“Lieutenant” is etymologically “placeholder”, and was originally used for top-level underlings; in a sense a “Lieutenant” is that for a Captain, a “Lieutenant Colonel” is that for a Colonel, a “Lieutenant General” is that for a General (or, in some systems, a “Colonel General” – in which case you can think of a Lieutenant General being a Lieutenant Colonel General, in which case it falls exactly where you’d expect in relation to a Major General, given how Major is just below Lieutenant Colonel.)
The context here is that at Meta there is a perverse incentive to create management roles where none are needed.
I can say with direct knowledge that some orgs within Meta are insanely vertical. To the point where one has to wonder what all the middle management types are even doing. They are about to get Zucked.
You mean the meta point? :) The same applies to a lot of companies like Amazon, Microsoft and Salesforce etc I'd think it would be more optimal to prune at these middle layers and reassign the leaf nodes, instead of pruning at the nodes if a company wants to has overhired.
So does he want a smaller company or a wider management span at each level? Because managers managing managers is a natural consequence of sane limits on span of supervision plus a large organization. At 15 direct reports (around what is typically cited as a reasonable upper bound), maxing out every manager, you reach (not counting the CEO as top manager!) managers managing managers at 15^3 = 3375 total employees. At 50,625 employees you’ve fully saturated another level of managers. Meta is around 70k employees.
And those numbers all assume that all direct reports of a N-level manager are N-1 level managers, where often they’ll have some N-1 level managers, some (if they aren’t at level 2) lower-level managers, and some non-managers, which makes the total number of employees smaller for any given number of levels.
I think what he means is he wants to avoid the situation of management chains of like 5 or more. Your example shows a large company should ideally only need 3 or so levels of management, yet there are plenty of long, inefficient chains everywhere.
The math here is useful, but I think there's typically a subdivision by ladder as well, so you have product, (technical) project managers, eng, etc. each report up through separate department heads who report to the CEO. That creates another "level" but I assume having SVPs and/or C-level executives doesn't count as "managers managing managers".
FWIW, regarding staffing changes and recent layoffs, I am very biased as an engineer who still writes code but I do agree with the sentiment from that article and think there's a lot of bloat at the top of large tech companies. Seems like there's significant cost savings left on the table if you're not laying off senior managers and directors of nebulous realized value while laying off ICs who did the actual concrete work and dealing with the attrition, knowledge gaps, etc. associated.
> The math here is useful, but I think there's typically a subdivision by ladder as well
Yeah, in a real org:
* You don't want every manager at the upper limit of span of control,
* Staffing levels for different functions don't line up for a perfectly equal distribution,
* Not every direct report of a second-or-higher level manager is a manager of the next level down, some are individual specialists or managers of small teams that support the higher-level manager. Often, even with 10-15 direct reports, you'd only want 3-5 full-depth subordinate units, with the remainder being specialists and small units supporting the org. So, realistically, depth should be significantly greater than that math, which is deliberately a limit case.
I’m sure there are plenty of org inefficiencies that need to be restructured or eliminated. Early in my career, I would simply nod along to a headline like this. Now, I realize any single individual lacks the imaginative capacity to accurately know how much stuff large orgs naturally end up doing in service of even a limited set of simple goals. A limited number of simple goals at a large enough scale require an explosion of seemingly unimportant work. That’s just entropy.
This is a pretty bad headline. The article quotes him as saying “I don't think you want a management structure that's just managers managing managers, managing managers, managing managers, managing the people who are doing the work.” Zuck doesn’t actually think Meta could get rid of everyone between him and the ICs, like the title makes it sound.
23 comments
[ 3.1 ms ] story [ 64.4 ms ] threadA Major General is a Major to the Lieutenant to the General.
This actually makes sense.
I’m guessing the answer to “why do the ranks not line up among branches of service” is “history”?
“I’m a Captain!” “Army or Navy?”
I can say with direct knowledge that some orgs within Meta are insanely vertical. To the point where one has to wonder what all the middle management types are even doing. They are about to get Zucked.
I suspect the decision to prune leaf before branch nodes was strategic and intended to expose dead branches.
Unfortunately, coordinating directions between 10,000 people means that you have managers managing managers somewhere.
Maybe General AI will fix this some day, and executive decisions will directly be translated to everyone in exactly their job function relevance.
And those numbers all assume that all direct reports of a N-level manager are N-1 level managers, where often they’ll have some N-1 level managers, some (if they aren’t at level 2) lower-level managers, and some non-managers, which makes the total number of employees smaller for any given number of levels.
FWIW, regarding staffing changes and recent layoffs, I am very biased as an engineer who still writes code but I do agree with the sentiment from that article and think there's a lot of bloat at the top of large tech companies. Seems like there's significant cost savings left on the table if you're not laying off senior managers and directors of nebulous realized value while laying off ICs who did the actual concrete work and dealing with the attrition, knowledge gaps, etc. associated.
Yeah, in a real org:
* You don't want every manager at the upper limit of span of control,
* Staffing levels for different functions don't line up for a perfectly equal distribution,
* Not every direct report of a second-or-higher level manager is a manager of the next level down, some are individual specialists or managers of small teams that support the higher-level manager. Often, even with 10-15 direct reports, you'd only want 3-5 full-depth subordinate units, with the remainder being specialists and small units supporting the org. So, realistically, depth should be significantly greater than that math, which is deliberately a limit case.