After 12 years of software development, I've come to the conclusion that software managers are not needed. From what I can tell, they have meetings, try to get people to work more, and approve time off.
The best team I've been on didn't have a manager. The lead developer handled communication with the IT Director about project status; and that wasn't very often.
We had no meetings, or KPI goals, and other such nonsense. That is, until the company was bought. Everything changed after that and traditional management took over. Most people left within a year.
After 17 years of development and 8 years of management I've come to the conclusion that developers are delusional about the relative importance of their role. I shall be the first to admit that - I used to be one myself.
Your run of the mile non-technical manager sure is not very good at understanding the intricate complexity of all the moving parts of creating software.
Developers on the other hand appears blind to the fact that they play only a minor role in the bigger picture. While technology is difficult it pales in comparison to orchestrating people across sales, strategy, business transformation, creative, development, QA, operations and infrastructure.
That's just the horizontal alignment. In a single entity. In a single timezone. With a single vendor. A scenario which never happens because in real life you have at least 3 levels of technical management, multiple divisions/departments involved, spread across continents and multiple vendors participating.
You need competent people to align, coordinate, communicate and pick up the stuff which falls through the cracks. Developers are not good at this work.
You need a manager. And of course they must be competent.
EDIT: I don't actually think developers are delusional. I worded the phrase to mirror the absolutism of the statement "I've come to the conclusion that software managers are not needed" which is just plain silly.
> Developers .. appears blind to the fact that they play only a minor role in the bigger picture.
> Developers are not good at this work.
Developers aren't some special creatures with one dimensional fixed mindset. They can do whatever that needs to be done.
This is kind of silly stereotyping precisely why makes developers hate managers. I hate managers who think of their team as mere code monkeys that are "not good at" some mythical management work.
I am a manager myself but I never diminish role of my team members as "minor players", "not good at work" ect. They are my peers and partners who are equally as "major players" as anyone else.
> They are my peers and partners who are equally as "major players" as anyone else.
I love this. The best teams I've been a part of are those that had a mutual respect for everyone else's work. I was an engineer on those teams, and thank god for the product manager, tech lead, design work, ux researchers, etc. Things work so much better when there's a real culture of "we're all in this together, and we all have important roles to play".
My reply was worded to mirror "...I've come to the conclusion that software managers are not needed." which is obviously nonsense.
"Developers aren't some special creatures with one dimensional fixed mindset. They can do whatever that needs to be done."
Yes, but it is ok to generalize in a broad discussion like this one. Otherwise we can have no meaningful conversation. In general managers do not understand developers. And in general developers do not understand management. Those that are good have been exposed to both sides.
I never diminish anyone but it is naive to think that everyone contributes equally.
I think I've observed enough of this sentiment over the years, from some very smart and successful people, to come to the conclusion that, if it is nonsense, it is non-obvious.
> I never diminish anyone
Perhaps you never _intend_ to diminish anyone, but to some, your statements may reasonably appear to be diminishing some cohort.
but in the end, you decide who gets promoted or not, you decide who has to put the "extra effort" to satisfy whoever you work for. So of course everyone is a major player, it's your best interest.
but you're also the one who gets the time to see the big picture and hopefully realize that, yes, the development side of a project, while the most fundamental, is also just one of the part. The other being : handling customer expectations, making sure the budget use can be explained, making sure people stay happy even when confronted to "absurd" customer's request, making sure to understand the project ecosystem to make sure you get the right customer contact person in front of your team, making sure the one dev with a broken back gets a proper chair, making sure good project are rewarded, you make sure HR's department craziness don't alienate your devs, you-f*g-name-it (been there, done that, got the tshirt : I've been in dev, business consulting and management :-))
Yeah, I work in an industry where software enables things, so it is central. But even in my industry (business apps), we could just go on with pen and paper like before. So programs are not "absolutely fundamental" but in the 21th century, well, they are :-)
But I can admit it's not like that everywhere, sure. But I've yet to see industries where people pay programmers to work on IT project which are not fundamental in a way or another or in the process of becoming fundamental. The point is when programs get deployed, they usually transform (and hopefully improve) a situation. Once the transformation is accomplished, the program becomes fundamental...
You understand, that company goal is not to "build a good product", even less to "write a good code", it's to increase a company value (in the sense that "good code" is orthogonal to the goal)? It might be unfortunate for developers, but quite often (almost always from what I've seen) developers' role _is_ minor here. Not in derogatory sense, just money don't come from writing code, they come from sales. Also no, not everybody can be a good manager, and not everybody can be a good developer, as those two roles require different mindsets and different training, so very few people can effectively do either (and not at the same time). So a good developer normally can't do "whatever needs to be done". You don't want a developer to do your surgery, do you?
As a CEO and long time developer before then... its complicated. High performance teams are killed by managers unless they stay out of the way and then whats the point right? Managers are great for managing clients, but not devs.
Low performance, bare minimum devs, sure need reminding and pro-ding to get anything done... yes, you need managers to babysit.
Companies with heavy management usually have mediocre at best devs and little to no innovation. They have great sales and support - this keeps them alive.
Captain obvious here - there is no one size fits all. Do what makes sense at your org and take home a paycheck or leave and start your own business - there has never been a better time!
'High performance teams are killed by managers unless they stay out of the way and then whats the point right'
The point is two-fold. First, creation and maintenance of a high performance team as the company grows is worth someone focusing on, not just hoping it happens organically (sometimes people, especially new hires, need to know "yes. You are really free to make this decision yourself. I'll back you on it." Likewise, sometimes people doing good work need someone who can spend the time to highlight the value of it to the rest of the business, to effect a promotion). Second, ensuring other needs of the business inform that high performance team, WITHOUT interrupting it, are also worth focusing on.
These are both true regardless of the environment, but become bigger needs the larger the company is.
"creation and maintenance of a high performance team"
Managers do this? right, right. I would say managers at best dont mess it up. Creation and maintenance of high performance team comes from within. The surrounding support certainly helps, but to say a manager creates and maintains this is a bit of a stretch.
"ensuring other needs of the business inform that high performance team"
I have found if it needs to be known or communicated it will happen. High performance teams are not just coders (another fail from managers) they are intelligent people who can read and write emails and know how to speak to other humans. This may be the stereotype of some hot pocket eating teenager in a closet from the movies but that is certainly not the case in most professional environments.
So for the first, I am explicitly describing managers in an agile, team lead style. But, yes. Managers, first and foremost, hire the team. Ideally from feedback, but ultimately the manager is responsible for the team's formation. But then the manager also has to help smooth out bumps along the way. Having someone objective who can help smooth out disagreements can be HUGE. Likewise someone who can give feedback to individuals; a large personality can be a huge asset to the team, but can also leave others feeling alienated; someone who can help carve out space for those others to speak up, and to help coach the person filling the room how to recognize that and also to carve out space for others, is immensely helpful. And maintenance includes progression, both for individuals and the team; devs are often focused on what they're working on, they're not evaluating how to get their work recognized, and more junior ones often aren't evaluating how to grow themselves. It also includes HR issues; having someone who can do the legwork for your Visa sponsorship, for instance, frees you to do real work. A good manager is basically the team's admin assistant as well.
For the second, of course they are. You have to take the two statements I made together. There is a spectrum; at one end are devs working on what they feel is important, but not knowing they're not working on what the business feels is important. At the other is the business deciding everything the devs do, oftentimes changing priorities every week. Having one person as a middleman there can be immensely helpful. That may look like a manager saying "Hey Bob. Jane has this need; work with her to understand and implement it", and handling everything off to a dev, but the point is that the manager, A. Gave Jane (and everyone else) one person to reach out to initially when it came to their needs of the team, B. Helped determine that Jane's need was a priority against all the other needs, and C. Gave Bob permission to ignore all other incoming requests and direct them back to the manager, to focus on Jane's need. That is not to say Bob is not reading and writing; it IS to say that Bob can benefit from someone keeping him from being interrupted by every person in the business with a need who thinks he might help.
This feels right. I also believe that what people need are LEADERS and not MANAGERS. I'm sure there are plenty of examples of the two on the web.
The challenge I see a lot of managers have is that they look to much at an engineer as a widget and not literal human that is just as capable as the manager (most the time). There are often times no lines between roles and responsibilities and every shop does that slightly different, and not based on some framework they masterly put together, but just something based on how the pieces fell in place with some help from previous experiences.
I just wanted to say also - _most_ engineering managers are not good managers. They were good engineers that are in the unfortunate process of the Peter Principle.
This is very true. Most shops have an R&D team for long term innovation and other devs like myself who love customer feedback and want to deliver solutions now.
I have not heard this about "most shops", in fact I have never worked in an org with a functioning software R&D team (unless you count enterprise architects choosing vendors).
No but I've seen similar dynamics play out at most companies. Broadly, "product teams" focus on customer feedback and delivering value to customers, at the expense of technical debt and lots of fickle requests. "Infrastructure" teams are heavily insulated from customers and deal with a more long-term roadmap and fewer changing requirements.
just an observation: if you moved from developer to CEO without significant time as a Dev or Eng manager you have very different skills and focus. CEOs of all but the smallest companies are playing checkers; Dev Managers play chess. This isn't some sort of insult; we need to focus on individuals with distinct skill sets and characteristics. CEOs can't and shouldn't be concerned with this level of detail.
It's also uncharitable to split teams into either high-performance special ops teams vs. mediocre, low skill / low potential teams. Two trends make this distinction meaningless: GROWTH and SCALE. If these don't apply then you can totally let the team self-manage. FANG companies typically have all those rockstars you're looking for and multiple levels of technical management.
Even if CEOs "can't and shouldn't be concerned with [individuals with distinct skill sets and characters]" (not a point I'm conceding, what makes you think a CEO's job is "checkers" compared to the chess of a Dev Manager who is realistically 2-4 levels below them in the organization?
High performance teams are usually created by good management. If you change the management then bad management can easily undo that. My best managers have been deeply technical people with good people skills, an interest in management, and willingness to get out of the way and no longer make the decisions they made when they were a developer. Management is a lot more than babysitting it generally involves a lot of coordination up and down the organization to ensure the team is doing the right work in terms of organizational and customer priorities. Good management can give you what to work on while leaving you to decide the how.
The role of the manager is to ensure the team can get on with their jobs, to celebrate their success and have their back when things don't go according to plan. Unfortunately most manager can't help to start interfering with the team and go for a top down driven management style.
oh, and any given dev can switch from high performance to bare minimum depending on the project, week of the month, what they ate for dinner last night, etc.
Same could be said for a manager.. or a manager that gets bad devs or devs that get a bad manager. This is the value of a hands-on ceo that does not have their head in the sand.
This is true but I've also seen developers add ten percent more revenue to an organization on a rewrite of an existing billing system and it wasn't due to any insights from management. It was many conversations with key business users and developers. Management knew enough to step back and loosen the reigns. Everyone can make a huge difference on a project, and a team of non hero's that just care about learning the business and the craft can deliver greatness.
Sometimes you can plan for success/heroic actions (in the good way) but often it happens by accident.
The right person digging into a weird problem they encountered, two people sitting together to review a design or someone doing a firmware update that fixes a bug in the RAID controller giving you double the IOps...
I have tons of anecdotes from colleagues who did some unplanned work just because they talked with someone else and that work then turned out to have massive positive impact.
The results are lauded but it wasn't even clear when the "little" project started that it would actually be a massive boon.
At a prior job (some website selling hotels online) I was responsible for the project that ended up giving the whole fleet a 25% capacity boost. Instead of 2000 machines we only bought 1500 next quarter. If we calculate that in perpetuity my project was a massive success. Millions of EUR saved since 2011.
The project initially started out as "tons of people tried building a decent perl RPM but couldn't get it to work, could you have a look?". The guy originally tasked with it was busy fixing a broken LDAP server.
So I built some RPMs, wrote a bit of tooling around it and we ended up with a Perl environment separate from the system perl which was great because the business was nearly exclusively running on Perl 5.8.9. At that point I looked into using that to get us off CentOS4. With the help of two colleagues we got a CentOS5 environment running and could migrate to a x86_64 environment.
That migration than gave us a 25% capacity boost and happiness ensued all around. The OS upgrade wasn't part of the original project scope, we just went with it because it made sense and management understood that it's a good idea.
I guess the problem are not managers per se, but non-technical managers. Somehow it is OK to have a pure salesperson / MBA manage engineers, but you would never put a developer in charge of the sales department.
I think many developers (not just coders, but in general technical people who make the actual product, designers, engineers, ...) feel there is an unfair primacy of managers / business people over developers / technical people. This is even reflected in the job descriptions: I hate the term "individual contributor" for example, it implies replacability and that non-ICs (managers) contribute more than one person.
Not every company has "bullshit" managers and for sure some developers are delusional. But these bullshit managers exist, and I believe this happens when you put people who are far away from making the product in charge. If the managers and executives are not intimately familiar with the product and the customers, then the product becomes some interchangable thing that you hustle. This can be disastrous for the work climate but also for the success of the company.
I always wondered if it was possible to build an inverted company organization, where developers/engineers were formally at the head of things, and much of the administrative tasks that management works on were subbed to specialists reporting to the developers.
If you look at a lot of pretty incredible projects that often were finished ahead of schedule or with groundbreaking capability, they often happen to be led by a deeply technical person. e.g. Hoover Dam, the SR-71, etc.
If you look at CEOs of tech companies, I believe a majority of them have an engineering undergrad degree of some form.
I work at a company that operates much like this. I've passed up on more lucrative opportunities in the past because it's such a great place to work as a developer as a result.
Actually, it implies that managers are multiplicative. You -hope- that it's a value > 1.0. I quite like the term individual contributor - it implies you're actually doing work (and that it falls on the rest of the business to make sure you know what work will be valuable).
But, yes. Bullshit managers invariably choose to take empowerment onto themselves, despite a lack of knowledge, and oftentimes without any real responsibility, either (i.e., they don't bear the consequences of their decisions).
I even worked at one incredibly toxic organization for a year, before I left, where sales people would literally roam the office, grab a dev, and task them with a customer request.
This company was in a very specialized niche market, and was basically the only player. The only thing I couldn't figure out is why someone else hadn't come along and eaten their lunch, because they were suffering from chronic under-funding. Getting them to commit to migrating to AWS was a major eye opener for them; since they suddenly had to actually pay bills or they'd lose the whole operation.
It sounds like you've gone from one extreme to another. None of the things you mention pale in comparison to software engineering at all. They are all solvable problems that many people can do. Sometime they are difficult, often they are not. Managers are certainly not useless and do work for certain types of organizations. They are not always needed though. Believe it or not there are people capable of managing and programming at the same time. You seem to be one of them so your view point is very strange.
No, I've actually mostly worked in well balanced places. It is my experience that technology is the lesser issue and the one with the smallest impact. That's not to diminish the highly skilled developers - the work just have less impact that we seem to believe.
I do not believe in self organizing teams. I have not seem them work. Without direction the team will fumble until someone assumes control. I have also not seen many happy in such an environment.
The opposite is strict top-down control, such as what is often described and lamented on HN, is equally bad. No one wants to work under a dictator.
There's a balance to be struck between authority and autonomy.
Large scale software development without management is proven to be possible by the existence of functioning FLOSS projects like the Linux kernel, various GNU tools, KDE and plenty of other examples.
I have not heard of a single software development company with 0 software developers. Not even Oracle manages that.
Managers are necessary for the parts of the company other than development. You can have a FLOSS project developed organically, and separately a company in the business of support, services, etc. Managers are also necessary in order for the company to exercise control over the development and extract value. And this may be needed for a company to survive. But they are not needed for software development itself.
Even FLOSS enthusiasts need to eat. Money don't come from writing code by itself, they come from sales. Or from donations. In both cases there is a lot of non-coding work. Do you want to do it? I don't. How do we call those guys who organize donations or sales? Collect requirements? Yes, there are requirements in FLOSS too, otherwise you wouldn't get donations, or support contracts, and those requirements are coming from the people who pay. Let's call everybody developers, only some won't write code and would organize coders, gather requirements, distribute payments. Wait, those people are currently called managers.
I'm reading: "You need a manager to coordinate a deeply hierarchical and separated structure with production at the bottom and the clients at the outside."
I mean no disrespect, but from my perspective I see a deeper problem here that is being patched over by exceptional people like yourself. Shouldn't the creators and clients have a major, high impact role and the coordinators a supporting role in terms of decision making?
To make an exaggerated joke it looks a bit like a dictator complaining about all their responsibilities and the hard decisions they have to make. Well maybe they shouldn't even be in that position to begin with.
And yes I've seen the diagrams and the theory behind this. But I've become more and more suspicious of bureaucracy and process that tries to fight the symptoms of lack of trust and human connection.
>And yes I've seen the diagrams and the theory behind this. But I've become more and more suspicious of bureaucracy and process that tries to fight the symptoms of lack of trust and human connection.
This resonates but what I have seen is when you engage a company that has this model you have to emulate their model to fit within their needs, then you get infected with it.
Ha - yes, you're right it sounds like a tautology. I worded my answer to reflect the absolute statement of no manager is ever needed.
"Shouldn't the creators and clients have a major, high impact role and the coordinators a supporting role in terms of decision making?"
No. The creators usually answer "how to do it". The more important question is "what to do". Both groups are important but doing the right thing is more important than how you do it.
Then make sure developers are deeply aware of all the concerns you listed. A large part of why developers think they are the center of the world is because they are meticulously isolated from said world. This is exactly what the article addresses as one of the main concerns.
As a developer, I’ve noticed a pattern: devs complain about too many meetings -> devs are left alone (aka isolated) -> devs complain about being left out of decision making process/chain of comm.
It's a good observation. IME you can't be in all the meetings you need to be in to obtain alignment on complex problems in a changing environment AND be effective in writing code, very generally speaking. My most frustrating career points have been when I was a tech lead, doing a bit of both. Being more purely IC or purely managerial, it is much easier to get important work done.
Its also a grass is geener issue. Work is generally more enjoyable as an IC, you get to focus and write code. Until you end up having to build something really stupid, and you want more control over product direction. Then you end up as manager, and long for the simplicity of being able to focus and write code. At least that's how I've felt at times.
Meetings are to align people. Complaining about too many of them can be a sign of many different things, but one of the biggest is needing to align with so many people to get anything done.
If you hold the same number of meetings but stop inviting some stakeholders, those stakeholders are going to be upset. If instead you remove some meetings from the process entirely, by e.g. not requiring multiple design reviews for internal products, or by not letting platform teams force everyone else to use their software, or by letting teams spin up their own low capacity services without meeting with ops, or anything else that lowers the complexity of getting work done, then your team will be happy and the previous gatekeepers will be complaining.
I agree. And then if the same devs have reduced meetings, are included in more cross dept initiatives, and included in decision making, they end up upset that their freedom is suddenly gone. Its not even just devs this is a common thing from call center reps to the top of the broken pyramid. Everyone is the linchpin in their perceived contribution to the world.
Both of those problems can be solved to a degree with more efficient communication. Async communication channels like email and other messaging are great at not wasting peoples time while keeping everyone in the loop who needs to be, properly used. Sometimes I think phpbb style forums would work better than the usual tools. But overall I find people don't experiment enough or put enough effort into optimizing communication and it leads to this false dichotomy of either isolation or time wasting. There's a middle ground of optional participation where you communicate widely but don't force peoples attention, and that's how you can reduce meetings without reducing valuable communication. You can always put a time limit on feedback if needed, mark some things important... Just find what works!
That isolation is often as much self-imposed as not. While the GP may have been a bit aggressive in his/her characterization of it, I agree with the general thrust that developers tend to be both narrowly focused on their own technical work and prone to inflation of its worth.
Of course it isn't: most individuals tend to inflate their contributions' importance. There are matters of degree, though. My experience is that developers tend to overestimate more often and by larger amounts.
My answer was intentionally phrased to mirror the absolutism of the GP. A large part of my time is spent explaining why certain process is in place or why we cannot get a license for this or that tool or why one needs to be considerate of the other crafts.
> Your run of the mile non-technical manager sure is not very good at understanding the intricate complexity of all the moving parts of creating software.
> Developers on the other hand appears blind to the fact that they play only a minor role in the bigger picture.
Y Combinator was largely founded as a rejection of this premise. Paul Graham believed it was much easier to teach great Software Hackers how to business, than it was to teach business folks how to software. I think it turned out pretty well.
> A scenario which never happens because in real life you have at least 3 levels of technical management, multiple divisions/departments involved, spread across continents and multiple vendors participating.
And your organization with 3 levels of technical management and all those divisions and departments will often get their lunch eaten by a very small group of people working in a startup, moving faster and accomplishing more in less time.
"Paul Graham believed it was much easier to teach great Software Hackers how to business..."
I agree
"And your organization with 3 levels of technical management and all those divisions and departments will often get their lunch eaten by a very small group of people working in a startup, moving faster and accomplishing more in less time."
You misunderstand. It is not my org - it is the client orgs. And none of those are software companies. The speed with which you can move is irrelevant because you are constrained by the pace of the org you're working for.
I'd like to make a wild inference and suggest that you and the parent poster are talking straight past each other, because you're drawing your experience from two different business paradigms.
kbmunchkin wrote:
> The lead developer handled communication with the IT Director about project status
The fact that the ultimate person being reported to was the head of IT strongly implies to me that it was an in-house development team working on in-house projects.
Your list,
> orchestrating people across sales, strategy, business transformation, creative, development, QA, operations and infrastructure.
Strongly implies that your experience comes from building consumer products.
Personally, I've spent time working on both successful and unsuccessful projects on both sides of the fence. What I've seen is that the most straightforward way to create a dysfunctional team is to try to take what works from one context, and apply it uncritically to the other. I personally hate working with a dedicated product manager on an in-house project every bit as much as I hate working without a dedicated product manager on consumer-facing stuff.
Consider that those two statements may be about in the same wheelhouse, but "developers aren't needed" is way, way more rediculous than "managers aren't needed".
> developers are delusional about the relative importance of their role
I think it's that developers fail to grasp that they are in the enterprise world. Web development and SaaS has blurred the lines, but at its core it's the same: the product you work on is a tool for another business. It's a bit more sexy because it's not mainframes, cubicles and suites anymore, but in the end the job is the same. And in that setting, the role of a developer is indeed minor in the bigger picture. There are exceptions in R&D, cutting edge projects, or in an early stage startup, but for the vast majority of developers the above is true.
Strongly agree with this. I've been in situations where my team had no manager, a bad manager, and a great manager. The situation where I had no manager was the worst, by far.
I've found that if you throw a bunch of Alpha Geeks together without constraints you may as well be herding cats. They'll want to work on their individual wishlists of strategic research problems (throw in this new tool! new methodologies I saw at this conference!) with less concern for the unromantic tactics, socialization, and alignment that makes a project succeed.
The issue with this argument is that a significant part of the "bigger picture" is artificial bloat which emerges from the fact of having layers of management. Managers spend much time actually managing themselves. It's of course not all black or white.
I don't think it's an issue with my argument but I do think the difficulty of grasping the bigger picture is a major issue.
It can be reluctance to understand why you're doing what you're doing. It can also be because no one cares to explain it to you. Most of the time it's probably between those two extremes.
And yes some times it's bloat. But rarely all of it. I don't have a solution to this.
A delusion is a sincerely-held false belief. Please consider: If all of the developers left your organization, could it still produce products? How about if all of the managers left? Your implicit devaluation of labor is not congruent with the actual reality of whose hands work the keyboard.
(And before you write a knee-jerk response, double-check your reasoning. Without managers, developers produced the Free Software ecosystem. Without developers, what did managers produce?)
I tend to believe these arguments just curve right into confirmation bias.
I’ve been on teams that worked fine with managers and those that did not. It depends on the commitment of the actors in question.
Anything else is reading what we want to in the tea leaves.
Edit: Been working on distributed systems since the late 90s, from public uni to big corp, to day 1 startups. It’s just computer configuration. Not magic.
Clearly some management is required somewhere in the chain, and if they're any good then they're important. But certain tiers of management seem to largely be pointless intermediaries; especially if they're not as technical as the developers.
I've stopped being skeptical of the general concept of management. I'm no fan of LinkedIn, but Reed Hoffman said something in a talk that instantly changed my entire opinion about the idea of a developer moving up in the hierarchy and no longer writing code. Paraphrased from memory: "You're still programming; just at a higher level of abstraction."
Regarding selling - I've know amazing people who could sell all kinds of products that don't even exist. So selling by itself doesn't seem to be the hardest thing - selling something that relates to the reality of the product is far more difficult.
Drain a company of managers, and the devs will self organize and save the company. Drain the company of devs, and the managers will desperately try to hire anyone to replace them. Devs are vastly, vastly, more important than management. Any given manager could leave a company and literally no metric would move.
I agree, you need a manager, but through my limited experience I feel as though the issue is developers are mostly detached from the buisness end of things. When things go wrong or some issue occurs I've always had managers step in to aid from the buisness side. You almost need a manager who has developer experience like yourself to make for a good manager. I suppose even a manager who is able to understand the general overview of what the developers are doing would be good.
Basically it seems the average manager in buisness would be a bad fit to manage developers. overall a manager is still needed for sure. I guess its the same everywhere though a bad manager vs a good manager makes a world of difference.
This is a similar mentality of people who get very upset over incentives sales gets and/or "company provided trips".
We all forget, at least in a for profit company, that we are all employed by selling things. Like it or not without the good salespeople, some of us don't have jobs.
This is all true, but reads more like a PM-type role, which is absolutely critical.
I understood GP's point as more directed towards "pure Eng. manager" type positions, often when there's also already a PM present. And I've witnessed first hand how his criticism rings true in that case.
I'm coming up on 15 years of professional career in software (development experience being 25 years or so).
In small projects, a lack of managers can be pretty effective when there is good team cohesion and lack of conflicts over priorities and expectations.
In larger projects and teams, where your team has to interact with other teams and navigate competing pressures within a larger organization, while still delivering on internal goals and keeping turnover reasonable.. managers are crucial.
The best managers make you feel like there is no manager. They run interference for you. They talk to you, understand your individual priorities as a developer (project priorities as well as career priorities), and coalesce that information across their entire team to quietly craft a path forward that preserves cohesion and effectiveness and productivity and happiness. They also communicate progress to higher ups, and interpret (to the executives) the day-to-day operations with relation to broader organizational goals.
Managers are the interface between larger organizational priorities and a team's individuals. For a company, managers serve the role of both implementing policy top-down, as well as understanding operational dynamics from the bottom up and communicating it to the higher-level positions in the org.
They are fundamentally a structural necessity once you step beyond a certain scale of project or product or company.
My personal thoughts on managers used to be closer to yours, but it has evolved over time. These days, I'm more inclined to think that _everyone should get an opportunity to be a manager_. Maybe even a structure where management roles are cycled between different people on the team with a willingness to take on the role. I think more developers need to understand how difficult good management is, and to get a personal exposure to that problem space. I think it would make developers into better developers.. who, rather than reacting blindly to their local perception of the symptoms of management (poor or good), get a firsthand view of the other side, and learn to better interact with it in the future.
> everyone should get an opportunity to be a manager
One way to test drive being a manager: read the Ask A Manager column (https://www.askamanager.org/) for a few days or browse the archives. Read the question submitted by the reader, consider how you would deal that problem, then compare your response with Alison's (and the commentors below.)
Yes, most the problems are not technology or software problems. Some are downright silly. But that's largely true of software development, too.
> The best managers make you feel like there is no manager. They run interference for you.
This is the most important thing, and its hard to recognize as its invisible to the developers. "Do I need a person running interference so I can focus and get stuff done?" is also a useful metric to decide if/when you need a manager.
We're going through this at the moment, company got bought, original CTO & IT Director left. Now the new bobble-heads try and introduce nonsense processes, much to the dismay of everyone already working productively.
I think the problem in large part is that most "managers" are brought in later on and then feel they have to go above and beyond to justify their existence/position/salary. In the end they do things just to be able to talk about all the amazing things they are doing and how their shiny new processes are helping everyone so much.
I have yet to see a successful founder introduce nonsense simply for the sake of it.
There are some jobs with a goldilocks element. Such that the job can be done just right, but it is easy to overdo or underdo it. Management is the most common job in this scenario, and we tend to really notice when it is overdone. Toes get stepped on, trust is not felt, everyone is unhappy. Unfortunately it many of the people that get into management try and push themselves to achieve as much as possible, which is what leads them to overdo it and make their teams unhappy.
I absolutely hate useless management layers and unnecessary ceremonies but I've been on teams with no hierarchy and it was Lord of the fly until the most power hungry developers became the effective lead/manager.
Nature abhors a vacuum (c) We are social animals and there is no such thing as "all people are equal" or "flat organization". There is always a structure, only it can be explicit and regulated (and usually more efficient and easy to navigate due to explicit rules), or it can be implicit (aka Wild West).
I’ve been in the startup scenario where the most power-hungry developers successfully rebelled against the CTO and had him resign; “Lord of the Flies” is exactly how I described the maturity-vacuum period that followed.
I was on such team too. There was cycle of power hungry leader, revolution, everyone for himself which special kind of dysfunction, someone else takes power again, revolution, everyone for himself which special kind of dysfunction, ... .
That lead developer was effectively the manager then.
But you can't just leave all choices to developers, there has to be a commercial point to it all - that isn't something a lot of devs even consider. Technical changes may be great, but the time spent on that has to paid for somehow, not just in 'job satisfaction' for the techies working on it...
When I was younger I wondered what all these other people in the business did with their time. Surely it was all about the code and all these other corporate drones just needed to respect that. Then I guess I started moving over to the dark side and got a bit more empathy for the real jobs that other people have to do - so devs don't have to spend ALL their time talking to customers, paying the bills, handling the marketing etc. Strangely many of these people, even customers, were curious about likely dates for shipping the work. Turned out their jobs depended on it too. I recently interviewed an engineer working at a startup run by engineers. There is no management, which he thought he'd like. Instead they use commitment. After 18 months of very long hours and weekends guess why he wanted to move on? Organizing the work and the people is an art, lots of people get it wrong even when they try hard to get it right - but these articles? I hope my competition is reading them.
> The lead developer handled communication with the IT Director about project status; and that wasn't very often.
I think depending on the organizational context, and how much external communication a project needs to succeed, this can mean that the lead developer becomes a manager in responsibility but not in resources. This can harm the project and its participants both because the lead developer's attention may be siphoned off towards management-related activities, and because their advocate to upper-management is less equipped.
E.g. Can the lead developer recognize that the project needs a person with a skillset not already present, write a JD and oversee the search for a suitable candidate? Or will the organization say "tech leads don't hire; managers hire"? Does the lead developer have the same access to the decision-making process of the larger org that a manager would, or will they be denied access to certain meetings, documents and resources? ("Sorry, we keep that personnel info in the same place we keep compensation info, which only managers should see").
I think the article makes the point that the problems preventing an org from making good products or decisions extends all the way up to the CEO, and I agree. And in the many organizations, software managers are a necessary evil to try to compensate for those broader dysfunctions.
> this can mean that the lead developer becomes a manager in responsibility but not in resources.
I think you've nailed it here - what's been described is a manager under another title, and one who will be hampered in that role in the exact ways you describe.
ive seen this happen alot... a lead dev is a manager without a title, and is not recognized or have any true authority, and they arent compensated for it either... its the worst of both worlds...
After 20 years I realize managers are important to keep other managers away. If you have ever worked without a manager in a department you become the manager as other departments will need someone to meet with someone. More importantly secret deals, backstabbing and positioning are going on above you. Your department might be chopped up next, you need someone to fight those battles.
> From what I can tell, they have meetings, try to get people to work more, and approve time off.
Do not under appreciate the importance of managers attending meetings, because a good manager attends those meetings so the devs don't have to. A good manager of developers seeks to shield her developers from the busywork and politics imposed by the larger organization so they can be as productive as possible.
Exactly. Once the number of stakeholders crosses a certain threshold, your lead developer will be so busy with them that they will essentially just become the manager because they wouldn't have any time to code.
Well, maybe you can call it manager. But it would be a "project manager" and no one would report to them - at least that is what I meant with the alternative that I suggested.
Bad managers are actively harmful to a team. Great managers will make a team shine with productivity and satisfaction. You'll never accept a bad manager again once you work with a great one. Sadly, they are few and far between. And to truly be great, their managers all the way up to the CEO also need to be great. So a role under such a leadership team is rare. But they do exist. It is worth your time to seek them out.
The best teams I’ve been in have been the ones where there was no filter between the developer and the client. The more filters, the harder everything becomes.
My previous role, I served 100s of clients, each with 1000s of end-users. If those users had direct access to my development team (5 devs, 2 test engineers, 1 BA) they'd spend the entire day fielding requests and get nothing done.
So, to avoid that, you add a product management team. And an SDM, who acts an umbrella over the dev team to prevent shit raining down on them from tech support, product management, sales, etc.
I suppose that's the key - if an SDM doesn't see themselves as a shit umbrella, they aren't doing their job. That's not their only role, but from the developers' perspective, it's perhaps the most important.
Edit - the team did interact with customers via various channels, but it was more controlled than "unfiltered". We had a forum where both sides could communicate (though the BA typically did the lion's share of that work from our end). We also had periodic beta review meetings with select customers. And of course, when tech support issues got to our level, we were often on the phone, VPNed into their systems, etc. But, between the BA and I, we triaged most inbound communication, and there was zero expectation a developer would respond unless specifically asked by me or BA (though they were free to do so if they wanted - most preferred not to do so).
Yeh Back in 94 I and another developer got flown up to Edinburg and delivered using RAD/DSDM in a month what a another traditional team said would take 2 years.
One of the other ideas at the times was a custom BT version of HTML - that got stamped on very had by the Labs thank god.
the illusion of a management structure.... ive seen it a few times in very dysfunctional corporate cultures where they value the illusion more than the reality...
in those cases the managerial "class" is more like an old-boys club, and its the grunts who do the real work (and the boot when things go south)
So the developers take on the responsibility of a manager so they don't need a manager? The value of a good manager is that they can do some of the stuff on that list so the developers don't have to (ideally including all the 'fighting').
> So the developers take on the responsibility of a manager so they don't need a manager? The value of a good manager is that they can do some of the stuff on that list so the developers don't have to (ideally including all the 'fighting').
Incompetent managers fail to unblock their teams and instead push pressure on their reports to "fight". Often blocking career growth, promos, compensation changes with these reasons.
I'm sorry. If the dev ultimately does this work, the manager should be fired or made an IC where they can demonstrate their value.
While I see your point and get where you're coming from, I think it would be made explicit that "management isn't needed" is a gross oversimplification that is only possibly valid below a certain threshold of company size. I currently work for a very large international company and none of the things we do would be even possible without management.
We're also lucky enough to have pretty good management, at least locally in my department, I haven't really been exposed to the higher levels of decision-making (which is a sign of good management IMO, that stuff is mostly irrelevant to me)
I have admittedly less experience than you, (~10yrs) but from what I've seen, my conclusion is that really bad management is worse than no management at all, fairly bad management is about as bad as no management at all(and from here can the opinion that management isn't needed really form in developers head) and good management is way better than no management at all.
I also think that software/tech management is a different beast from basically all other types of management. Experienced managers from other sectors tend to fall in one of the first two categories above in my experience.
Good managers are perceived from the developers point of view as someone who just helps organise tasks, and help you find the right person to talk to if you need something from outside your team.
In reality of course they do lots of other stuff too, but those things mostly shouldn't bother the developers IMO.
Never should a manager ever act as a go-between. Doing that puts you firmly in the "worse than no management" category of managers
I've been programming professionally since 1993, and I can tell you that some of the software managers I've have had been invaluable. When they are good they can be very propulsive. I don't necessarily have time to collect up all the different and sometimes contradictory specs from the product owners and make sense of them. When the manager is technical and can do this for you it's golden.
Although I am not completely sure I agree with the thesis, I think it's worth pointing out that one of the largest software projects in existence, Linux kernel, doesn't have software managers either.
In any case, when I first read the headline, I thought, well, in a worker cooperative, the developers can fix the bad management. :-)
Yeah - I think that's good insight. You need someone that has all power and resolves conflict if necessary. But even for something as big as the linux kernel, Linus did the dictator part _and_ continued coding, doing PR reviews etc.
How much time did he spend on coding as Linux got bigger?
And you could argue PR reviews was how Linus performed his management function. Which could be an interesting way to organize technical management in proprietary software projects, too...
> How much time did he spend on coding as Linux got bigger?
Yeah, I don't know. But even just keeping up with PRs seems quite time intense.
> And you could argue PR reviews was how Linus performed his management function
You certainly can, but just because the word "management" appears in there, it is wholly different from the "management" that the original post is talking about. But yeah, you can call both management. You can also call a software developer a manager of the codebase I suppose.
I think the original article is predicated on the notion of "management" as a role (specific person so designated), and not a skill, that can be taken up by any developer. If it was the latter then the whole argument would fall apart.
Personally, I think managers are primarily intermediaries between capitalists and laborers. So they are needed under capitalist mode of production, where surplus value is extracted (according to Marx, at least), but they are probably superfluous under different modes of production, like that of open-source software.
A significant amount of work on the Linux kernel is done by contributors who on the clock for Intel, Red Hat, etc., and I'm sure have software managers.
My experience is about 1/3 of developers are completely lost without a manager checking up on them periodically, between 1-5 times per week. (Varying depending on seniority, while the frequency changes it seems like the percentage does not; junior developers are more likely to get lost but not always in the ways a manager check-in helps.) By this I don't mean a one hour meeting, but something more like a standup, or just an email "hey, last week you were working on X and I didn't any updates about it, how's that going?"
If you get a team of purely the of developers that don't need this, it's great! But that means somewhere else there's probably a team that's 2/3 or worse that, and that's hell. If there's a manager, it's luckily just hell for the manager.
One factor I think is span of control. If the director you mentioned starts getting project statuses from 15+ teams then the director might not scale. The director will not be able to scale for things like appraisals, burning issues, hiring.
> The best team I've been on didn't have a manager
I feel exactly oppositely. The best teams I've been on have been ones where the manager has a strong focus on being a crap-umbrella for the team, sheltering us from anything coming from above, being an advocate for our needs, and a fierce seeker of true requirements for the projects that our team needs to work on.
I think I would be very anxious being on a team that didn't have some manager who took seriously the role of sheltering and nurturing the rest of us. It's obviously not an easy job to do well (else we wouldn't have had bad management experiences).
Now imagine that setup with a bad tech lead who didn't know what they were doing. No oversight, no control, no feedback, no communicated goals to keep people accountable, etc.
Every single approach works when you have driven, intelligent people who all mean to do the right thing. The validity of an approach is only tested when those things stop being true and you need to notice and recover.
Well, no approach works when you have incompetent people doing the bare minimum and trying to subvert anything for their gain, so I don't known where you want to test anything.
The truth is that not all approaches work on your perfect scenario. Many are known exactly for turning competent cooperative people into incompetent individualists.
My first questions would be (1) how big is/was the team? (2) how experienced? (3) how fast were you growing/changing? (4) without any sort of process or documentation how did you handle performance reviews (good and bad), promotions, growth? I'd also bet that your lead was (a) really good and (b) doing 2 jobs.
IME smaller organizations with experienced & stable teams can do fine without formal management, especially if the system is small enough to fit entirely in you head. Beyond that you need some form of management.
Aside: you come off at best somewhat ignorant or less charitable as kind of a jerk; there's a lot more to running a software company than writing code by yourself. We get it; you think all managers are useless MBA suits, but some of us actually still self-identify as devs that just want to improve the systems that build software. You can only get so far focusing solely on individual improvement.
Sure, at very small sizes where everyone personally knows everyone, there are a lot of things you can get away with.
I like the story of google, however. They theorized they didn't need managers, tried to prove it, and ended up proving the opposite. It's almost the kind of thing only google could do, that is a large enough company with the resources to run large scale studies internally, yet is curious and cares enough that they're willing to run such a study to definitively determine the direct value of managers and manager traits.
Managers are not needed but management is. There are many ways to fulfill the task of management, one way is to have a manager, one way is to have the developers handle it themselves.
If you're lucky you can get a team of devs who are capable of self-managing and communicating with the outside world and handling everything else they need to do to be successful besides just writing code (writing JDs, talking to customers, dealing with budget, etc.)
But hiring a team like that is hard. Developers like that cost a lot of money. The law of comparative advantage is a real thing. If you can't hire a team that can manage everything themselves you can at least piece together people who can fulfill the different functions.
And that's how you end up with developers and managers.
Seems like it depends on how you define ‘manager’. Your lead developer is what I would call a manager. In my experience, most managers have been team leads who set priorities for the group, help arbitrate technical decisions, advocates for more resources for their team, negotiates with other teams over who will do what work and take responsibility for which parts of the system, advocate for promotions for their team members, and deal with underperforming team members.
I don’t know how a large company could function without that role. If you have 200 developers working on 100 different projects.... who is deciding what to work on? Who is deciding how many people should work on what? Those people are managers.
I like how Engineering Manager is defined in "The Managers Path" -- which is what you are describing. They shoud be someone who does an occassional bug fix to maintain a boots on the ground approach, but is spending 95% of their time on the items you listed. I think one issue w/ Managers is that we get manager's who should be at this technical level, but are instead so hands off (or fully non-technical) they can't even reason (or delegate) priorities when architecture or tech debt is involved. To be fair, I have worked with fully non technical managers who had a good understanding of how much time to allot towards architecture, refactoring, etc. But those are rare.
I've come to the conclusion that software managers are not needed
I think some layer of management is needed for strategy, business operations, market positioning, etc., but they should get out of the way as much as possible, similar to how Joel has done it at Fog Creek.
Many years ago I was visiting a startup being incubated at Stevens Institute in Hoboken, and there was a group of graduate engineer students on-site looking at a demonstration; the CEO asked what the most important factor in building the product was, and the very smart engineers talked about output, efficiency, portability/miniaturization, etc. The CEO stopped the brainstorming and said "whether or not someone will buy the product".
Aren't there a lot of things that need to get done that are annoying to developers? I personally enjoy having someone else in charge of all the manager responsibilities such as planning who works on what wand when, handling expenses, etc
I bet you would hate to spend 4+ hours per day in negotiation meetings, stacking up plans and writing reports, some days not even having time to start working on development tasks. You'd probably give all this to someone who wants and has experience to do it, someone like...a manager?
If your lead dev didn't have to spend that much then I'm afraid that experience does not extrapolate on what larger companies do; if they did then I think you're missing the elephant in the room that the lead dev was de-facto the manager at the expense of their direct duties.
Depends on the team experience, morale, complexity of product, organization, and customers.
I also agree that a high-performing team tasked with a clear mission don't need more than a glorified secretary.
But if any of those start to fail then it starts degrading everything. Without decent managers, development/progress can collapse.
Decent managers can paper over the other deficiencies to a surprising degree.
Good software devs can also paper over deficiencies in the chain as well to a surprising degree.
Bad managers can be part of the problem too. Bad devs will obviously kill the whole thing.
And of course everyone in that chain has their own agendas, from selflessly devoted to the macro-organic corporation to machiavellis edging for any economic and power advantage.
Capitalism is supposedly about the free market, but organizations strangely don't align themselves to the basics of microeconomics. They are centrally planned top-down pyramids. Incentives will never properly align to maximize organizational productivity because of the totalitarian nature will reward those with power, not productivity.
I know agile and scrum and so on carry a lot of baggage these days but what you're describing is Less Agile. You flatten the org back out, create product teams that pull from a single common backlog and developers interface directly with their customers. That's not even the most controversial part! You also do away with pay tied to individual goals: everyone succeeds or fails together and are paid the same way.
I liked this until everyone got paid the same. That's the fastest way to lose all of the people who can get more elsewhere. And it's not going to be the laggards who jump ship...
I think below a certain threshold of competence, a manager will negatively impact the productivity of a team mostly by eroding any ability to focus on a single task at a time. A good manager should shield the team until it is appropriate from the very things a poor manager would be dumping on the team as they arise.
KPIs are generally set on a timeframe where they usually become meaningless by the end of that timeframe. For software, goals set for a year out are almost never going to be meaningful by the end of the year.
Pick your flavor of Agile or whatever process your team can agree to, but I doubt the term "KPI" will ever come up when doing real life planning and work.
1) It's pretty clear then that you don't have much experience with planning at level beyond a few people if you've never heard the term KPI come up in software
If that was true developers would make their living just by sitting in front of their personal laptops and coding. Instead they tend to apply for jobs at companies that have a lot of managers
My manager seems to defend the team from a lot of bureaucracy. I see his schedule and task lists and other paperwork and I am extremely happy to have him.
Well, in my experience I'd say it depends. It depends on the developer. But it takes a good manager to understand if a developer needs management or not... :)
The best team I've been on didn't have a manager. The lead developer handled communication with the IT Director about project status; and that wasn't very often.
We had no meetings, or KPI goals, and other such nonsense. That is, until the company was bought. Everything changed after that and traditional management took over. Most people left within a year.