Comment by choeger | Hacker News Reader

Comment by choeger | Hacker News Reader

When comparing leadership that (mostly) works (in the military) to leadership that (mostly) doesn't, I notice a few fundamental differences:

1. Flat hierarchies. In SE, there are officially only very few levels between a grunt and the CEO. In the military, hierarchies are deep. You have squad leaders, platoon leaders, sometimes even more staff, company staff, and so on. That means the gradient between a leader and someone being lead is rather small. Ideally, everyone can step up to a higher position immediately (useful because of the, ahem, high fluctuation).

2. Planned career advancement. The military has always agreed that leaders (that is, officers) require special training. But that training comes after they have learned the fundamentals. In SE, most managers come from dedicated management backgrounds. If an engineer becomes manager that is considered unusual and difficult.

3. Strategic training. Officers have wargames and even colleges dedicated to improve their decision making. SE Managers have what? A few books on amazon? It's not just the how that is important in great leadership, the what should not be forgotten.

Your observations are interesting to me because #2 and #3 have virtually zero overlap with my life experience.

#2 - working at a major tech company, there was a big emphasis on progression for each level up the engineering org, as well as the expectation that anyone (with at least say, 6 months experience at a given level) would be capable of stepping up to fill a hole at the level above them on short notice. This happened from time to time and often was an opportunity to accelerate career progression -- if you did well filling in short-term at the next level up, you'd probably get some extra training and get promoted to make that permanent in the near future.

Also #2, in my 10+ year career I've never met an engineering manager (at any place I worked, or in my extended network) who got in via any path except spending >= 5 years as an engineer first. I'm aware that such managers exist, but I've never encountered one.

#3 - Every place I've worked has had a management training program of some sort. Where I worked most recently, if you wanted to move into engineering management you must be a Staff Engineer and you must apply (and get a supporting recommendation from your manager) to the Apprentice Manager Program (AMP). During AMP, trainees get assigned a small team to lead, and that team is part of the training program, as they give regular feedback. The managers in training attend weekly training sessions, have homework, etc., and are getting feedback from multiple directions throughout. At the end the candidate may or may not be approved to move to the manager track - and they also have the chance to decide it's not for them and go back to the IC track with no problems. They can also apply again in a year if they fail the first time through.

My understanding is this kind of stuff is pretty standard among major software companies.

> leadership that (mostly) works (in the military)

I'd say by everyday metrics, military leadership doesn't work. It is incredibly wasteful, structurally leading to the $5K hammers and an entire year to design an official face mask.

It also utilizes and requires a "punishment is one misstep away" dynamic - without which there would be no functional lower levels of the pyramid, and higher levels would look very different.

You treat an engineer badly, they leave their job and possibly go to your competitor. You treat a soldier/pilot badly, and ... you can keep doing that for quite a while.

Steve Jobs' management style was, perhaps, the closest SV can get to military style. It works, but I wouldn't work there. And Jobs paid people handsomely to endure that.

On the other hand, there's Linus Torvalds. He has incredibly effective leadership, over almost 30 years now, without hiring people, without even paying them, yet still getting a lot done. Not entirely flat hierarchy, but anyone is one email away from Linus (though not one commit away); No planned career advancement or horizon, and no strategic training.

I've seen many projects on the spectrum between Jobs and Torvalds, and in my experience the Jobs side of the scale is not generally more successful.

> On the other hand, there's Linus Torvalds. He has incredibly effective leadership, over almost 30 years

He is not effective on my scale: he has alienated a huge amount of contributors with his abusive style of communication.

And please do not say that it is necessary for "filtering" and quality assurance, it is equivalent to justifying mobbing.

Another unsuccessful aspect I blame him for is the sorry status of the kernel security, "a bug is a bug" and similar stuttering nonsense.

Who is effective on your scale? I suspect Jobs, Gates, Ballmer and even Pichai have alienated more people than Torcalds, if that’s your metric (it’s hot mine). Though Linus is the only one doing it in public - other than some anecdotes about chair throwing and cursing, you don’t hear about the others.

It is not acceptable whether it is in public or not.

Yeah, I wonder if flat orgs are similar to open plan offices in that they're a good idea when you're small, but don't scale. On paper it sounds nice if there's only 3 people between you and the CEO, but that often becomes an excuse to avoid implementing any process for bubbling ideas upwards. "We're a flat company, just go talk to the CEO!" doesn't work when there are a thousand people trying to talk to the same person.

Not my experience at all. All kinds of random people - including HR office clerks, janitors and receptionists, this is no joke - end up sneaking into the product org via personal connections.

The bar to become a product manager is incredibly low in tech, as demonstrated by the mere existence of roles such as “junior PM”, often filled by kids straight out of arts college.

Not in the top tech companies, specially in a technical product, most PM roles prefer previous software engineer expeience.

Regarding #2 - is that really true? My understanding has always been that while NCOs might well steadily work their way up from grunt to rank N, COs start immediately at N-1 (or N-2, whatever), minimal non-managerial/leadership training required, with indeed no expectation that they would ever fulfil any lesser role.

Then they work their way up from there, the NCO's maximum conceivable career goal being for them just some minor stepping stone on the way to whatever - or they get stuck/give up on the way, and go off and do something else. Up-or-out is not uncommon in non-military lines of work either.

Apples and oranges. Officers start as O1 as second lieutenant or equivalent. Enlisted men including NCOs start as E1. An O1 always outrank any enlisted man. They go to completely different schools. Officers hope to go to eg Army War College and learn about commanding great units, tens of thousands. Enlisted schools are more tactical.

No. And not to the parent comment. Senior enlisted schools are tactical and strategic and philosophical, with a heavy emphasis on ethics. Even technical schools have both tactical and strategic subject matter to provide the basis for reasons and rationale for systems specs and design.

For example, before they ever see a fleet unit, junior Marine officers have had 10 weeks of OCS (not for academy grads), 26 weeks of TBS, then they are sent to the school house for their specialties, which can be for a few months to over a year. OCS and TBS are essentially run by senior enlisted personnel.

And I sat in two schools where officers and enlisted go the the same school, and where NCOs were coveted by the officers' after-school study groups.

Guess again, Batman.

OCS is Officer Candidate School. Graduates become O1. Not sure what your point is.

Here is a link to the Army War College if you want to see what they teach.

https://ssl.armywarcollege.edu/dmspo/index.cfm

I don’t know which world you live in but in SV from my 10 year plus experience, 90% tech managers were software engineers in the past.

Reply on news.ycombinator.com