I've never worked a place where I've liked the implementation of engineering managers. Often it feels like the EM is both the tech lead / architect and your manager, which conflates things too much. Also they're often too busy to do either part good enough.
I've much preferred when the managerial role is outside my team, or at least a different person. Makes it easier to discuss things with them.
I'm not saying I don't value their work. Just that I don't like the way of putting multiple responsibilities onto one person.
At the last job I was managing 14 people (about 40% of the project), baby sitting the dysfunctional project manager, developing, and providing QA support for after hours deployments. I don’t miss it. I had too many bosses each with too many diverging goals.
My number one goal when looking for the next job is to identify immaturity early. The result is I am still unemployed.
Reading Meta wants a 1:50 ratio is wild. Let’s put some simple numbers down.
I like a 30 minute 1-on-1 per engineer per week. At 1:50, 25 hours out of my 40 are now 1-on-1s.
Assume a usual team is 5 engineers, so I’m covering 10 teams, effectively.
So I have 15 hours to keep context of 10 teams. Am I expected to contribute as an IC, too? Usually when I’m directly leading a team, putting code into prod is how I feel the pain.
Finally: am I getting 10x the salary for doing the job of 10 EMs?
I’m all for cutting bureaucracy but is AI really this magical?
Not that I'm defending that ratio, but I can't imagine if they have a 1:50 ratio, they're expecting to do 1:1s every week. Maybe do it every 2 weeks. Now you've got half your time back.
For better or worse AI also enables unprecedented levels of surveillance of your team. You can churn through what every single one of those 50 people is doing every day, what they're actually delivering, what their working habits are, what conflicts they're having with people, etc.
You can churn through all that, but I'm not sure you'll understand the people in a relevant way. At this ratio it's more like a warden or evangelist role or so, rather than actual people management.
Assuming you're not doing project status updates and you're actually talking about them rather than the work, that seems like quite a lot of dedicated time. I do half that with my reports.
I don’t necessarily disagree on the ratios but I think that’s because we’ll see more software companies overall rather than because companies will engage mass layoffs.
I used to work in an organization where a senior director had 2 direct reports, one of them being a director who had 2-3 direct reports (senior managers), and these senior managers had maybe 3-4 managers each.
Getting anything done was a nightmare because everyone up to the senior director level wanted to have an impact on the decision, so the approval process took ages.
When done right, the flattening is good (and I say it as a middle manager) - it should remove layers that don’t add value, and empower people at lower levels. In this case removing the senior director or director wouldn’t hurt the organization.
Unfortunately what I often see is that an organization will reduce number of people in one layer from 3 to 2, increasing the number of direct reports, but keeping the same number of layers leading to no improvements (besides cutting costs)
The article is from Sep 1, so still before Opus 5.5 release date, and I guess the mindset of authors was formulated long time before Opus 5 release, which I consider the turning point of software engineering industry. Things changed a lot since then, and it was a point of no return.
The business model changed since then, you no longer need a small army of devs to solve problems, we no longer need to play LARP of Prussian army to work in the industry.
Article says someone is attacking the role, but the brutal truth is different: the demand for such service is decreasing rapidly.
Pretty typical for a manager; out of date and irrelevant. Probably had to wait 2 months for approval from the hierarchy of bureaucracy. When the team can be agile on its own, the management process gets left in the dust.
I'm most certainly working for a less happy company where managers insist on a lot of reporting where already understaffed teams spend their time filling and updating jira instead of doing the work to get a project done. Not sure if it has anything to do with managers, but no matter the amount between, there finally has to be more trust again. Whenever I have to speak to a manager, I'm met with disinterest in what has to be done, how and how much it costs. Seems like its best to do nothing and produce no costs except a wage. There is always some sort of haggling and personal things going on as well that should be far away from building a working system.
At this point I believe that even if we had AGI/ASI, we'd still hit our head on the decision making processes, since it seems to be an interpersonal people problem instead of simply providing solutions to very real problems people pay for.
Even former engineers who move to such management roles change their frame of mind with enough exposure to the upper layers and they forget what it all is about.
"One of the things managers repeatedly ask me about in workshops is: “How do I keep the team together when they prefer to talk to AI rather than talk to each other?”
Now that's a new problem. Usually articles like this re-hash classic problems.
(I recently had to chat with customer support for a retailer. The other end was human, and one of the better AIs would have been far more useful.)
Developers do sometimes think they can do without managers until, as the article mentions, there's anything that needs deciding or a "people problem". It does make me laugh a bit that at each level there is a non-understanding about the adjacent levels such that destructive changes seem good at first glance.
IMO it is valuable for people to occasionally do things out of their own area to generate that understanding but what business would ever allow that?
Perhaps developers can decide on their own? In my experience most managers are major bottle neck, and just "diversity hires". They are not capable of comprehending the problem, let alone deciding.
I so much prefer to just exchange markdown folders with other developers and their agents. After a few rounds we get decisions and can move on. No weeks and weeks of meetings needed!
This is taking past the parent comment and shows that parts of the divide are on the engineering side. Your point is valid and definitively weighs much, but I don't understand how engineers don't reflect what their contribution in this is and I am sad for all who haven't worked with good people managers. Those really elevate an organization to a next level. Seen both cases (good and bad managers). Similar for product managers. Engineers think they need to build. Period. But the sentence continues with "for customers" and they can't understand customers and their problems at all.
> Engineers think they need to build. Period. But the sentence continues with "for customers" and they can't understand customers and their problems at all.
And managers are working for customers? In my experience they only focus on revenue, and squeezing out more out of customers. The MBA/management culture is cancer that is eating industry. With AI developers can simply bypass it!
As engineer i got in troubles for adding minor features or improvements, customers asked for. It was planned as next "major" release/contract for tons of momey.
Most will because you will simply not need to hire as much people. I used to run a tribe of 4 teams to do something I'm today doing with me and another guy. These are real projects that have better outcomes and get done faster with 2 people vs 4 teams (20 people) because this way we also have no coordination or communication problems. And I believe I'm a better manager than I am an engineer, so this isn't even bragging about anything, its just what it is.
I've never worked a place where I've liked the implementation of engineering managers. Often it feels like the EM is both the tech lead / architect and your manager, which conflates things too much. Also they're often too busy to do either part good enough.
I've much preferred when the managerial role is outside my team, or at least a different person. Makes it easier to discuss things with them.
I'm not saying I don't value their work. Just that I don't like the way of putting multiple responsibilities onto one person.
At the last job I was managing 14 people (about 40% of the project), baby sitting the dysfunctional project manager, developing, and providing QA support for after hours deployments. I don’t miss it. I had too many bosses each with too many diverging goals.
My number one goal when looking for the next job is to identify immaturity early. The result is I am still unemployed.
Reading Meta wants a 1:50 ratio is wild. Let’s put some simple numbers down.
I like a 30 minute 1-on-1 per engineer per week. At 1:50, 25 hours out of my 40 are now 1-on-1s.
Assume a usual team is 5 engineers, so I’m covering 10 teams, effectively.
So I have 15 hours to keep context of 10 teams. Am I expected to contribute as an IC, too? Usually when I’m directly leading a team, putting code into prod is how I feel the pain.
Finally: am I getting 10x the salary for doing the job of 10 EMs?
I’m all for cutting bureaucracy but is AI really this magical?
Not that I'm defending that ratio, but I can't imagine if they have a 1:50 ratio, they're expecting to do 1:1s every week. Maybe do it every 2 weeks. Now you've got half your time back.
For better or worse AI also enables unprecedented levels of surveillance of your team. You can churn through what every single one of those 50 people is doing every day, what they're actually delivering, what their working habits are, what conflicts they're having with people, etc.
You can churn through all that, but I'm not sure you'll understand the people in a relevant way. At this ratio it's more like a warden or evangelist role or so, rather than actual people management.
I like a 30 minute 1-on-1 per engineer per week.
Assuming you're not doing project status updates and you're actually talking about them rather than the work, that seems like quite a lot of dedicated time. I do half that with my reports.
They’re delusional (and will soon pay the price for their hubris), just like Microsoft
Obviously, they did let Zuckerberg get away with the Metaverse idea for 10 years ;-)
Well on the other hand only these megacorps will need managers because for 99% of other software companies, 10 people will be enough.
I don’t necessarily disagree on the ratios but I think that’s because we’ll see more software companies overall rather than because companies will engage mass layoffs.
I used to work in an organization where a senior director had 2 direct reports, one of them being a director who had 2-3 direct reports (senior managers), and these senior managers had maybe 3-4 managers each.
Getting anything done was a nightmare because everyone up to the senior director level wanted to have an impact on the decision, so the approval process took ages.
When done right, the flattening is good (and I say it as a middle manager) - it should remove layers that don’t add value, and empower people at lower levels. In this case removing the senior director or director wouldn’t hurt the organization.
Unfortunately what I often see is that an organization will reduce number of people in one layer from 3 to 2, increasing the number of direct reports, but keeping the same number of layers leading to no improvements (besides cutting costs)
I’ve seen a director with no reports.
The article is from Sep 1, so still before Opus 5.5 release date, and I guess the mindset of authors was formulated long time before Opus 5 release, which I consider the turning point of software engineering industry. Things changed a lot since then, and it was a point of no return.
Good as it is, I don't see opus 5.5 replacing managers
The business model changed since then, you no longer need a small army of devs to solve problems, we no longer need to play LARP of Prussian army to work in the industry. Article says someone is attacking the role, but the brutal truth is different: the demand for such service is decreasing rapidly.
What does that have to do with the article?
In fact, what are you trying to say? Seems like a generic response you could try to post in every submission.
Haven’t you noticed that after the Opus 5.5 release, colours are more vibrant, interactions more pleasant and turds more pungent?
Anthropic astroturfing? That marketing budget has to be spent somewhere.
Pretty typical for a manager; out of date and irrelevant. Probably had to wait 2 months for approval from the hierarchy of bureaucracy. When the team can be agile on its own, the management process gets left in the dust.
Sorry, managers, for being unnecessarily harsh.
I'm most certainly working for a less happy company where managers insist on a lot of reporting where already understaffed teams spend their time filling and updating jira instead of doing the work to get a project done. Not sure if it has anything to do with managers, but no matter the amount between, there finally has to be more trust again. Whenever I have to speak to a manager, I'm met with disinterest in what has to be done, how and how much it costs. Seems like its best to do nothing and produce no costs except a wage. There is always some sort of haggling and personal things going on as well that should be far away from building a working system.
At this point I believe that even if we had AGI/ASI, we'd still hit our head on the decision making processes, since it seems to be an interpersonal people problem instead of simply providing solutions to very real problems people pay for.
Even former engineers who move to such management roles change their frame of mind with enough exposure to the upper layers and they forget what it all is about.
I can't tell if this comment is satire.
"One of the things managers repeatedly ask me about in workshops is: “How do I keep the team together when they prefer to talk to AI rather than talk to each other?”
Now that's a new problem. Usually articles like this re-hash classic problems.
(I recently had to chat with customer support for a retailer. The other end was human, and one of the better AIs would have been far more useful.)
Developers do sometimes think they can do without managers until, as the article mentions, there's anything that needs deciding or a "people problem". It does make me laugh a bit that at each level there is a non-understanding about the adjacent levels such that destructive changes seem good at first glance.
IMO it is valuable for people to occasionally do things out of their own area to generate that understanding but what business would ever allow that?
Perhaps developers can decide on their own? In my experience most managers are major bottle neck, and just "diversity hires". They are not capable of comprehending the problem, let alone deciding.
I so much prefer to just exchange markdown folders with other developers and their agents. After a few rounds we get decisions and can move on. No weeks and weeks of meetings needed!
This is taking past the parent comment and shows that parts of the divide are on the engineering side. Your point is valid and definitively weighs much, but I don't understand how engineers don't reflect what their contribution in this is and I am sad for all who haven't worked with good people managers. Those really elevate an organization to a next level. Seen both cases (good and bad managers). Similar for product managers. Engineers think they need to build. Period. But the sentence continues with "for customers" and they can't understand customers and their problems at all.
> Engineers think they need to build. Period. But the sentence continues with "for customers" and they can't understand customers and their problems at all.
And managers are working for customers? In my experience they only focus on revenue, and squeezing out more out of customers. The MBA/management culture is cancer that is eating industry. With AI developers can simply bypass it!
As engineer i got in troubles for adding minor features or improvements, customers asked for. It was planned as next "major" release/contract for tons of momey.
The article makes a case that EMs do useful work but fails to make the case for why EMs can't be replaced by AI.
Most will because you will simply not need to hire as much people. I used to run a tribe of 4 teams to do something I'm today doing with me and another guy. These are real projects that have better outcomes and get done faster with 2 people vs 4 teams (20 people) because this way we also have no coordination or communication problems. And I believe I'm a better manager than I am an engineer, so this isn't even bragging about anything, its just what it is.