An L6 manager and an L6 IC do different work though. And it's rarely (read: never) the case that the person in charge is most knowledgeable about all the details.
If an L6 manager could keep all of the details of all of the things all their reports are working on organized, they aren't handling enough scope. Imo that's the difference between 5 and 6. You can no longer track all the details in one person's head.
> If an L6 manager could keep all of the details of all of the things all their reports are working...
That's not what they said, is it? There is a huge difference between what they said and how you interpreted it. They said "you need to be able to build the software and review diffs and write serious ones now and again" and "it’s very difficult to manage a process that you don’t understand with some sophistication" (emphasis mine). How do you get 'know every little detail about everything' from that?
It seems obviously true to me that any manager who can't "build the software", "review diffs", and "write serious [diffs] from time to time" is useless. Anyone who disagrees with this is probably dead wood, spends their time fighting political wars about issues they don't understand, and their team is more likely than not constantly fire fighting and in serious trouble (in my opinion).
If I had a dollar for every time some clueless manager lectured me about how to write software, and I rolled my eyes and ignored them, and they never found out (because they had no idea what was going on to begin with, and were just posturing based on something someone said) I would have at least $100. I obviously have a chip on my shoulder here, but managers who barely know what is going on still tend to want to 'contribute', which of course they can't do in reality, so they end up playing keyword matching and 'helping' the team avoid 'duplicate effort'. For example, they will see one team building something, and another team building something, and notice some of the words are the same, and then make a big show of 'avoiding duplicate work', but in reality the use cases are extremely different and nothing was being duplicated. Once they have alienated enough of the team, people just start telling them nonsense and they have no way to know it. Productive, smart people start to leave or lose motivation. A fast pace is replaced with constant excuses and a team that is basically no longer showing up for work. The manager can't tell the difference. This is what managers who aren't in the details are like, pretty much without exception.
You seemed to skip their second paragraph, which is what I was replying to.
Someone who can review diffs and even write a serious change every now and again isn't the most knowledgeable person on the team. The person who is writing and reviewing all of the serious diffs is. And for many L6 EMs, you'll have three of those people reporting to you, they're all working on different projects, each of which you have partial but imprecise knowledge of (and sometimes, very little because L5s and FAANG are expected to be able to operate mostly independently). So you spend your time ensuring that everyone on the team has career growth opportunities, and that your less experienced people have mentorship, and hiring and arguing about headcount allocation and prioritization, which all matter, but which are all things that most engineers don't want to touch!
Yes, managers should still have engineering experience (at least up to like director/vp levels, where IDK maybe they don't need it) but having a baseline knowledge of how to do a software is not the same as being the most knowledgeable person on the team.
I made a concrete statement about what I think the minimum technical bar for EMs is up to at least L7. I also remarked that "clearly this isn't your main focus" at that seniority, which you seemed to skip.
The second paragraph, the one you seem to have an issue with, a combination of 2 things:
- a few examples of extremely senior, extremely successful engineering leaders who stayed at or near the top of the game technically, and those are but a few examples from a very long list
- an observation that in other fields, law for example, they call in the highly-knowledgable, highly-expensive, person capable of solving problems few in anyone else in the organization can and that this person carries titles like "partner" and makes the most money.
I know this is a touchy subject and I've been trying to be less flamey on HN so I didn't go hard like the GP, but they're fundamentally right even if the language is a bit intemperate: there is a prestigious and important job track for people who are pretty damned technical/quantitative but not wildly hands-on and generally concerned as much or more with coordination and communication, particularly in cross-functional or externally-facing scenarios than software systems per se: product manager. As a wild oversimplification: when an EM becomes senior enough they end up as a CTO, and when a PM becomes senior enough they end up as a CEO. This is natural and healthy division of labor.
An EM is concerned first and foremost with the health, happiness, and therefore productivity of engineers. On the foundation of the trust and rapport and deep knowledge that comes from that kind of engagement with their team, they are able to also be concerned with how their team fits into the bigger business picture: is this the right team for the needs of the business, what hiring and performance management would be necessary to make it so if not, what is a realistic schedule for the work that needs to be done given the strengths and weakness of the team members both generally and at this moment in time?
When I'm wearing my hacker hat I have no interest in reporting to an EM who couldn't do my job in a pinch, I might respect that person on a lot of levels but I won't be interested in their opinion of how I should do my job. And at no time in my career has it been so easy to identify such managers: they are the "back to the office damn the torpedos" crowd. When the task tool, and the code review tool, and the oncall/incident situation, and the build wiki are not sufficiently comprehensible to an EM to form an opinion of who is doing a good job, the instinct to do "ass-in-seat" performance evaluation is strong, the instinct to be "visible" is strong, and narrative that there's value add looks very threadbare over Zoom.
This bloc is probably too big and too entrenched to dislodge, but WFH for 2 years working out just fine is the best chance we're going to get. IMHO.
> - a few examples of extremely senior, extremely successful engineering leaders who stayed at or near the top of the game technically, and those are but a few examples from a very long list
But all of these are the exception, not the norm. You said you've been in a FAANG style org, so you've been able to view the org-chart. For every Jeff Dean or John Carmack, there's three-dozen directors and VPs who manage large orgs whose names you've never heard of and who haven't checked in code in half a decade. They're still usually very good managers.
(A reasonable opinion I've seen, btw is that if you're managing a team of say 5 or more people, if you have time to make regular code contributions,you probably aren't focusing enough on your other managerial responsibilities and are letting your reports down)
> - an observation that in other fields, law for example, they call in the highly-knowledgable, highly-expensive, person capable of solving problems few in anyone else in the organization can and that this person carries titles like "partner" and makes the most money.
This is pseudo-true. Partners are called partners not because they're capable of solving problems no one else can, but because they carry an ownership stake, and I'm not a lawyer (and presumably neither are you) but the impression I get is that law is far more delegatory than software, where a partner may call in some favors to address a problem, but will also have 20 Jr. associates investigate 20 different possible approaches and write 20 different briefs to then choose between.
> When I'm wearing my hacker hat I have no interest in reporting to an EM who couldn't do my job in a pinch,
Maybe we have different definitions here, but I've never, or perhaps once, had a manager who I felt could do my job in a pinch, and I was a new grad and he was the worst manager I've had at the time (although he very quickly got better as he learned to be more hands off). A manager who could do my job in a pinch is too micromanaging. Maybe you mean something different in that they could, given a week or two to turn-up on the project take over, but that's not really what I'd consider "in a pinch".
> Maybe we have different definitions here, but I've never, or perhaps once, had a manager who I felt could do my job in a pinch,
The chief of Air Force still flies the fighter jets just as well as the junior pilots. The head of surgery in a hospital still operate just as the junior surgeon. Why does tech have to be different?
An air force officer may have to qualify on their aircraft each year, but they're unlikely to be "the most experienced" fighter pilot, and it's common in the military for enlisted people to make fun of officers for being too abstracted away from the life of a soldier.
> The head of surgery in a hospital still operate just as the junior surgeon. Why does tech have to be different?
Because the load we put on medical personal is abusive and we shouldn't copy it in tech?
You've got some reasonable points and I don't just want to completely gang-tackle you here, but there is a part of your argument that I think is sufficiently incorrect in a sufficiently harmful way that I'm going to sort of keep at you about it, in what I hope is an open-minded or at least respectful way.
Some people like the book "Coders at Work" (I do), some do not, but almost anyone would agree that the people interviewed are absolute luminaries [0].
There are 15 chapters comprising 15 interviews. You have to get to #6, arguably #7 before you find someone who at the time of writing wasn't both a world-renowned hacker and currently or at one time a demonstrably successful engineering leader. The first 5 being: jwz, Brad Fitzpatrick, Douglas Crockford, Brendan Eich, Joshua Bloch. It starts to get a little blurry in the second half because it's so thick with CS academics (who in a different way also do engineering management), but you've still got VP-types who still code like Peter Norvig. The ~50% who aren't demonstrated engineering leaders are super hard-core CS researchers like Donald Knuth. The book doesn't even interview Cliff Click, or John Carmack, or talk about the fact that Larry Page and Sergey Brin wrote the first version of Google themselves and continued maintaining parts of it well into hyper-growth. Eric Schmidt wrote `lex`. When Jack Dorsey was recruiting me for Square over lunch he made an incredibly eloquent argument about why he uses OCaml rather than Haskell for his personal hacking.
At the time I was an L7 EM, my L9 boss didn't have much time to write code, but asked probing questions about everything from the merits of various binary classifiers given imbalanced underlying Bernoulli distributions to the algebraic properties of the data structures we were using for distributed systems convergence.
I don't dispute that plenty of successful leaders in technical organizations have become rusty as hackers when they hit the mega-seniority, but the idea that some L6/L7 manager shouldn't be able to lift some of their team's serious code off the ground, let alone some undergraduate dynamic-programming interview question as was the original point of my original post is simply contradicted by a mountain of evidence both generally-available and anecdotal to numerous people in this thread.
You can get ahead as a mid-level EM without knowing the frib-frobs from that whatsits, but God-willing I'll never work for one again. That's a visibility game, it's a popularity game, it's a schedule-too-many-meetings game, it's a post-too-much-on-the-internal thing game. Fuck that game.
> I don't dispute that plenty of successful leaders in technical organizations have become rusty as hackers when they hit the mega-seniority, but the idea that some L6/L7 manager shouldn't be able to lift some of their team's serious code off the ground, let alone some undergraduate dynamic-programming interview question as was the original point of my original post is simply contradicted by a mountain of evidence both generally-available and anecdotal to numerous people in this thread.
I think this is the root of the disagreement. I want my managers to be technically capable with an engineering background. As far as I know, my entire leadership chain, arguably excepting my SVP and CEO (at Google) have such a background. Yes they should be able to pass a tech screening and certainly system design.
They should have enough knowlege to know why you're making certain technical decisions and be able to see that you're justifying them well. That's all true. What I'm saying is that a manager who is staying so deeply aware of the details of all of their reports work that they can drop in and take over in a pinch probably has too much context. I'm not making an argument on abstract technical ability or knowhow.
Like in the same way that my entire management chain has technical knowlege and background, practically none of them have submitted code at work in the last 3-4 years at minimum, for some that's their entire career at Google. I expect that, with enough effort they could all work on the projects I work on and make contributions, but they'd need to learn all of {language and codestyle, libraries and relevant tactical design patterns and norms, various technical constraints to the design that are important but nonobvious}, so in most cases it would be similar to onboarding someone who was a technical contributor from another team or an experienced new hire. I don't really consider some random employee of Facebook to be someone who could "do my job in a pinch" either, even if they're more technically competent than I am.
If an L6 manager could keep all of the details of all of the things all their reports are working on organized, they aren't handling enough scope. Imo that's the difference between 5 and 6. You can no longer track all the details in one person's head.