Hacker News new | past | comments | ask | show | jobs | submit login
Ask HN: What do middle-level managers do?
118 points by vagab0nd 18 days ago | hide | past | web | favorite | 85 comments
I'm an individual contributor at one of the big software companies (FANG). I'm curious about what middle level managers do in these companies. Specifically, my manager's manager, the ones between the CEO and first level management. They seem busy all the time (they are usually not at their desks). But they are never in any meetings. They never send any emails. And they don't seem to affect the direction of projects in any way. And I'm just too scared to talk to them. So asking here: what do they do everyday?



> I’m just too scared to talk to them.

No reason to be! They’re just another human being trying to do their best for your company and team. Assume best intentions!

If you’re nervous because you have no idea what to say to broach the topic, here’s something you could try.

“Hi, I’m Charlie. I work for Susan over on the infrastructure team. As part of my own professional development, I’m trying to build out an understanding of how all of the people at our company move towards our common goal, as it’s not always obvious to me at my level. Do you have time for a quick one on one where you can share with me your roles and responsibilities, and what some of your challenges are here at our company?”

Basically, you don’t need to say something as blunt as “What exactly is it that you do here?” Instead, just come at it from a position of positive inquiry where you’re just genuinely trying to learn and grow.

I say all this because no answer here is going to really tell you what that person does. Everyone’s roles and responsibilities vary, regardless of their title. You won’t learn unless you learn how to ask!


Maybe it's just me, but it's always been my experience that it's a better career move to A) always avoid using fancy business-isms outside of meetings, and really whenever possible and to B) try to get information from people first by making them like you as a person and comfortable around you, then asking.

It's been my experience that most people find "HR speak" to be scary, pretentious big words, kool-aidey, or just hard to decipher. But I don't work in FAANG so I just wanted to put this out there to see how everybody else feels.


Fair enough. My language when I’m doing writing tends to sound a bit over the top because I’m trying to use super generic language (given I have no context about someone’s role).

How’s this:

“Hey! I’m John on the infrastructure team. I’m still fairly new here, and I’m trying to gain a lot of context around the different teams and team members. Any chance you’re free for a quick 1:1?”

Then, in the 1:1, probe them a bit more about how they fit into the team.

“Oh. That’s cool! I didn’t realize your team took care of that. How does your role fit into that work?”


Fully agree. And this is actually a well known fact. People following the rules perfectly as seen as losers in business. You are expected to break the ice with a more personal non textbook approach and therefore showing you are "street smart".


Based on what I know and have heard about middle managers, I don't have a sound basis for assuming best intentions.


At my company I feel pretty comfortable just asking "Yo, what's your job?". Why not? Would you be offended by someone else's curiosity?


> Would you be offended by someone else's curiosity?

Me personally? No. Do I know others who would be? Yes.

Always a tricky line trying to be thoughtful about how someone might react to a particular question, especially when you're just genuinely trying to learn something about them. But, given this person is likely a relatively unknown party, not having any trust built up means that there can be misunderstandings even from simple questions like, "Yo, what's your job?"

That being said, if your company culture is such that everyone you hire responds well to that style of question, great! Certainly no problem on my end with that type of questioning. I merely was being a bit more elaborate in my earlier response to account for someone not necessarily taking something that brief, well.


Do you and they not have a baselevel implicit trust simply due to being coworkers?

Do people some people think it's appropriate to guffaw at simple questions like, "What do you work on here?"

If somebody were offended by that question, I would imagine it is not due to any overreach on my part. It would take an especially megalomanical (or highly confidential) work environment where superiors would dismiss work related inquiries outright.

I can imagine in other countries with clear hierarchies, this might be different. But I imagine even in such places it can be advantageous to be social and informed in the work place.


> Do you and they not have a baselevel implicit trust simply due to being coworkers?

In some cases yes, in other cases, no.

As an example, say you're a product manager in Department A. You, previously, had to work with Software Team B on a deliverable, and it was a really rough experience. The team was curt, challenging, and often spoke down to you because of your role.

There's some turn-over on Team B over the year, and there are a few new hires. You know of them, but don't really know them at all, as you no longer interact with that team. One of the new hires walks up to you at lunch one day and says, "Yo, what is it that you do here?"

Now, a rational approach is to just take that person as just being curious. Assume best intentions, assume they're just curious, assume it's very genuine. In that brief moment though, your frustrations with Team B surface up, and you react, rather than act rationally. You assume that someone on the team was bad-mouthing you, as you had heard the team lead previously say "I don't know why we even have product managers here anyway." Now this new hire is coming over, asking what it is that you even do, and now you've got to deal with this stuff all over again...

Or, perhaps someone who is from my generation, hearing "What is it that you do here", would just immediately think of Office Space, and immediately enter into a "defend your role" sort of mentality, rather than just taking it as some honest inquiry.

Just to reiterate. I'm not saying one should react the way I'm describing above. But, I will personally say, I don't always react to questions in a way that's healthy. How a question is worded to me can make or break my response. So, whenever possible, especially when giving generic advice on HN, I try to give suggested phrases that will hopefully avoid any unnecessary reactions, within reason. Obviously, one can't be possibly remotely aware of the challenges each and every one of us have had in our lives, and so misunderstandings are inevitable- But, thoughtful phrasing can go a long way and help to establish great relationships over the long term.


* Do you and they not have a baselevel implicit trust simply due to being coworkers?*

Depends highly on the context. Highly dysfunctional organizations make it easy to be distrustful of coworkers.


I must admit that when a complete stranger's first interaction with me is a breezy "Yo" followed with a blunt request for me to do something for them, I am not as predisposed in their favour as I might be if they had chosen other wording.


I know you’re downvoted, but I agree with this 100%.

Technical folk usually are fascinating people but more often than not (in my experience) are introverted. It’s a shame, because as an extroverted passionate engineer, it’s amazing to see, build and play with abstractions above software. Stepping outside your bubble in a not-timid way provides access to this.


I can't speak for the downvoters, but I don't think people are implying "Don't step outside your bubble in a not-timid way", but rather, "There's probably a better way to phrase this question in a way that's less likely to be misinterpreted."

So, I 100% also agree with the sentiment! Ask people what they do. Be curious. Learn from others (that's why we're all on HN right?)! Try to step outside your comfort zone when you can. Good things will happen more often than not.

Just, be mindful of how you phrase questions. That's my primary point.


“It’s not what you said, it’s how you said it.” At least that HR gobbly-gook is less prone to misinterpretation. (Works for sexual harassment too)


Working with other middle managers to align resources for upcoming projects according to company goals. For example, if you know you need to implement a foobar next quarter that will require another team to implement a bazblee that you depend on, they can work with the bazblee senior manager to make sure the bazblee gets done, instead of the plumbus being requested by another group (that is, if the foobar and bazblee are more important to the company's goals than the plumbus - otherwise, they'll do the plumbus and you won't get your bazblee - so you probably shouldn't work on your foobar project)

Convincing other managers of your project's alignment with company goals is usually not an exact science, so your middle manager's ability to get these things committed to and done can be highly variable, depending on their connections with other middle managers and their ability to influence.

You can tell if you have a good middle manager if things you need from the rest of the company tend to fall into place, versus constantly being disappointed when you ask for stuff from other groups in the company.

Of course this all requires skillful ICs and team managers to manage up and make sure the middle managers know what they are doing and what they will need from the rest of the company.


Interesting. We design a project inside the quarter where we committed to build it, middle managers could not possibly be fighting for dependencies during the planning cycle because engineers haven’t looked yet to see what they are.

I’ve long suspected we should be writing design docs and putting them on the shelf one or more quarters before committing to execute on them, but what do I know.


At middle-management levels, dependencies are probably more abstract than system-level dependencies. Imagine your product has no mobile app yet, but one is in the roadmap for next year. Then someone in marketing is in talks with Apple about a chance to do a demo in a keynote or something. This would bump up the priority for the mobile app for this year, but other features previously in the roadmap have to be moved around. So there’s a level of managers which will go around negotiating the options.


(Note: my replies below assume a sane situation at a FANG company with good management.)

#define MM middle manager

> But they are never in any meetings

Do you mean they are never in any meetings that you are in, or do you mean that you have access to their calendars and can see that they have no meetings at all?

I have never seen a MM who's calendar isn't back-to-back meetings all day every day. Their job is basically 100% talking to people, whether that's face-to-face or via email or whatever. Note that depending on how senior they are, their calendar may be made private, since the names of their meetings and who they're meeting with might be sensitive ("Sync about possible layoffs" would be a really bad meeting name to leak).

A large number of these meetings are (1) people trying to influence the MM (i.e. get them to approve something) and (2) the MM trying to influence other people (i.e. get a partner team to do something, or get their boss to approve something).

> And they don't seem to affect the direction of projects in any way

The fact that your team exists and is staffed by N people was an MM's decision at some point. It may have been someone else who had the idea to create your team, but they had to make their case to an MM and convince them that it was worth staffing.

Has your team ever grown, i.e. gotten a slot for one more person? If so, an MM made the final decision that it was worth allocating another person.

MMs spend as much time managing "up" as "down". An MM's boss is asking them to justify why the company has 100 people working on X. To do this, the MM has to demonstrate that there is an important business problem to sole, and present a convincing strategy to solve it that requires 100 people to execute.

If your basic experience on the team doesn't seem to be directly influenced by MMs on a regular basis, that may just indicate that you're in a stable part of the org solving a clear business problem and the MM is satisfied that you are executing. You would see an MM's influence by (big) changes in team priorities, project direction, reorgs of your team, etc.


Fantastic and accurate description of what they do and the problems they solve.

Oddly enough, just as there are those who think MMs twiddle their thumbs all day, there are those who work with or are MMs that think executive leadership twiddle their thumbs all day.


You should try to get in to some meetings where these guys are. Its good to be known by these people, you shouldn't be scared of them, esp if you're working hard for them. Things they do:

Talk to execs to find out what direction of the company has been decided

Choose strategy, approve strategic technology choices, plan for the future

Making sure first line managers & their teams are doing what they're supposed to

Allocate money to projects, decide which ones need resources, which ones can be cut

Make a budget, decide how much money to ask for next qtr/year

Hiring policies, hiring managers, plans for hiring staff

Make sure division is following company policy, implement internal controls, auditing is happy.

Keeping in touch with other divisions and regions to swap ideas and work together.

Firefighting problems that come up with people leaving, systems not working properly, projects running late.


From first principles, as a manager at any level, your output is the collective sum of your team's output. You maximize your effectiveness by finding the right balance of things to do that (a) can contribute directly to the team's output and (b) improve the team's capability to produce.

Managerial activities fall into 3 key categories: Strategy, People and Habits.

Let's contrast these activities for a Level 1 manager and a Middle manager.

Level 1:

Strategy - The projects that the team needs to do are defined bottom up and top down, and the Level 1 manager's role is to split it into right chunks, allocate them and make sure they get done. The operational aspects of tracking and ensuring things get done is more important here.

People - You have 1:1's with team members to make sure they are at their personal best and are aligned with the company.

Habits - You ensure that the team is following the right practices in their day to day that help them be effective.

Middle manager:

Strategy - You try to make sense of the company (or Business Unit) strategy (both product and Go to market) and then boil it down to the key projects that your teams needs to accomplish. The important aspects here are high level definition, prioritization and allocation. You make sure that you allocate the right projects to the right teams based on their capabilities and past projects and that they are crystal clear on the the 1-2 big things they need to deliver per month/quarter.

People - You are heavily involved in hiring and staffing the teams and making sure that your managers are being effective through skip level 1:1s.

Habits - You ensure that right habits (also called processes, but I don't like that word) are being followed by teams. eg. Is the weekly planning meeting happening properly, is work getting tracked, are their development practices that helped one team which could help other teams, and so on.

I hope that gives you are better picture.


When I worked as a director for a large tech company I had some similar questions. Because I joined through an acquisition, I never truly adjusted to the network of managers both above, below and alongside me.

Eventually, about a year after leaving to launch a new startup, I read High Output Management. Suddenly everything made sense. I'd strongly recommend reading it.

Many Bay Area companies are led by people who read this book, and I think you'll gain an understanding of what the mid-level managers worry about in their day to day.


Was Director of IT. My manager would be considered Middle Management. She had a half dozen directors reporting her and would meet with them to get updates and set goals for each department every week. She would pour over all the financials and accounting to make sure things were on track. She would take the results of meetings back to Upper Management and tell them everything was fine. Upper Management would then go play golf and figure out some new way of expanding our lines of business, or meet with random groups of staff to make their presence known, put on a friendly face. No idea what else Upper Management did with the rest of their day


What's funny is that the further removed up the totem pole a role is, the more people assume they do less and less. Sure in some cases that might be how it goes, but it would be worth spending some time exploring how things work at the top. For example, what appears to be a useless golf game may be relationship building for a future major partnership.


Now I wonder which FANG you work at. At the one I worked at (N) you would definitely see the middle managers all the time, and you'd be having skip level meetings with them on a fairly regular basis.

Their job is to facilitate coordination. The same thing your manager does, but for other managers. They make sure everyone is working towards useful goals, make sure hiring and budgets are in line, and communicate with their peers to avoid overlapping work and most importantly make sure to communicate context down their teams about why they are doing what they are doing and what the business hopes to accomplish in regards to their specific area (and overall).


Gee, I don't have a direct answer for you on that question; but tell ya what, I'll run it up the chain and circle back to you on it in a few weeks!


You can only effectively directly manage so many people. So as you hire more people you need more managers but then you get too many managers to directly manage so you need another level.

Imagine you had 10,000 people working for you. Just saying hi to them would take up all your time. How many people would you say a person effectively directly manages? Some would say 5, others 15. I think the upper level of estimates might be 30-50. (Yes, someone is going to say 200.)

Let’s go with 10 for even numbers. So after you hire 100 people you need a manager for the managers. When you hit 1000 you need another layer. When you hit 10,000 you need yet another layer.


The analogy I like using that of a carriage drawn by a large team of horses. Workers are the horses, and CxOs are the drivers. Management is the harness the keeps all the horses pointed in the same direction and running together.

A majority of the physical work is done by the horses, of course, and it looks like the drivers (Top level management) are slacking off - but everything is really on them. "Strategy" is really just where to go and which route to get there. And if they choose wrong or time things badly, everyone else in the organization has done their work correctly, and the failure is purely executive.

The workers are great at running (execution), but left to themselves are horrible at self organizing and running in the direction required in tandem, with the correct beat and stride. It doesn't matter how strong or fast the individual horses are, it's much more important that they run together without friction or distraction and support each other.

That's where (good) management comes in. They're the blinders that keep the horses from individually pulling in different directions and chasing distractions, and the harnesses that keep the entire team fixed together. The exert pressure on horses to both slow down and speed up, and both are vitally important. Done well, management is critical to making sure that the whole org works smoothly. They also act as a communication bridge between strategy and execution. Strategy is great for the all hands once a week or month, where the execs sing about the promised land, but the weekly plans, budgets, and responsibilities of making sure everyone is fed and has toilet access are all management.

Bad managers typically suffer from delusions of being either horses or drivers. They either try to work hard like the horses (bad idea - there's no harness or blinders), or think they're in charge of strategy (they're seen as just slackers, and no harness or blinders again). They don't temper executive vision with their knowledge of reality, nor do they transitively inspire workers.


>The workers are great at running (execution), but left to themselves are horrible at self organizing and running in the direction required in tandem

This is an ideological claim that is counter to my experience. Workers are fine at self-organizing, they just might organize in ways that management does not agree with because workers and management may have different ideas about what the company needs. In my experience the usual reason that workers have trouble organizing themselves is that they are not given control over resources like hiring and equipment purchases, specifically so that they won't be able to organize without management.


Can you share more of your experience?

Mine tends to be like the OP. For example, suppose a company has a new product and everyone believes there is a lot of opportunity with new product. So then each team figured out a strategy on how to approach said product independently -product managers build new features and iterate -marketers start launching campaigns to drive adoption and engagement -data science is building new models to find the best customers for -customer support is trying to figure out what the hell new product is and what all these complaints are referencing

Etc

Depending on the product / size there may be some self organizing collaboration but many times not, or certain teams are left out unconciously. At least one reason is that each team doesn’t yet trust other teams to help set their strategy - not maliciously but just from a lack of evidence. If you are trying to maximize impact the intuitive thing is to work out what you can contribute and ask for what you need from other teams vs developing a joint strategy / approach. Collaboration takes time with an unknown return. Now if you happen to know the other teams and already have worked a lot together you are more likely to reach out from the start.

It’s the senior managers / directors that are calling out that teams should be talking with each other sooner than later, getting buy in, etc.


My experience is that senior managers don't have the time or inclination to be aware of operational issues. They are unable to perceive something like there is a structural issue between units A and B. These issues are actually seen and surfaced by workers in units A and B, who actually experience the result of the mismatch. If they are able to cross-communicate directly they may be able to resolve the issue without intervention from above. However, they may not be able to, typically if this involves hiring.

But the only way that a senior manager could even understand there is an issue is if someone in unit A or unit B makes them aware of it. This is useless overhead. The workers have to be able to successfully communicate the issue up to senior management and usually also suggest the resolution, then get buy-in from senior management on implementation. If instead the workers were more empowered to make decisions themselves the information and decision-making bottleneck could be avoided.


It depends on the job, and the phase of the company’s life.

When times are good and things are growing, self-organization works great because doing what you do better is easy to grasp, easyish to rally around, and whatever fuck-ups happen get buried in growth.

When things are more challenging, answers aren’t always so clear. Somebody has to lose something. You need accountability, someone who owns the strategy and execution. Enter the manager.

The other issue with self management is that it is usually bullshit unless you have a great culture and the good fortune to pull together an amazing team. That is incredibly difficult to do. There’s always a pecking order based on relational based power, and whomever holds the positional power must understand what that means for her team.


> It doesn't matter how strong or fast the individual horses are, it's much more important that they run together without friction or distraction and support each other.

This.


Analogy breaks down because “horses” are just as smart as the “drivers” (sometimes more and sometimes less). Sometimes the horses work their way up to be drivers because of luck and drive. Sometimes the drivers were born into it because they went to posh boarding school.


Not always. Like the prototypical draft horses, the people have blinders on. The scope of their observations are limited to what is in front of them, or what they hear in the stable after work.

The horse may think moving forward on the road is the wrong direction, but the horse also cannot see the obstacle off to the side.


> Analogy breaks down because “horses” are just as smart as the “drivers” (sometimes more and sometimes less)

Sure, but horse is horse and driver is driver. I can give excuses all day for why I'm not the driver.`


The Peter Principle applies here for any large-scale company that has been running for a good amount of time: https://en.m.wikipedia.org/wiki/Peter_principle


They provide a human (inter)face to a group of people, and allow people to talk about/attribute behavior to an organizational unit as if it were a person. The most useful thing they(we) do is deal with bullshit drama on your behalf, like Kelly isn't giving you the thing your boss asked you to get from her last week because Ellen told Kelly to do something different because John is an asshole, and documenting everything to cover asses appropriately. Seriously I just relay information between people and assess whether they've actually received that information correctly.


Thank you for being honest.

Couldn't that purpose be served by hiring people who know how to communicate and are not political, and using technology for group communication and planning?


Why don’t you just ask them? Not in a confrontational way, but like “hey what are you up to this week?” I’ve worked in orgs from 10-30,000 people and have always been able to at least get 5-15 minutes with this level of management.


Right, but maybe even go a step further, and be like "Hey Josh how you doing? Busy day?" They'll likely elaborate, and if they don't want to then it means your question is overbearing in their eyes.


Highly recommend the book “The Leadership Pipeline” written by Ram Charan. The book focuses on developing a pipeline through which leaders within an organization move and describes accountabilities at each level, along with red flags that indicate that a manager at a given level isn’t operating correctly at said level. Very helpful in understanding what happens (or should be happening) at each level.

Ram Charan is a well respected author and consultant, who has according to Wikipedia consulted with some of the largest businesses in the US.


These books usually describe a very idealized version of the real world. I think the only way to find out what mid level managers do is to ask them or ideally to shadow them for a few days.

I think shadowing other roles would generally be very beneficial in a lot of companies so people understand each other better. I have had a lot of occasions where I basically thought that some role is useless until I worked with them and found out what they have to put up with the whole day.


Certainly they do. And the book admits as much. In my opinion, it is still a valuable read as it gives a generalized lens through which a large org’s structure, and activities at each layer, can be understood.


They manage managers. Per every ~8 people, you need some sort of a map/reduce operation. That is a manager. An organization is essentially a giant map-reduce hive mind.

Some managers are there due to them just cruising slowly up the chain.

Management is a really hard problem. There are very very few managers I have met in my entire life that I would work with them again. Some just love the power of ordering people around. Some are just plain assholes.

Seriously, life is too short to be treated like shit.


Really depends on the organization.

Middle managers manage managers.

They are definitely in meetings, just meetings you're unaware of.

My experience dates to IBM in the 1990s so is lacking whatever has developed since then. At IBM first and second line managers still had some understanding of the code or product or services being developed/run by their organizations (and I knew first lines who were still active contributors).

Once you got to third line, those managers are more focused, almost solely focused on "business" issues and macro operational issues.

As you get higher up an organization, the managers could care less about the day to day or even month to month issues faced by their organization. They're focused on will their organization meet or exceed revenue projections? Are you running below or over budget? You've maxed out space in Blue Cube corners and at the rate of growth only need 10,000 square feet next year but they can only rent new space in units of 100,000 square feet. The CEO is demanding 5% cuts across the board, can you do it without firing people? Or is there a particular organization you've been hoping to get rid of anyway that you can terminate en masse?

Even higher up at director and VP level they're focused on profit & loss, because that's what they get measured on, maybe secondarily customer satisfaction. Employee satisfaction is mostly irrelevant once you're above a second line organization.

Depending on the _jurisdiction_ a manager is in, they may have certain legal responsibilities or authorities that a lower level manager may not have. Even if you're not the CEO, you might have to sign off on certain contracts (especially with governments, governments love to have an executive's signature on things, because they can go after that executive when the $100MM payroll system costs $300MM and still doesn't work).


Thats interesting, it feels like that upper level you mention are more like individual contributors, except optimising very slow responding systems indirectly through policies.


Totally. I’ve worked with very senior level people in corporations and government. The people at the top who aren’t on the letterhead are simultaneously powerful and powerless.

All of their information is 2-5 degrees from the source. They are held accountable for outcomes, and need to be able to navigate through all sorts of stuff.

Senior government middle managers are the most fucked. They are either civil service or staff people who moved up the ranks and have extensive expertise in their programs or field or political appointees who know anything from less than nothing to being preeminent in their field.

IME, the government people are smarter, but get ground up by the machinery of bureaucracy. Often times you have layer of formal leaders who are in a symbiotic relationship with the next level down who really run things. Sort of like how in many ways sergeants run the army.


Yes. If you’re a middle manager you’re typically getting directives from on–high that you have to implement indirectly by pushing buttons and pulling levers which themselves are pushing buttons or pulling levers.

Picture driving some sort of vehicle remotely. There's a slight delay to everything, right? You turn, and the vehicle turns a few seconds later. By the time you see an object on the horizon you’re marginally closer to it. And you know this all so you build in certain factors to change response time. But there's still a lag. And now picture driving the vehicle by telling the remote operator how to react. There's now two levels of lag, and depending on how you communicate, two more chances for mis–communication.

It’s why bottom–up communications get to be so bad in large organizations. The classic production line “the product is a piece of shit” transmutes, through absolutely normal and routine communications softening and filtering, into “our product brings things to life through its fertilization qualities.” And conversely, the top–down message gets filtered and manipulated by each layer of management.

It’s also why escalation culture becomes a thing in large organizations. IBM (in the 1990s, can't speak for today) had a notorious escalation culture: if I didn't do a thing you wanted me to do, and was in a different organization, you’d “escalate” up to some middle manager or executive somewhere who had enough pull to scare my manager into doing something, regardless of the business value or need. The executive doesn't particularly care, he or she just wants the issue off their pile, and it makes them feel good that they're doing something to help their product area.

Every organization is susceptible to this as it grows. It’s human nature.


I like that - I think that's about right. Working on the business as a machine, not only to drive the machine (the CEO and his staff), but to optimize the internal architecture of, and to optimize its various inner workings and component-to-component linkages, is a thing.

But - it's a machine made out of people. Whom are inherently squishy emotional things to build a well-oiled machine out of...


When you talk to very senior leadership they'll talk about the difference between "strategic" and "operational".

Directors set strategy. Senior managers try to implement that.


Sure. I've heard that a lot. But that's very abstract. What actual work is involved in "setting a strategy"? And why does it seem less visible compared to other work?


I like to think of it like this: one layer prioritizes which products, another layer prioritizes which users, another layer prioritizes which features, another layer chooses architecture that supports all features, another runs sprints.

Consider allocating resources to an email client vs. a web browser, then figuring out how much to focus on enterprise vs. small business vs. consumer, then figuring out user stories.

All require different analyses - a PM who should be asking specific customers for feedback would be ineffective if they also needed to agonize about the financial consequences (market size, biz dev team readiness to sell into that market, CoCA) of choosing those customers. And even the latter decisionmaker may be better able to focus by having a departmental budget, and letting the C Suite decide if browser dominance is something that maximizes shareholder value.

As a developer these things should seem self-obvious because ideally they were anticipated a year prior, headcount was allocated, and you were recruited at the right time. It’s rarely perfect, of course. But it doesn’t have to be - people don’t need to row in sync down to the millisecond, but without a coxswain there’s no hope.


I like the coxswain analogy because it raises the question of why the coxswain is typically paid orders of magnitude more than the rowers.


I did a bit of research out of curiosity, and these quotes from coxswains at https://www.sfgate.com/sports/ostler/article/Steering-toward... stuck with me:

"My (two) guys were very mechanical. Every 30 seconds, they wanted the stroke rate and the position of the boats that were even with us or ahead of us. Guys really need to know where they are... We control the psychology of the athletes in the race ... You try to wring every last bit of power out of those athletes. If you do that, you can lose a race and eventually come to terms with that."

Any manager out there, at any level, would nod and approve.


A 'strategy' is a high-level coordination on what types of trade-offs get made. Strategy is also remarkably delicate - if you have an organisation with 100 different people, there are somewhere like 100 different opinions on what the best trade-offs to make are. Imagine the chaos; and if you can then containing that chaos is what middle managers do.

Middle management is a fairly involved process of finding people who are doing things that aren't according to strategy and stopping them.

Most of middle managements time is monitoring that everyone is doing what they said they would do, and motivating people to do what the business needs - by hook or crook. A good manager will make the supervisors uncomfortable at the right moment on the right subject. The supervisors will then get things done that the strategy requires in conjunction with their team.

It is remarkable how quickly people will start working against each other without management. I worked on a mine site where the miners sometimes tried to dig out the access to the mine; and while that was extreme people working at cross purposes to each other is common. Someone in any large company always wants the equivalent of rewriting the software from scratch; they need to be found and stopped.


Anything that isn't within a couple of levels of your own role is basically 'less visible'. That's sort of the point, it's about abstracting - divison of labour etc. So you can't really see the work that someone two levels above you is doing. Likewise for the same reason people higher in the org struggle to get information on people a few levels down are doing.

The basic answer is that operational roles are focused on the current phase of work. Middle level roles are focused on co-ordination. Higher level roles are focused on future phases of work (and managing external stakeholders).

For many of the senior roles the reasons the work is less visible is because it's either in situations employees aren't, or considered confidential. For example, CEO meeting with investors is a big chunk of work, but it's not that visible to employees. For an example of the second, I'd say 'organisational design' and areas of investment are often confidential as they have implications for hiring/firing.


Take a typical org.

The board meets and talks and decides everyone is going mobile and that becomes a goal.

The senior level managers (ceo/coo/vp of departments) come up with a general strategy around mobile for their departments. Marketing VP might decide to change CRMs . They come up with a plan that involves getting a budget approved for the assessment phase (meeting with vendors, sme, etc). Based on that they put a presentation together to get support/budget for the next phase. During this time they might try to see synergies with other approved projects with budgets and try to fit in the CRM changes within them. If approved they move to the implementation phase where the work goes to the development manager and/or vendor management. A pm gets assigned the project. They meet with the vp and keep them updated. When the project is finished reports are written and sent back to the VP. The VP makes year end presentation on what they did to achieve the goal. They would report things like mobile engagement up 10% after CRM upgrade. CEO would report to board 80% of customer facing sites are now mobile friendly. Next year we plan to do 100%.


Work includes but not limited to : - Competitive analysis ( what’s happening in the market today) - scenario planning ( what happens if Google gets onto our turf) - Operational readiness ( how do we get to where we want to be ). Will add more a bit later.


I was more levels from the CEO at Microsoft (up to Ballmer) than I was as a lowly intern working for the US government at NIST (up to . . . well, Reagan at the time).

I could see value up to about three levels above me, and maybe a couple at the top, but those three or four middle layers were pretty confusing. We had director-level and partner-level people do flybys of our engineering groups, saying things like "We'll see each other every few weeks, you're doing great stuff!" and then vanish for six months, a year, or forever. They never seemed to do anything that mattered.

I work in a flat organization now. Not necessarily better, but there's a whole lot less dead wood, avoidance by delegation, and simple quit-in-place behavior.


These middle layers are doing a lot more work than you probably realize. I am a little more privy to some of what they do being a senior IC at a FAANG (6 managers away from my CEO) - sometimes managers take on projects that have to be done but are outside their team's scope. The scale of the project may vary depending on what level they are. My own senior manager was in charge of one such project that ended up affecting the whole company (100k+ employees). They might be in high level discussions with executives about plans on a weekly/monthly/quarterly/yearly basis (potentially all 4 of these levels, like my own is), and doing advance planning before first level managers and senior ICs become involved. These managers often have to have an acute sense of what kind of product may be needed, but good ones will leave the planning of the details to their reports. Their job is to make sure other teams/orgs are aligned and that their team/org delivers on expectations set by all parties.

There becomes a point where it becomes impossible to have everyone be aware of all the knowledge and nuance that is going on without it creating immense noise - when this happens, these middle layers become vital. You still ideally want as little of them as possible, but they all should ideally serve very important purposes. I'm very thankful for all the levels of management in my own organization and company at least (experience may vary of course), it is the best management team I have been under at any workplace - other than my direct manager, each level of management up my chain has over 10 years of management experience apiece.


My personal experience was at Siemens VDO (Automotive). And here was the layout: teams (me part of one with 5 other programmers) -> team leader (programmer also, my direct boss) -> groups (my group was BCM - Body Control Module) -> Group leader (non-programmer, the middle manager) -> Department (Hardware for me) -> Head of department -> Board of directors -> CEO. After a few weeks I started to learn that the middle managers roles are to make sure to set broad goals within their groups/teams and to make sure resources are available to programmers to make those goals achievable before deadline. So my interaction with the middle management was very easy - give them the list of resources I need it to make sure the goals are done before deadline. And in doing so I've learned the absolute best truth about this - power is a balance. I have the power, as a programmer, to make life miserable to a middle manager if he failed to fulfill my list of demands. Also he had the power to make my life miserable if he fulfilled his part of the bargain and I do not achieve those goals before deadline. So after learning this truth instead of taking a "sir, yes sir" ass-kissing attitude like usually people do when they meet somebody above them on the ladder, I started to treat them as buddies with "Hi George, how's the house redecoration working" style. Definitely dispelled any tensions and made both of our lives easier. Also I had on other organization I worked for a stuck-up-ass type who expected everybody to bow before him. Since I was never the bow-before-you type sure we did a collision shortly after I started working there. My direct boss also told me to suck it up because he was afraid of that guy too. I didn't like it so next thing I did I went directly to CEO and complained about that stuck-up-ass manager and that he was making my life hard without a real reason. Long story short, for the remaining of my time there (another year) that guy was only honey and sugar with me while still remaining an a*hole with the rest. It all depends of how you crush the bad behavior on 1st encounter in the end.


Most are processing information and communicating. A CEO may have a goal. It takes and army of middle managers to break that goal into actions for individual contributors, and communicate progress against it. And then they work on process improvements to make the above work better. It’s a mix of analyzing, meeting and communicating.


The main function of managers is routing information in the organization; this model was developed in the 1940s. In most large corporations today, there are a handful of C-level officers, the "executives" - their role is to make decisions. The conception of the modern corporation is that these people do NOT need any specialized domain knowledge in order to be effective, they are just generalized "decision-makers", that is, they perform the executive function; what they need is to be fed a reliable stream of information to act upon.

However, in any large organization there are usually many thousands of times more workers than there are C-level executives. This produces a huge, two-way problem of information flow - how can the executives know what all of the workers are doing, and how can the workers know what the executives want them to do?

The solution is a hierarchical model; this hierarchical model has been used throughout history to organize polities, armies, etc., under the control of a small number of individuals. The basic principle is each unit has a single manager; each set of units has a ur-manager, etc. This allows information to flow from the worker to their manager, from the managers to the mid-level manager, and from the mid-level manager to the executives.

Of course, the information remains bandwidth-limited. Each manager can only, practically, assimilate information from a limited number of people. This number probably doesn't change at different levels of hierarchy - the manager can deal with 10 workers before becoming overwhelmed, the mid-level manager can deal with 10 managers before becoming overwhelmed, etc. This means the larger the organization, the more levels of management you need to try to manage the information flow. Roughly, if you have N workers, you'll need log(N) levels of managers.

The efficacy of this strategy, however, is limited by a key constraint: the compressibility of information. If 10 workers are giving information to their manager, and that manager is passing all of this information on to the mid-level manager, the mid-level manager is going to be overwhelmed; they cannot process information from 100 workers. It must be reduced; the job of the manager is to winnow down the information from 10 workers, determine what requires intervention from above, and pass this on.

In practice this is extremely difficult; the information might NOT be compressible (you need, for example, a two-hour presentation to understand a particular security problem and its implications, not a two-paragraph summary). A friend of mine who works at a high level in an org said her observation is that C-level officers are routinely drowning in information; their answer is simply to ignore the vast majority of email they receive and to punt on their executive function.

The same problem exists the other way - a C-level officer cannot possibly make decisions for all of the workers in the hierarchy. They can only give a small amount of information back to their managers; this information does not magically uncompress into a sophisticated plan for workers. In reality, people have to do work to fill in the blanks (i.e., make decisions) at all levels of the hierarchy.

The result is that information often gets lost - C-level officers are unaware that their workers feel that the product is broken, that engineering or security problems are overwhelming management goals, that timelines can't be met because there are too many constraints.


Top notch comment. Really feel it captures the problem succinctly. I’m curious how you built this model. Everything from the notion of compression onwards is novel to me.


Thanks! I'm not sure how exactly I arrived at it, this is mostly the result of my own observation, I think - I don't have much background in organizational theory. My primary training is in population genetics, which also deals with notions of information flow between individuals in a group, so I think this is where some of that thinking emerged. The rest came mostly from my experience (working for a large university) and my wife's experience (working for a large corporation), where both of us observed similar dynamics over the years with regards to communicating with management (supplemented by observations from friends in tech and elsewhere). Both of us also have a significant problem with authority, which has caused us to avoid management positions and maybe be skeptical of their utility, which might lead us to this kind of analysis.

In particular I observed early on that there was a nominal organizational structure, based mostly on control of funds, which often had little to do with the operational structure, based mostly on the need to coordinate work. This separation led me to pay closer attention to who was often doing the actual work that made projects succeed, which often required communications outside of the official management channels - because these channels were simply insufficient to the effort, not through any malicious intent but simply because of bad design.


Excellent stuff. I love when interdisciplinary accidents like this happen to bring insights.


The compression up, and the necessary expansion with inventing gap-filling down (inference) is one of the reasons (together with the general specification problem) that I don't think that AI is going to take over the world soon. That is: The generalists don't know how to specify in detail - they rely on successive layers down to invent what goes in the gap. And then ultimately, at the bottom layer - the code is the specification. Attempting to describe to an AI what it needs to generate code is already what we use compilers and interpreters for. If we aren't clear, in detail, about what we want, what good would an AI even do? The whole AI will replace coders is an age-old dream that used to be expressed by drag-and-drop code-modules that would allow codge-gen to be directed by business domain experts. See how that worked out...


Fascinating view. Also somewhat novel to me, but acts in the same manner as something that management experts like Deming and safety experts like Sidney Dekker talk about: the magic is in the human details at the point where the rubber hits the road. Those humans are actively making conscious decisions about things. The engineer deciding to encode some data model in is using his human intuition and intelligence to trade off future proofing vs ongoing cost etc. using some picture he’s designed in his head about the product. He can’t always ask the CEO about the deets. He’s filling in gaps because the compression is guaranteed lossy.

Interesting view that that would hamper an effective AI. And cool analogy to the drag and drop thing.


It's like involving the product owner in deciding whether to use a linked-list or a bucket or a hashmap for object/DS traversal (say, a list of triangles in a 3D scene). Most of the time - you just wouldn't involve them. The coder decides and implements and may or may not even report off what was decided and why. That's the gap-filling that I think it would be hard for an AI to make (of course there are a lot of space-time-memory tradeoff costing matrices that could support an AI decision process).

What I find fascinating about humans is how idealistic we are (as a species). "Wouldn't it be nice if a business domain expert could just "specify" applications without the need for technical specs or coders?" (It's just like a compiler, but smarter, right?) Or (not to get too political), "Wouldn't it be nice if we could all share economic outcomes more equally without specific consideration as to contribution?" Well, yes, it would! But, as the old trope goes: Capitalism is for sinners, and socialism is for saints - and we ain't no saints.

So, I believe that most humans do their best every day to do the best, in their situation, with what they know. But individually, we can't know everything. So, heirarchal management structures are a hack to sorta/kinda solve a problem with a constraint. 1) No individual is a panopticon or even if a polymath, capable of industrial-scale output, and 2) We are apparently tribal hierarchal monkeys, and we operate better (as individuals of the human species) when there is a chain of command - where no individual can effectively directly manage more than 8~15 reports.

Thanks for your comment. Hopefully, I didn't drive too far off into the ditch in my response.



I don't know how you made that leap - but this is pretty spiffy! Thank you!


That AI uses multiple layers and feedback loops to gain an understanding of it's input environment in order to predict changes in the environment.


Certainly neural models do!


I can tell you that "they are never in any meetings" is highly unusual for someone that's a "manager of managers". The places I've been that's pretty much all they do.


By the way, when I did a poll of how much time during the day people actually worked (including meetings), senior people worked the same amount as juniors: https://news.ycombinator.com/item?id=17068804. Time spent working might be irrespective of seniority on average.


I can tell you that much of "Middle Management" today falls under the category of "Bullshit Jobs" and "Leadership BS" (books have been published on both). It is mainly due to mindset and the way that company hierarchies are structured. "Management" is considered prestigious and hence is given more money then they are worth. Everybody wants to be a "Manager" in one form or another (there is something in the human psyche which wants to lord it over others whether needed or not).

The erroneous belief that if you have a team of 4-6 people you need somebody to "manage" them is ingrained in the industry. Thus the proliferation of management "levels" as axioms.

Ideally they should grow organically only as a need arises otherwise a single level between Strategic/Vision <-> Middle/Operational <-> Engineering/Production should suffice (think of them as concentric circles with Engineering at the core). In a organization with a such a structure true middle management exists and you will see them involved in everything since they are the conduit between the end-goal and the means of production.


Instead of asking what they do (which could indeed be an odd question), ask what their vision is. Most leaders like to talk about it. While they explain their vision, you can ask questions around how to get their, what their role is in this plan, etc.


Odds are you can simply add their calendar on google calendars and see when they've booked meetings.


Managers of every kind have to deal with personnel, meaning humans, and this is a huge part of what they have to accomplish. They divide their time between humanistic goals and project goals, and it's very squishy.

Managers have to devote a certain amount of time to understanding you and your influence on the process and your mental state.

Managers have to have influence with other people. They build relationships and they form a big picture about the work being done and the timelines imposed and the human assets they have to motivate to meet the important goals imposed by the business.

You've heard all this before but it's a very squishy job. There is no template for it. Management is very, very difficult to do.


Bullshit jobs :)

These are jobs that are useful sometimes, mostly not. At best you can say they keep an eye on the organization.


For some reason Middle Level Managers are the strongest opponents of remote work.


They probably spend most of their days sending emails, you just don't see them.


A mix of supervising, teaching and inspiring the people in the trenches


Holidays or home office




Registration is open for Startup School 2019. Classes start July 22nd.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact

Search: