesting blog posts on what that means? Ideas about what the new world may look like? Will existing software engineers be running "dark factories"?
In less than five years it will be clear that unaugmented humans are a bottleneck.
So many jobs as such may be limited to humans with high bandwidth BCI connected to AI, that effectively gives them an AI exobrain. But even then the AI would need to do most of the work or the human would just slow things down.
I was just discussing this yesterday. The only time when Mythical Man Month isn’t a problem is when C-suite/management needs to burn their budget up and projects are behind. Usually you’ll end with teams with too many juniors. The leads are only focused on the harder problems and dont have capacity to properly onboard juniors. So management hires contractors from Europe/Asia. The juniors offload their work and unknowns to contractors who are often also very unskilled.
This has been going on for a long time in tech. Now with LLMs you have a bunch of juniors who are unskilled and a bunch of management trying to fill the skilled spots with unskilled+LLM workers. The seniors are starting to become plagued with people who are excellent at being unskilled in the work place. They maintain shrewd hiring practices until the projects are behind enough and we arrive at the original need to hire contractors.
“Mythical Man Month” is just something people say to indicate status. No one is using it’s lessons in actuality otherwise we’d have to acknowledge over half the work force in software are a bunch of frauds delegating their work and decision making to others.
> The new thing is on the supply side. Companies are already reporting staff+ vacancies that take 66 days on average to fill ... The shortage is already, in 2026, exclusive to senior-level roles.
I am not sure I agree with this being caused by lack of candidates. I have 25+ years of experience, I was caught in an office closure layoff in early 2025 for the second time in my career (the last was in the mid 2010s, where it took me a few weeks to easily find a new position), took the spring off, spent the summer preparing for the leetcode hazing ritual and interviewed a lot in the fall: despite having very up-to-date skills (I always keep up-to-date personally, also my last position was all k8s, llm integration, you name it) and long stints in my resume, interviewers did not care.
In the positions where I managed to pass the automatic resume screening (which was probably 10% of the time it felt), the interviewers just did their typical 5-rounds-sudden-death interview loops, with sudden death happening if the solution was not optimal. Sometimes sudden death was "oh I see you have experience in cloud 1 and cloud 2, but we run on cloud 3, sorry", other times it was a friday afternoon slot where the interviewer obviously just wanted to be done with the week.
It also did not help that many positions were "remote but only if you live within 50mi of one of our offices", and others were trying their best at downleveling (I have 15 years of my 25+ at principal+ both architect and IC, and some recruiters were pitching me senior positions, as in senior senior, not senior staff / senior principal). Often the hiring mgr would bemoan the fact that so many resumes are fake, and so many candidates just outright lie, but none of this seemed to make a difference when it came to the interview setup.
Ultimately after about 3-4 months of searching (I'd say I applied to ~100ish positions where I thought I'd be a good fit, and had actual interviews in less than 10) I ended up finding a local position for a significant pay cut, it is also the first job where the mandate from high on up is AI-everything, velocity first (like it's mentioned in this article). Given my long time experience I am able to be very productive with claude (because I know what it should be doing) but even with really tight reins it is a daily occurrence for it to prioritize expediency over correctness requiring rework.
I honestly don't know how people starting with AI agents will manage to improve, I have been through the transition of assembly->higher level languages, and the "spec driven agentic is just like using a compiler" rings very superficial to me: nobody would accept a compiler that 5% of the time (being generous here) miscompiled your code.
The current push seems to be "it's fine to have juniors on claude code producing tons of PRs, we'll just put a claude / codex reviewer action in GH so velocity stays up" which will lead to very interesting codebases in a few years. LLMs are a transformative technology, but it feels that management's perennial obsessions with cutting costs is going to use it mostly for that, as opposed to a way to create new products / improve existing ones.
2031 is in 5 years. There will be plenty of seniors even if the influx of juniors is exactly 0. Not saying 0 influx of juniors is good, it harms young people. But it wont cause lack of seniors issue in 2031.
Seriously. I've been programming almost 30 years and I feel like I could easily have another 10 in me. I'm sure many reading this will assume my skills are deteriorating at an atrocious rate. But I'm still the lead developer on a team that leans on me heavily to figure out the hard stuff. So I can't be too far gone.
I'm 57 and have no desire to retire. I know plenty of other programmers in their 50s who aren't itching to retire.
Programming, at least for me, isn't like some factory job where you can't wait to get out at your 30 years or whatever. I've always enjoyed what I do and found it rewarding.
That is absolutely absurd. Most people, especially those outside the FAANG ecosystem and in literally the rest of the world outside of SV, are not trying to retire at fifty. Jesus.
I was a CS major in early 2000s before switching. English. However my tech abilities are what got me the career path - MS Word is powerful, few people learn how to write well, much less with business acumen.
This is now in your neighborhood; it was in mine 5 years ago but NOT blamed on AI. RFP response specialists are a “cost center” of human capital. It’s also a great entry level role to move into a VP role after 3-5 years. Learning the business, warts and all, to help win more.
The Baby Boomer generation cut the path completely. They pocketed the savings or passed it on to investors. Now it’s nearly impossible to get a decent gig in the field - 10 years minimum experience, salary about $80,000 USD maybe, except in Defense.
Go look at how many software “solutions” there are in the RFP space. Software has claimed to be able to do the job, or let one person do the work of three, for a decade now.
I’m not sympathetic to the tools created by this industry now actively realizing “the leopards are eating my face?!” type of situation. Even Senior Devs now will start to face age discrimination or wage suppression. This was written without any AI by the way. I’ve never touched a clanker, my hands are clean.
It seems like throughout human history, every piece of technology has had a complete chain of human knowledge for how it worked. No one person knows how to build an EV starting from dirt, but someone knows each step in isolation.
With LLMs, some people are claiming that the time for understanding code is over, and that leaves a gap in the chain of knowledge. I can't think of a time when such a gap has been adventageous, and can certainly think of times when it has come with risk or loss. Many of us have seen this manifest at small scale when the one person with obscure system knowledge leaves a company without adequate prior knowledge transfer.
At large scale, well, I think it will be a great time to be a dev who knows how to hand-code.
At every place I've ever worked that hired juniors, the juniors always have to leave by about 2 years, because the company will never (roughly) double their salary to what they can get on the open market.
That first year, the juniors are probably a net drain in productivity - for their work and the time other devs spend teaching them. Then the 2nd year, they're underpaid for as long as you can hold on to them. But the good ones will be out the door pretty quick.
The system always seemed kind of broken to me. But a shop without any junior devs is also kind of sad. If you hire the right person, it's rewarding to the other devs to teach them and watch them grow.
It also teaches way more to mid-levels and seniors than just grinding code would.
> I have been tracking junior software-engineering listings on LinkedIn and Indeed since late 2024.
I don't think such a short timeframe is indicative of anything at all. This data could be interesting if comparing to say before Covid or even before 2008, but as is, this is far too short of a timeframe to draw any actual conclusions. How would you know whether we're currently under hiring juniors or we were previously just over hiring and are now returning to the norm?
Read TFA these concerns are addressed.
No they're not? Also, it's AI slop.
> This isn’t a dip. It’s a structural collapse
For what it’s worth, I’ve been in companies not hiring juniors in 2016-2020, 2023-2024. My current company just hired a few, plus interns.
This is not new.
IMO: This says more about companies than the market. Companies are willing to spend more on salary rather than improving their internal processes.
Good point to a degree, but it applies to all work, not just software engineering, and even before they get into the workforce. Young people are not going to have any depth of skills or knowledge if they literally just act as a go-between for AI to do everything.
Maybe they become experts in wiring up AI and robotics?
But 5 years out, we might be into a new compute-in-memory paradigm. And we may have exceeded the complexity of the human brain with models greater than 100 trillion parameters.
We already see performance that is superhuman in task specific ways with 1-10 trillion parameters models.
In under 5 years, most small and many medium business will be handled directly by AI agents. You won't even necessarily know what kind of software they are running. For small-medium business, it may have very little or no code and just be a semi-structured database that is neutrally rendered into any needed form on-the-fly.
I feel like this is very possible eventually, but under 5 years seems unrealistic to me. The compute power is already running into limits. And a few more major problems need to be solved before agents can have that kind of autonomy to create software from scratch and run entire businesses.
Given the intense pressure to find more efficiency, 5 years is probably enough to see deployment of radical new compute-in-memory paradigms such as nitrides-based ferroelectric devices that have a few orders of magnitude more efficiency.
I think the biggest problem is jagged intelligence which I would bet would be solved largely with bigger models, which comes back to rapidly evolving hardware paradigms.
Is a corollary to this article that if you get into IT now and you manage to land a job, you'll make bank in 2031?
Deep knowledge and patience will always be rewarded. Most people give up quickly.
> Deep knowledge and patience will always be rewarded.
No. Have you every worked at a business? Sometimes they are, often they aren't especially if decisions are being made by people who don't know you.
Oh, this is talking about the junior to senior pipeline. I thought it was going to be about improvements in AI capabilities.
I think it's reasonable in 5 years that most of what I do as a senior dev (yes, even turning vague notions into concrete deliverables, giving high level ideas of tradeoffs, and so on) will also be automated away.
Does anyone have interesting blog posts on what that means? Ideas about what the new world may look like? Will existing software engineers be running "dark factories"?
In less than five years it will be clear that unaugmented humans are a bottleneck.
So many jobs as such may be limited to humans with high bandwidth BCI connected to AI, that effectively gives them an AI exobrain. But even then the AI would need to do most of the work or the human would just slow things down.
Agree strongly with the premise, but the obvious AI writing is painful.
I'm sorry for the juniors but that's more jobs for me to choose from. They will make it so somehow.
I was just discussing this yesterday. The only time when Mythical Man Month isn’t a problem is when C-suite/management needs to burn their budget up and projects are behind. Usually you’ll end with teams with too many juniors. The leads are only focused on the harder problems and dont have capacity to properly onboard juniors. So management hires contractors from Europe/Asia. The juniors offload their work and unknowns to contractors who are often also very unskilled.
This has been going on for a long time in tech. Now with LLMs you have a bunch of juniors who are unskilled and a bunch of management trying to fill the skilled spots with unskilled+LLM workers. The seniors are starting to become plagued with people who are excellent at being unskilled in the work place. They maintain shrewd hiring practices until the projects are behind enough and we arrive at the original need to hire contractors.
“Mythical Man Month” is just something people say to indicate status. No one is using it’s lessons in actuality otherwise we’d have to acknowledge over half the work force in software are a bunch of frauds delegating their work and decision making to others.
The points raised may be interesting, but the ai-slop writing here really detracts.
What is with all the articles taking a dump on AI being written with AI's assistance?
Perhaps ai bros are saying it to discourge people from reading it. Idk just a hypothesis.
I come here for the comments first, then decide if it's worth reading the article.
> The new thing is on the supply side. Companies are already reporting staff+ vacancies that take 66 days on average to fill ... The shortage is already, in 2026, exclusive to senior-level roles.
I am not sure I agree with this being caused by lack of candidates. I have 25+ years of experience, I was caught in an office closure layoff in early 2025 for the second time in my career (the last was in the mid 2010s, where it took me a few weeks to easily find a new position), took the spring off, spent the summer preparing for the leetcode hazing ritual and interviewed a lot in the fall: despite having very up-to-date skills (I always keep up-to-date personally, also my last position was all k8s, llm integration, you name it) and long stints in my resume, interviewers did not care.
In the positions where I managed to pass the automatic resume screening (which was probably 10% of the time it felt), the interviewers just did their typical 5-rounds-sudden-death interview loops, with sudden death happening if the solution was not optimal. Sometimes sudden death was "oh I see you have experience in cloud 1 and cloud 2, but we run on cloud 3, sorry", other times it was a friday afternoon slot where the interviewer obviously just wanted to be done with the week.
It also did not help that many positions were "remote but only if you live within 50mi of one of our offices", and others were trying their best at downleveling (I have 15 years of my 25+ at principal+ both architect and IC, and some recruiters were pitching me senior positions, as in senior senior, not senior staff / senior principal). Often the hiring mgr would bemoan the fact that so many resumes are fake, and so many candidates just outright lie, but none of this seemed to make a difference when it came to the interview setup.
Ultimately after about 3-4 months of searching (I'd say I applied to ~100ish positions where I thought I'd be a good fit, and had actual interviews in less than 10) I ended up finding a local position for a significant pay cut, it is also the first job where the mandate from high on up is AI-everything, velocity first (like it's mentioned in this article). Given my long time experience I am able to be very productive with claude (because I know what it should be doing) but even with really tight reins it is a daily occurrence for it to prioritize expediency over correctness requiring rework.
I honestly don't know how people starting with AI agents will manage to improve, I have been through the transition of assembly->higher level languages, and the "spec driven agentic is just like using a compiler" rings very superficial to me: nobody would accept a compiler that 5% of the time (being generous here) miscompiled your code.
The current push seems to be "it's fine to have juniors on claude code producing tons of PRs, we'll just put a claude / codex reviewer action in GH so velocity stays up" which will lead to very interesting codebases in a few years. LLMs are a transformative technology, but it feels that management's perennial obsessions with cutting costs is going to use it mostly for that, as opposed to a way to create new products / improve existing ones.
LLM written slop article about LLM slop in bigtech
2031 is in 5 years. There will be plenty of seniors even if the influx of juniors is exactly 0. Not saying 0 influx of juniors is good, it harms young people. But it wont cause lack of seniors issue in 2031.
There will be a lot of super seniors on the doorstep of retirement, but no seniors.
How long do you think careers last?
Seriously. I've been programming almost 30 years and I feel like I could easily have another 10 in me. I'm sure many reading this will assume my skills are deteriorating at an atrocious rate. But I'm still the lead developer on a team that leans on me heavily to figure out the hard stuff. So I can't be too far gone.
People are looking towards retirement by the time they are 50 usually
I'm 57 and have no desire to retire. I know plenty of other programmers in their 50s who aren't itching to retire.
Programming, at least for me, isn't like some factory job where you can't wait to get out at your 30 years or whatever. I've always enjoyed what I do and found it rewarding.
Ah hah hah. Wouldn’t it be nice to have that option?
That is absolutely absurd. Most people, especially those outside the FAANG ecosystem and in literally the rest of the world outside of SV, are not trying to retire at fifty. Jesus.
I was a CS major in early 2000s before switching. English. However my tech abilities are what got me the career path - MS Word is powerful, few people learn how to write well, much less with business acumen.
This is now in your neighborhood; it was in mine 5 years ago but NOT blamed on AI. RFP response specialists are a “cost center” of human capital. It’s also a great entry level role to move into a VP role after 3-5 years. Learning the business, warts and all, to help win more.
The Baby Boomer generation cut the path completely. They pocketed the savings or passed it on to investors. Now it’s nearly impossible to get a decent gig in the field - 10 years minimum experience, salary about $80,000 USD maybe, except in Defense.
Go look at how many software “solutions” there are in the RFP space. Software has claimed to be able to do the job, or let one person do the work of three, for a decade now.
I’m not sympathetic to the tools created by this industry now actively realizing “the leopards are eating my face?!” type of situation. Even Senior Devs now will start to face age discrimination or wage suppression. This was written without any AI by the way. I’ve never touched a clanker, my hands are clean.
[dead]
[dead]