Hacker News Reader

Comment by wpietri | Hacker News Reader

> There is a worrying trend of treating staff as equally capable but untrusted cogs in a wheel, and metricising performance.

That's a great way of summing up what I've heard called "managerialism".

There's a theory behind, e.g., the MBA that a competent manager can manage anything. That what one manages makes no difference as long as you're a real manager. One of the corollaries of that theory is that all one has to do is find the right short-term metrics and make them go up and to the right.

I don't think this theory makes much sense when you examine it. But it's excellent at justifying a caste system, one where all workers are controlled by wise managers, and where there's a stable hierarchy of executives above that. I note that the hierarchy established is basically equivalent to those of feudalism. It also has the handy property of justifying high compensation for managers and executives; since workers are cogs and higher-ups do the important thinking, all gains should flow upward.

There are other ways of organizing things, of course. But this has become such the default that people have a hard time even imagining alternatives.

View on HN

> But it's excellent at justifying a caste system

exactly. the managerial class claims dominion over the rest of us, and then swiftly constructs a rigid social system that locks in their position on top of it. this necessarily includes promulgation of unfounded myths and beliefs taken on faith to support the ossification of the system.

>promulgation of unfounded myths and beliefs taken on faith to support the ossification of the system.

a.k.a. neoliberal economics

I tend to find people with your view have never been an executive managing the managers, or even a manager. When you are (like myself) then you see that good managers / leadership (including tech team leads, etc.) really do make a massive difference in performance with the same people.

Things like metrics, values, and re-orgs might seem like Dilbert-esque corporate speak (a view I used to share), but they're incredibly important once you're trying to get a bunch of people to do something with excellence and speed.

I would be more inclined to believe this if the pro-manager view didn't come from the managers the vast majority of the time. It seems too convenient and self-service. Of course managers want to think they make a massive difference.

Certainly some aspects can: keeping politics away from your group so that they can work, focusing work tasks by eliminating extraneous ones, advocating for people in the company. However, once you start getting into things like values and metrics it starts becoming a lot more grey. On the one hand metrics are a great idea, a tiny step towards making management more science driven. But on the other, the metrics so often used are such a poor and overly simple representation of a very complex system that it's often hard to see how they provide an accurate view of any given work place.

Goals and values are often completely lame. We value hard work! Your goal is to do your job! Somehow managers must know that they are ridiculed for this. After all entire movies are made doing so. Yet they still seem to want to believe this rhetoric works.

I believe "The Zen of Python" is one of the reasons for Python's success. I also believe that most "goals and values" that CEOs obsess over are bogus, but there's a kernel of truth that setting the philosophy for the company has a subtle, but strong effect.

Your skepticism based on who's speaking is warranted. After all, you shouldn't ask your barber if you need a haircut. Yet, maybe your hair stylist has the most expertise. Should you trust your dentist on how often you should get your teeth cleaned?

Sometimes companies are forced into doing metrics. If you need to fire someone for poor performance, in order to avoid a lawsuit, you'll need to document it, and that requires metrics. The same goes for lawsuits over who gets paid how much.

As long as there are regulations and lawsuits over this sort of stuff, there will be metrics.

From ground level, though, it's really easy to see the dysfunctional cultures growing up around metrics. At the factory where I used to work, we'd zig-zag from metric to metric. Yield one quarter, Cpk the next, scrap rate the quarter after that. Special cross-departmental teams to manage the metric. Regardless of which metric was chosen, people would do really dumb things to try to meet it. If scrap rate was the metric, we'd send terrible material through the line to keep the scrap rate low. If yield was the metric, we'd scrap anything that was even slightly abnormal. Cpk was the worst (or maybe the best), because we just adjusted our spec limits until Cpk was acceptable.

The better managers translated the quantitative metrics into qualitative guidance for their teams. The worse managers provided no such buffer and instead made their teams achieve metrics by any means necessary.

> The better managers translated the quantitative metrics into qualitative guidance for their teams. The worse managers provided no such buffer and instead made their teams achieve metrics by any means necessary.

This seems like a key insight.

Metrics are key to good management – but metrics are for the managers, not for the teams. Managers that pass metrics directly down to their teams as goals aren't doing their jobs; they're trying to make their team do their job.

Maybe. Thoughts?

Agreed. What I left out is that being a good manager can be really hard. If this quarter, the metric is scrap, and you've already exceeded your scrap budget, you have to be willing to tell your people to scrap bad material anyway. You have to do that, knowing that you're going to get raked over the coals in the managers' meeting. Knowing that it's going to count against you at review time. Knowing that the people you manage mostly don't understand that you're all that stands between them and the torrent of abuse. Middle management is a hard job.

I sense a "its hard but its good" theme in this post, and frankly, I disagree.

Making middle managers accountable for the failure of executives to manage their company, effectively just insulates those people who made decisions from those who know they are terrible decisions.

I was a middle manager who had to do this very thing, and even though my company was happy due to making gobs of money, every review I had was terrible because I wasn't hitting the metric of the month (that generally ended in horrible burn out and hiring even more replaceable people) and I was constantly warned that "You wont bonus this month."

I never got a bonus while I worked there, and yet was highly respected and valued by all my employees.

When I left the company the other great managers followed me, and the business folded, having hundreds of people lose their jobs.

So tell me, why would anyone want this or claim that this type of decision making is "good management"?

Why setup a system that punishes those that help the business and deliver actual value to customers instead of meaningless metrics?

> Metrics are key to good management

This doesn't follow from the GP's comment. All we know from that comment is that Metrics + some managers = bad.

Parachutes are key to survival when jumping from airplanes. Doesn't mean they always work or are always used correctly.

Maybe this is a matter of necessary but not sufficient? I have no strong opinion, I am just pointing at a grey area.

Guys, this one is self aware!

This article by Ben Horowitz seems relevant - he mentions his history of managing teams by metrics. I like this quote:

"Management purely by numbers is sort of like painting by numbers—it’s strictly for amateurs."

http://a16z.com/2011/07/19/when-employees-misinterpret-manag...

Hello! I have been both, as well as a founder. I also have been consulting on organizational process since before you started high school, so maybe lay off the condescension a bit.

I'm not saying good management is unnecessary. I'm saying that managerialism is basically the opposite of good management. Metrics can be valuable if used very carefully. But they so rarely are, because a lot of the drive for metrics is really is in service of a cultivated shallowness. Simple metrics allow somebody to swoop in, jerk some levers, provide apparent gain, and then jet off to something else, never mind the consequences.

This makes two big assumptions that can't be overlooked:

1. People can't self-manage (to some degree) 2. Performance is the end goal of management

I think both of these are wrong, or at least unfounded. I work in a more-or-less self-managed environment, and performance is actually pretty good. I get to pursue about 50% of my work independently and the other half is either collaborative and in line with a coworker's needs, or is one of those "must-dos". I don't see a possible world where I'd get more work done over the average day if there were artificial metrics like "time spent at desk", "lines of code written" or "number of tasks completed" dangled over my head - I'd probably just fuck off more and come up with clever schemes to break the game. This problem can be seen in academia as well - give a bunch of smart people a matrix of rules and yell "GO!". Watch them optimize.

I'd also argue that if it's only performance you're looking for, you're both ethically and intellectually off base. Good managers do increase performance/productivity - bad managers do not. But to assume that, as with teachers, academics, or software engineers, you can measure and reward behaviors without causing adverse affects on your workplace then you're dreaming. My manager is good because he backs off when needed, leaving me and a coworker to argue about an implementation design. My manager is good because he sets up an area of focus and says "ideas, go!". My manager is good because he will let people take a week or month to pursue a new solution that he personally doesn't understand, but trusts them to solve it aptly. My manager is not good because his team excels in certain metrics - his team excels in certain metrics because he is good. And last, my manager is good because productivity is not the end goal - employee happiness is. Happy employees do good work, and more importantly, they're happy for their own sake, which is much better than being surrounded by burnt out metric junkies.

So maybe my takeaway is that if we're going to do any kind of metric gathering it should probably be sub-perceptible and not be treated as a golden calf. If my CEO wants to devise a clever way of using me as a science experiment in pursuit of better management practices then I encourage that. But to tell people "here's what we're tracking now go off and do good work instead of that" is really fraught with problems.

I am also a manager of managers. I agree 100% that managers/leadership makes a big difference in performance.

But I don't see the difference being systems and metrics. I see it much more as interpersonal relationships, respect, and caring about the people that work for you. The manager's competence in the domain itself is also important.

Selecting the right person for the job is also vital, but you are specifically addressing different performance by the same team with a different manager.

I manage manager managers. I agree 200% the managers make a big difference in performance and we need more of them.

</clever_sarcasm>

That's great, but I'd love to hear what your manager says about your recent performance. Is there any way I can sit down with him or her and maybe come up with a game plan of how we can improve their performance in helping you meet your managerial goals?

> The manager's competence in the domain itself is also important.

Correct me if I'm wrong but this is a minority view in management circles

I tend to find that people with your view have a psychological and material necessity to believe the mythology.

"It is difficult to get a man to understand something when his salary depends upon his not understanding it." -Upton Sinclair

The need for this sometimes seems to be, itself, an aspect of managerialism. Sure, if you're going to the moon on a tight timescale you probably do need to marshal a larger number of people than can usefully self-organise in the available time. But most projects are much smaller scale. A feature of managerialialist organisations is that there's rarely any consideration of the possibility that a give project could be done by a self-organised group (let alone a single smart individual). Everything starts with managers assigning a "team", which then inevitably ends up requiring management.

Sure there are plenty of managers who do a good job but there are also plenty of useless managers who do more harm than good. Especially in complex jobs like academia or software engineering it's pretty easy to make up some metrics that look good but in reality do more harm than good. And somehow the sample people that created this system are then paid to fix it.

Yeah, but it makes a huge difference whether the managers have domain knowledge or are just dictating based on statistics. In the latter case, their expectations will be unrealistic and they won't be able to process the input from below effectively because they've already agreed to implement imperatives from above.

Professional managers can of course learn about different jobs, but there is a massive and largely unpleasant difference between being manged by someone who's been promoted for their domain competence and someone who has been parachuted in because of their expertise in ordering people about. Since it's virtually impossible for junior workers to fire a manager they dislike then what appears to be efficiency is often achieved through exploitation. To know this and perpetuate that pattern is to disavow moral agency, converting what is nominally a cooperative enterprise into a zero-sum game.

I can almost guarantee the managers improved nothing other than reporting of work. Before they arrived everyone was too busy doing it to tell execs what was happening. If anything I bet tangible productivity is reduced with the introduction of management structure and more detailed reports.

It seems like you're talking from a viewpoint that doesn't understand the point of the article. There is no manager I've ever encountered in a university who has been of any value to the academics they've managed or produced any increase in productivity or excellence with their ridiculous metrics.

True, to borrow the metaphor there are 10x as well as 1x and -1x managers.

The problem however, is that they all tend to be absurdly compensated whereas logic would support paying only the first group in that manner.

This thread reads like an argument on the merits of standardized testing in school.

I suppose if you value employees who are trained by rote and operate in a manner reflecting that training...

I think this argument against "rote learning" is basically defunct. The bottom line is unless we want to live socialistalically and force roles on children like in Germany we need to teach ALL the stuff to ALL the children, and the only way to do that is to blast through the material with tons of homework, rote learning, after school tutoring, standard testing, etc.

So it is a choice we make, tell everyone what to do from an early age and allow them to focus, or try and teach all the things to everyone, using standard testing to make sure everyone is at least keeping their head above water.

</best_post_ever>

Sarcasm? I hope.

If this is the case, then why are so many bad managers still managing?

> I don't think this theory makes much sense when you examine it. But it's excellent at justifying a caste system, one where all workers are controlled by wise managers, and where there's a stable hierarchy of executives above that. I note that the hierarchy established is basically equivalent to those of feudalism. It also has the handy property of justifying high compensation for managers and executives; since workers are cogs and higher-ups do the important thinking, all gains should flow upward.

Let's not throw the baby out with the bathwater. I agree that metricising performance is only as good as your metrics, and in my personal opinion, metricising knowledge work is a fool's errand (too many variables to accurately capture performance). That is _separate_ from having managers though. Managers can often triage work effectively, help engineers unblock their work, help identify people in the org that contain the domain knowledge necessary to help one of their direct reports, etc etc.

I think the whole urge to track velocity and points and stupid things like that to be absolutely dumb, though. Also I'm an IC and am coming up to the point in my career where I have to figure out a way to not be pushed into management.

I'm not opposed to managers. I am one, after all. But I don't think "management" is a universal skill. I think somebody who manages developers should probably be a developer. I think somebody who manages doctors should be a doctor. And I think there's a lot of merit in non-hierarchical systems, like the way doctors and lawyers are mainly accountable to professional organizations.

I agree that metricising knowledge work is a mistake. But I think it's also true for other kinds of work. In colleges I worked in a factory and there was no end of trouble that came from managing via metrics.

> I think somebody who manages doctors should be a doctor.

i could be wrong, but i believe in 'real' professions like medicine and law, there's some kind of board/bar rule that prohibits being managed by someone who never had the actual job they're managing. any one know if that's accurate?

I find physicians like the idea of working for doctors, and hospitals have shrewdly supported the role of 'chief medical officer", while keeping financial controls with the other C level officers. The Children's Hospital in San Diego is famously "Rady's" because the physicians ran it into the ground and the Rafy family bailed it out on condition they accept professional management practices. Not coincidentally, the Rady family also funded the the UCSD Rady School of Management. Also not coincidentally, many health care workers loathe working there, generally attributing this to a belief that management is more interested in the bottom line than patient outcomes, and this leads to a general distrust of Management.

For a professional engineer's license, you have to demonstrate competence and have 4 letters of recommendation from PEs who have either managed you directly, or have overseen your work. So there is not a requirement that you (as a budding prospective PE) may never work for someone who isn't a PE, however it does limit your ability to achieve your own license. Therefore, there is a desire from the younger engineers to seek out PEs to work under.

There isn't in law, but, in most (maybe all?) US jurisdictions, only a lawyer can own a law firm.

Agreed. The big problem in academia is that when you don't trust academics you need administration to provide oversight. Administration is now a huge portion of the cost of university education. Not research, admin.

To be fair, academics can also be childish, vindictive, sexist. There is also a real problem of hiring one's friends and students. BUT inter-organizational competition helps to some extent with these problems. Regardless, they justify administration and associated costs.

Managers, CEOs, police officers, border agents, and presidents can also be childish, vindictive, and sexist - but they're rarely subject to similar oversight.

There's something disturbingly paternal about the idea that unusually clever and/or creative people can't be trusted to act like adults and therefore need adult supervision from people with authority over them.

Seems you both agree but are talking about modern management practices vs the concept of having someone "manage". The workplace class of "Management" is much more than just the act of managing you describe.

What you describe is how I've seen software dev management properly work, managers are a support structure for the workers, like Admin. They have important and useful skills, but don't get the position of metaphorical aristocracy.

When you have MBA style dated general management they operate by putting the devs subordinate to their management "expertise" and then they don't "triage work effectively, help engineers unblock their work, help identify people in the org that contain the domain knowledge necessary to help one of their direct reports, etc etc." because it's not really in their interest to do so.

It's not just incentives either, there is a huge cultural component, which is why it's so easy to parody "tech management" and have the caricature be recognizable around the world.

What is an IC? I think it should mean something like senior engineer but I've only seen this acronym in the context of intelligent community

Individual Contributor (i.e. not a manager). It's a standard term at Microsoft, but I don't know how widely it's used elsewhere.

Individual Contributor, if I'm not mistaken

You're right. Thanks and fixed.

Internal Consultant?

Intestinal Combustion?