Hacker News Reader

> What's wrong with centralized decision making and planning out everything for years? The rest I can agree with, but 1) in any company, you need centralized decision making or you won't get anything done; the point of a company is not to be a democracy, it's to be a machine that generates profit, and centralized decision making accomplishes that.

No it doesn't. Top-level decision makers are poorly informed about what's possible and what's actually effective. Complete top-down decision making is Dilbert incarnate.

Top-level executives need to make the goals of the company clear to team managers, and that's not just "profit", and they need to set the constraints that any direction in pursuit of those goals must satisfy (such as legal or marketing requirements), and let the engineers find optimal solutions within that space. Engineers excel at finding effective solutions given a set of constraints, executives excel at defining business policies that effectively shape a successful business. Let each employ their appropriate skill set.

> And 2) what can possibly be wrong with planning things out far ahead?

Inflexibility to changing market conditions. Having invested time, money and effort into forming such an intricate long-term plan drives us to stupidly believing sunk cost fallacies, and ignore evidence of ineffectiveness.

No. Top management sees the company in quite a different view than its employee. They also see their own future. So, what they decide for the company might actually not much to do with what the company does. For example, the president of a big car company can have a big interest in making polluting car. So it'll make sure the company is a bit dysfunctional on environment protection. Of course he won't say it. In exchange, the top manager will be rewarded and might event get some more power to help the company in a more positive way later on. I experienced that 2 times. One in a huge japanese tyre company : we were given small book explaining us how the company was ecological, while a te the same time rubber extraction was pursued in an absolute non ecological way. In a gov't company, we're continually told to work for those we help but at the same time, the top mangement decision make that work super difficult.

My thought is : manager are human, and their in position where they have lots of pressure (they've chosen that). so most of them behave has human : they do mistakes, they act more selfishly than they care to admit and they mask the truth with lies or on-purpose ignorance.

As I prefer to act with real good intention and produce tangible results, I know I won't ever make it to the top. I'm not interested in making compromise.

Moreover, when I'm with people who are at the top or who want to be at the top, I can clearly sense that their core-values are very different, even opposite of mines... And I'm sure they instantly feel I'm not part of the tribe. Because that's a tribe with values, customs, dress, etc.

> They also see their own future. So, what they decide for the company might actually not much to do with what the company does.

That's exactly what I said executives should do: set goals and constraints on how to fulfill those goals. You're just saying that companies can change goals, and that's fine. You just tell your employees that your goals are changing, so they shift focus to the new goals. You're still not making decisions for them other than overall direction.

I think what Wiz21 is getting at is that the interests of the company and the interests of its senior managers are not always aligned.

Sure, but I thought we were talking about what makes for good management. Bad managers either need to be reformed or fired.

The big problem is that bad managers get their by sacrificing the company's success for their own success, and focusing on managing perception as opposed to creating value.

For instance if you were a V.P. at an mortgage company you could push your underlings to make bad loans. Even though in the long term it will hurt the company when those individuals default, in the short-term you will look awesome.

Or for instance when a V.P.s boss makes a wide ranging decision about the company will move in a new direction. For instance "To the cloud". You can argue about how that doesn't make sense for your part of the organization and have you're boss and grand-boss think of you as "Not a team player". Or you can cheer-lead a decision you know is wrong, knowing that afterwards you can produce some great powerpoints slides about how failure was outside of your control.

> For instance if you were a V.P. at an mortgage company you could push your underlings to make bad loans. Even though in the long term it will hurt the company when those individuals default, in the short-term you will look awesome.

Having metrics to measure success is of course important. While micro metrics like lines of code produced per day are easily gamed, metrics like number of successful projects divided by the estimated risk are not. For the mortgage, number of mortgages divided my estimated risk is similar, and would easily identify the managers that are putting the company at risk to game their metrics.

> Or for instance when a V.P.s boss makes a wide ranging decision about the company will move in a new direction. For instance "To the cloud".

Do you mean that the CEO is declaring that the company should use the cloud for all of their products, or that the company is to become a cloud provider? Because the former is exactly the wrong type of decision that I've been talking about, and the latter is exactly the kind of decision they should make.

>Inflexibility to changing market conditions. Having invested time, money and effort into forming such an intricate long-term plan drives us to stupidly believing sunk cost fallacies, and ignore evidence of ineffectiveness.

Oh please. I've worked in giant companies that planned things years in advance. The plans changed frequently, with market conditions. Planning things far ahead doesn't mean you're rigidly bound to those plans, it just means you have goals you've set and you're going to work towards unless things change, forcing you to alter those plans.

By your logic, you shouldn't bother going to college, because things might change and planning things out 4-5 years in advance like that is futile. Lots of people go to college and end up changing their major or doing things somewhat differently while they're there; not going to college at all because this might happen would be stupid.

>Top-level decision makers are poorly informed about what's possible and what's actually effective. Complete top-down decision making is Dilbert incarnate.

Centralized planning doesn't necessarily mean micromanagement. It just means not acting like Microsoft with the different business units all competing with each other and stabbing each other in the back.

> Planning things far ahead doesn't mean you're rigidly bound to those plans, it just means you have goals you've set and you're going to work towards unless things change, forcing you to alter those plans.

Except that's not what the original poster was talking about. The OP specifically referenced the centralized planning of communist states, which are poorly informed and inflexible to changing conditions, just like I said, and they employ deceptive propaganda even among the people that are supposed to help them realize their plans.

Nothing wrong with having a rough strategy that you openly share with your employees so you can all work towards realizing. That's exactly the kind of planning I described in fact.

Then the OP was making a strawman argument. I've worked in big companies, and I've seen long-term planning, and it's not like what he complains about. He's alleging that all big companies use this mythical rigid central planning, and it simply isn't true, he doesn't even have any examples.

If big companies were all like failed communist states, then these big companies would have failed long ago. They haven't. Maybe some have, and maybe some small fraction of those did fail partially due to overly-rigid long-term planning, but to claim that this is a common problem among big companies is quite wrong IMO.

As far as I know "podunk" is an insult, is that what you intended?

My problem with the planning obsession is that planning is really hard in software development. It's different in production. There you can predict the future fairly well.

When I tell some middle managers that I don't know exactly how long a software project will take they insist on endless planning meetings that hold up work even more. In the end it's best to make up an estimate so the team can go back to work. All teams are forced to do so so in the end you just wait for the first team to fall behind schedule and take the blame.

The whole system is pretty much everybody lying to each other and hoping they won't fall first.

> When I tell some middle managers that I don't know exactly how long a software project will take they insist on endless planning meetings that hold up work even more.

It's middle management's job to come up with the estimate and give that to upper management. They should not be asking team leads for project-level estimates, team leads are too close to see the forest for the trees. Any time a manager asks me for a number, I ask them to try to come up with one, because their guess is as good as mine, so he might as well come up with it.

To management thinking, not knowing how long something is going to take is indicative of not understanding the problem enough. The solution to a lack of understanding is to get the right people in the room together to articulate what they know about it, figure out what's missing, and work out a correcting plan. In other words, a meeting.

What's different about software development is that it more resembles research than manufacturing. If you were to ask Thomas Edison while he was testing thousands of materials for filaments how long it was going to take before he has the next light bulb, he'd have angrily thrown you out of the room. Yet we expect programmers to be able to answer that question all the time.

The best estimate you'd get for when Edison would be done would be from someone close to him, but not actually working with him, say his secretary or wife. They'd be able to notice all the little things that he can't because he's not paying attention to them.

So in software development, you'd ask the lowest level manager, rather than coder, for a ballpark number. All that person needs to do is look over the history of the group and come up with a number in line with other projects with similar scopes. If there's no project with a similar scope, then a meeting or two is needed in order to work out how to break up the project into more-manageable chunks.

It doesn't make a difference if you ask the coder, the secretary or some manager as long as they are allowed to give a ballpark estimate. That's easy. The problem starts once somebody doesn't accept that ballpark number and wants more precise data. Then you start pouring over velocity data and define the project architecture to a level you can't reasonably know at this stage.

The solution usually is to make up some stuff but tell people about some nonsense methodology you used. It works but it's just lying. It would be much better if management methodologies would be able to accept uncertainty. Agile was one way to do this but it has evolved into a micromanagement tool.

> The problem starts once somebody doesn't accept that ballpark number and wants more precise data.

That's crappy management. If my boss did that to me, it's time for a closed-door meeting where we have a come-to-Jesus on how software development is conducted. If I can't show them the light, I start sending out resumes because my working relationship with my boss is trashed and life is only going to get more painful over time.

Yeah nah. (That's australian for I see what you're saying but you're wrong[1]). One goal of senior management is to try to determine an organisation's trajectory and make decisions that put it on that trajectory in the hopefully medium to long term - but beware short term financial incentives. Junior management has the same kind of thing going fractal style. Bonus points if that trajectory doesn't work out (do you have one or more fallbacks?). While centralised decision making may enable the whole long term planning thing to be realised, in some situations it's not possible (e.g. medical organisation where you have to give substantial numbers of staff very high levels of autonomy) or desirable (e.g. where you need to make staff at the minimum feel that they're in control of their own destiny because for example they're difficult to replace). In fact regarding the second point, I was involved in a project a while ago where the goal was to improve the autonomy of steel workers in order to effect organisational improvement. So there's a whole movement in management which is being fairly widely adopted to avoid excessive centralised planning.

[1] c.f. "nah yeah" - "what you're saying sounds very wrong but it's probably right"

> Yeah nah.

Australian? Yeah nah. It's Kiwi as

"Yeah no" is now fairly common in American English too. See, e.g.: http://www.grammarphobia.com/blog/2013/05/yeah-no.html

The problem is when the decision makers are removed from daily events, and they stick dogmatically to the plan and try to change reality to fit said plan.

A non-soviet example may well be McNamara's time as Secretary of Defense.

As one anecdote put it, by the time the report of a shooting of a elderly lady with a bicycle in a Vietnamese village had reached the DoD, it had become a ageless, genderless, combatant, with a grenade...

> What's wrong with centralized decision making and planning out everything for years?

Read Hayek's "The Use of Knowledge in Society"; it's about exactly this.