Go to Making Light's front page.
Forward to next post: Refrigerator Pickles
Subscribe (via RSS) to this post's comment thread. (What does this mean? Here's a quick introduction.)
I don’t want to say “there’s a big buzz on the internet” about a subject you probably haven’t heard of unless you read business news. If you do, odds are you already know that the August edition of Vanity Fair will have a major article by Kurt Eichenwald called “Microsoft’s Lost Decade.”
Vanity Fair Daily published a short teaser/summary of it, Microsoft’s Downfall: Inside the Executive E-mails and Cannibalistic Culture That Felled a Tech Giant:
Analyzing one of American corporate history’s greatest mysteries—the lost decade of Microsoft—two-time George Polk Award winner (and V.F.’s newest contributing editor) Kurt Eichenwald traces the “astonishingly foolish management decisions” at the company that “could serve as a business-school case study on the pitfalls of success.” Relying on dozens of interviews and internal corporate records—including e-mails between executives at the company’s highest ranks—Eichenwald offers an unprecedented view of life inside Microsoft during the reign of its current chief executive, Steve Ballmer, in the August issue. Today, a single Apple product—the iPhone—generates more revenue than all of Microsoft’s wares combined.This strikes me as magical thinking: you make your company more competitive by making its internal departments and individual employees compete with each other. Wherever it comes from, IMO it’s profoundly dysfunctional. Business is about getting work done — unless you’re in a line of business where that work consists of figuring out who’s a star and rewarding them, which is rare.Eichenwald’s conversations reveal that a management system known as “stack ranking”—a program that forces every unit to declare a certain percentage of employees as top performers, good performers, average, and poor—effectively crippled Microsoft’s ability to innovate. “Every current and former Microsoft employee I interviewed—every one—cited stack ranking as the most destructive process inside of Microsoft, something that drove out untold numbers of employees,” Eichenwald writes. “If you were on a team of 10 people, you walked in the first day knowing that, no matter how good everyone was, 2 people were going to get a great review, 7 were going to get mediocre reviews, and 1 was going to get a terrible review,” says a former software developer. “It leads to employees focusing on competing with each other rather than competing with other companies.”
Companies and departments are by nature internally cooperative clusters of people who are working on the same projects and/or issues. Turning employee evaluations into a game of winners and losers and stars, and employees into competing gameplayers, is not a good way to get work done.
Another way to look at stack ranking is as a mechanism for concentrating rewards on the people most likely to be hired away, or to be tough competitors when employed elsewhere. I can see how that might appeal to executives who are into the theory that you have to pay inflated salaries to get top-gun executives, which has been the basic rationale for the ongoing inflation of executive salaries. I can’t construe that as a counterargument to my opinion that it’s not a good way to get work done.
The “grading on the curve” aspect of it is also defective. Basic managment theory limits the number of employees that can report directly to a single boss. Any department that’s small enough for everyone in it to be reporting to the same boss is too small a sample for that boss to be grading them on a rigid 20-70-10 curve.
Besides, as any kid who got curve-graded in school can tell you, it’s no guarantee of high-quality work. If all the students in a curve-graded class slack off, they don’t all get a D or F. Instead, it gets easier to get a B. If all but a few students slack off, it’s a good bet that the ones who don’t will get an A. Now translate that into the essentially cooperative workplace. Is it really a good idea to reward employees when their co-workers fail?
(If you know in advance that 10% of the employees in your department are going to get fired, one logical answer is to always keep a few redshirts around. This frees up the rest of you to stop worrying, and work on the stuff you were hired to do. Any good work you get out of the redshirts is pure profit.)
Stack ranking also fails to take into account what kind of work is being done. Sometimes fast-moving highly profitable achievements rest on an earlier foundation of slow incremental work on less-than-tractable problems. It’s not unusual for a department to do both sorts of work. Which kind gets rewarded for being productive? Which gets the bad reviews and firings?
I’ll absolutely question the use of stack ranking as a motivational device. Doing good work, looking ahead, helping to create a strong, smart organization, and refraining from doing evil should be enough to get any employee a good annual review. If what it gets them is a note in their permanent record saying it wasn’t enough, and they should have done more, they might feel motivated to try harder next year, and in a few cases may try harder the year after that; but mostly not, and sooner or later they’re all going to lose heart. People want to care about their work. If you break their faith in their job, it’s hard to win it back.
It’s a miserable system for managers, too. Say you’ve put together a great department — competent, well assorted, good work proprioception, with high productivity and high morale. Now impose a rating system that tells you that your department manages its people neither better nor worse than any other department. Be forced to label 20% of your people winners, without reference to the rest of the department’s work that makes theirs possible. Label 70% of them as timeservers and underachievers, no better than they should be. Label 10% of them failures, or even fire them, when you’ve spent all year trying to help them be good at their jobs.
I’ve managed employees. Having to do that to them would be a breach of trust. You’re the mechanism that applies a dishonest ranking system that hurts them. If an employee doesn’t already feel their performance warranted a bad review, there’s no satisfactory way you can explain to them why you left them out of the winning 20%, or the reasonably safe 70%. You may protest that the ranking system is flawed and arbitrary, but you’re still the person who arbitrarily ranked them.
Some articles I found interesting:
Vanity Fair Daily: Microsoft’s Downfall: Inside the Executive E-mails and Cannibalistic Culture That Felled a Tech Giant
Erika Andersen at Forbes: The Management Approach Guaranteed To Wreck Your Best People
IDC Analysts at Computerworld UK: Microsoft, Big Data and statistical idiocy. Stack rankings: An example of the misunderstanding and misuse of statistics
Steve Gall at Walden University, 2005: Human Resource Management: Issues related to the use of the stack ranking method in conjunction with the employee performance review for promotions and benefits
Eric Novinson at The Motley Fool: Stack Ranking and Value Traps
Julie Bort in Business Insider: Microsoft Isn’t The Only Tech Company Doing Forced Employee Ranking
Marco Chiapetta/Microsoft Insights at Network World: Microsoft’s ‘Lost Decade’ sensationalizes common issues among large corporations
Addenda:
Keith K. @46 recommends an interesting link:
One genuine question I have is how well [stack ranking] works with certain modifications. At Valve**, they use stack ranking, but it’s done by peers, not managers. I wonder if this changes the dynamic at all. Also, Valve explicitly cites teamwork as one of the components measured for stack ranking. I wonder if this changes the dynamic enough to erase the downsides of stack ranking.I’ve read some pretty darn amusing internal documents, but the Valve Handbook for New Employees is a goodie. Have a look. I’ve always taken it as an article of faith that your co-workers know you. I’m now chewing on Valve’s theory that you can run a company on that basis.**The game company responsible for Portal, among other classics. If you haven’t seen it, look at their employee handbook, which describes a ridiculous democratic utopia with no set internal structure. It is, by far, the most entertaining internal company document I have ever read.
===
Heresiarch @57 on blind competitive pressure as an evolutionary force:
One of the things it has consistently produced … is cooperation. Time and again, it’s turned out that in a cutthroat take-no-prisoners dog-eat-dog existential battle of all against all, the most winningest strategy is working together. What these competition-inducing schemes to improve upon the inefficiency of group production constantly miss is that group production originates in the first instance by out-competing everything else. Cooperation is where competition leads. Trying to use blind evolution to improve upon cooperative systems is like noticing that great square wheel you made is getting rounded on the corners from wear, and setting about sharpening them back up.(Emphasis mine.)
I'm thankful beyond words that this kind of nonsense was not corporate policy when I was a software development manager.
If I were forced to play this game, I'd try to label people who'd already been fired or quit as the underachievers.
I say this in the most gentle way I can, but... Doesn't the publishing industry "stack rank" writers by sales? Or is this a cri de coeur that the industry should stop that practice?
Or waited to fire them until after the evaluations went in?
Hal, that is so far off the mark that I'll have to think about it before I can even begin to respond.
One significant problem with this approach is that it rarely values catalysts and facilitators well enough.
Assume you have a team of eight top-flight developers. There's a good chance, at a competitive employer like Microsoft, that three of them are flaming egoists. Could be two, if you're lucky; not improbable that it's four.
If those flaming egoists are really productive during the year, it won't be because of their shining genius. It'll be because somewhere else in the team, either among the eight or somewhere on the side (QA, team leader, scrum master) is someone who is really good at smoothing feathers and soothing egos. (Or, when required, knocking heads together.) Otherwise, it's gonna be dominance fights all the way down.
The thing is that while they're facilitating communication among the egoists (Who, in this scenario, are genuinely good at what they do. They just don't play well with others.), that catalyst isn't doing anything else. If they're a team leader, or a scrum master, then they're probably OK -- making the team work well together is their job. But if someone else on the team is the one with a talent for facilitation, and is the one making the communication work, then they're not doing whatever else they're "supposed" to be doing. The team's work will be stellar, but their delivery will be in the tank.
Remove them as part of the failing 10%, and the whole team will suffer.
Hal,
I don't know what the publishing industry actually does, but it would have to be insane to 'stack rank' authors. Publishers have actual measures of sales for their authors, and they should be trying to publish all future books that will make a profit (and ruling out things that will be embarrassing in other ways)
In the just short of fifteen years I've been in high-tech, there were occasional mentions of implementing stack ranking.
It was generally limited to "It's the way they do things at FOO and BOSS is looking into it." I sometimes wonder if this was the management equivalent of a parent enthusing about the Smiths down the street, whose kids took cold showers and went to bed at 8:00pm and could only have two toys in their room, ever.
My current department is full of refugees from Intel, which has a similar hyper-competitive ranking system. They're glad not to be part of it any more.
Doing good work, looking ahead, helping to create a strong, smart organization, and refraining from doing evil should be enough to get any employee a good annual review
...and not one GODDAMNED thing more.
The article is on newsstands now. It's in the issue with Alec Baldwin on the cover.
Stack ranking is not the only problem the article describes. Microsoft invented an e-book years before Amazon and Apple, but was too creatively petrified to make such a different product. There is also the saga of Windows Vista, a case of bureaucracy out of control.
In response to Hal's post (#2)
1) Sales figures are a real (external) outcome not an internal assessment.
2) A publishing house's different writers don't have to cooperate as part of a team.
3) The publisher might drop writers that lose money or that they decide aren't worth their time, but they'd be foolish to use this sort of fixed (fire the bottom 10%) ranking scheme.
serge: just doing good (etc) and not one thing more should be enough, or it should be enough for just a good review and not one thing more?
My former employer implemented this sort of system some years before it was taken over (then disappeared as an entity). It struck everyone on my design team, including the managers, as a stupid idea at the time.
What I don't understand is under which circumstances is stack ranking a good idea. Maybe business management theory makes assumptions that hold true for some industries but not the one I work in? Maybe it makes more sense if you're trying to optimize the widget manufacturing? (Or maybe when you assume that Alice and Bob are equivalent to spheres of water that mass the same as they do?)
I will freely acknowledge I have no idea how publishing works from inside, either as a publishing employee or as a writer. I have, though, been reading a lot of works by both over the years, from at least Connolly forward.
With that in mind, consider the following quotes:
"Another way to look at stack ranking is as a mechanism for concentrating rewards on the people most likely to be hired away, or to be tough competitors when employed elsewhere."*^*
"Stack ranking also fails to take into account what kind of work is being done. Sometimes fast-moving highly profitable achievements rest on an earlier foundation of slow incremental work on less-than-tractable problems. It’s not unusual for a department to do both sorts of work. Which kind gets rewarded for being productive? Which gets the bad reviews and firings?
*^*
"Say you’ve put together a great department — competent, well assorted, good work proprioception, with high productivity and high morale. Now impose a rating system that tells you that your department manages its people neither better nor worse than any other department. Be forced to label 20% of your people winners, without reference to the rest of the department’s work that makes theirs possible."
Are auctions never held (or even sotto voce offers) to entice writers from one house to another? Has no one else heard of writers getting dropped because they don't write quickly enough, or generate enough sales, regardless of quality? Does that last quote sound like a bestseller list to no one else but me?
It may well be I am off the mark... But if so, I don't think it's due to lack of sympathy with the situation, or lack of exposure to publishers and writers talking about their business.
Hm. I meant, "no direct idea from inside..."
I worked at company that does ranking and rating, first as an individual contributor then as a manager. I left. Abi's point about it not rewarding collaboration is spot-on.
The process is that the 1st and 2nd level managers meet and review the "brag sheets" for everyone in the rank group. These are discussed and then ranked. The rank group may have people doing similar jobs who are different pay grades. If you are one of the higher pay grades, and you end up ranked below someone with a lower pay grade, this will affect your raise. If you end in up the bottom third or so, you could be in real trouble.
In one of these sessions, some of the engineers had been working on a high-profile project and had a lot of visibility. The voting was being done by a group of managers who didn't know everyone in the group, let alone the quality of their work. The votes worked out that the ones on the high profile project were golden and no one else was. This felt patently wrong but we mired down in arguing. A second meeting was scheduled a few days later. I was one of the first level managers, and I pulled the other ones into a small meeting and said "let's figure this out between ourselves and then present a list to the bigger group." We were able to agree on something more equitable. In the next meeting when we announced that we had a proposal, the higher level managers looked so relieved, and things were worked out quickly.
One of the people I worked with who was in a different department (and therefore different rank group) -- smart, knowledgeable, great team player -- told me that he was the only grade 8 person in his group. Therefore, he had to be ranked above everyone else based on his contributions, or he'd be dinged for a bad year.
One of my direct reports asked several people to give me feedback on him for his review. Two of those people were peers of his. One gave him a nice write-up. The other ripped him up. The one who ripped him up did not do any better a job than he did -- if anything, he was less technically competent and he was certainly less pleasant to work with. I was sure that he'd just done that because he understood that they were in the same rank group and I ignored his input. After I quit, I was appalled when the new manager made the backstabber the team lead.
I worked at the 'soft in the late 1990s and remember stack ranking, but in a less vicious way. I think we got trying-to-be-objective reports on whether we'd met our objectives, and what other employees thought of us; the raise-and-bonus pie divided up among `sibling' employees was fought over by our managers (so a great group's weakest player was supposed to get a small piece of a big piece of pie); and we were stack-ranked, but that wasn't the main determinant of pay and promotion.
Of course, it's reasonably easy to make your employees happy when they're all 28.5 and changing the world and the magic money firehose is on full.
shadowsong @ 11... Oh, I meant that doing excellent work should see you rewarded with more than the sweet little nothings whispered in your ear. :-)
I would also say that I agree stack ranking is probably ineffective, on its own terms. My point is, software ain't the only place it's been tried.
My feeling whenever someone suggests more internal competition would improve a company is "well why not break it up, then?" If divisions should compete rather than cooperate, then spin them off as separate companies. If employees should compete more and cooperate less, why not make them all independent contractors?
I've worked for several medium-to-large high-tech enterprises (Burroughs, which became Unisys, and Yahoo), and I've never seen a situation where a large company's problem was too much internal cooperation.
Relatedly, my in-laws talk about the "genius management strategy" that drove them to quit their original jobs, at a company where they'd been for decades.
They got some new managers who had heard of the idea that the bottom-performing 10% of employees at a large, established company are pure deadweight -- can be fired with no effect at all on productivity. So they fired the bottom 10%. This actually worked out okay to start with.
Then, a year later, they fired the new bottom 10%. And repeated again and again. With very little, if any, new hiring in between rounds of firing.
So people who had been productive were a) saddled with twice as much work, and b) terrified for their jobs in case they were determined to be in the "bottom 10%."
If you have a Gaussian distribution of productivity (which is a questionable assumption anyway, which I'll get to), and you chop off the left tail, the remaining employees aren't likely to redistribute themselves to form a new left tail. You're going to start chopping into people who are actually good employees, very quickly.
It's questionable whether employees' ability is likely to follow a Gaussian distribution in the first place -- because employees are not a random sample. In order to get hired, they had to show more than some minimum level of ability at their jobs.
In fact, some research shows that ability tends to follow a power law distribution, rather than a Gaussian distribution. That is, most people are at one level of performance, and then there's a long tail off to the right containing the few superstars.
I think what's missing in the article I linked there is the effect of selection. The research looks at groups like professional baseball players, Congresspeople, movie stars, and recording artists. I think there's pretty obviously a minimum level of ability to be considered part of each of those groups. So it makes intuitive sense that if you look at people in those groups, you'd get a lot of people who perform close to that minimum cutoff point, and then a long tail of people who perform better. The people who perform worse than the minimum cutoff point are excluded from analysis, because they never get a pro baseball contract, never get elected to Congress, never get cast in a feature film, etc. Imagine taking a Gaussian distribution and chopping off the whole left half of it, or even more -- you're really only even looking at the right tail.
I'd suspect that at most companies -- especially companies like Microsoft -- the selection process to work there is already such that pretty much everyone working there is in the right tail of ability/productivity/performance; otherwise you won't get hired in the first place. There is no reason to assume that any team made up of these employees necessarily must contain anyone with bad performance. Instead, it makes sense to assume that most people's performance will be very good or excellent, and a few people's performance will be beyond awesome.
And no, it doesn't work to say "Well, then, anything less than beyond awesome is UNACCEPTABLE! Everyone should be doing beyond awesome or they're just slacking off!" Because with different projects, different roles, and taking on new challenges, different people may be the "beyond awesome" performers at different times. The distribution of performance measured by a single metric, at a single performance review, doesn't capture that.
I second #11.
John Chu, some of it was the belief that competition always made everything better. At our annual rah-rah meetings, we were often reminded that we were all working with no tenure -- anyone could quit or be fired in any two weeks! -- and most of the crowd would cheer. Even at the time, I found that a little odd, but the attraction was that everyone there was the winner in a tough race.
The direct answer to the question posed in #2 is that no, successful publishing companies do not choose which authors to publish by any such simpleminded formula. You see all kinds of cases where unsuccessful authors have been given more chances, been encouraged to try different approaches, been repackaged and promoted to different audiences, and so forth. Editors and publishers will frequently tough it out through a lot of flops when they have a gut sense that this writer might produce a bestseller eventually -- or, at least, find a long-term profitable audience. And yes, sometimes publishing companies throw up their hands and say "This didn't work and we don't know what to do about it, good luck with your next publisher." But we don't simply slice off the bottom 10% and double the marketing budget of the top 20%.
However, the more important point to make about comment #2 is that its attempt at an analogy fails because most publishing companies are not in fact associations of authors attempting to help one another to a common goal.
Teresa is not saying that it's always, in every situation in life, wrong to "high-grade," to drop less successful ventures and focus on more successful ones. She's saying that it's borderline-nuts to run an organization by constantly firing the lowest-ranked employees in each department or division, even if the department or division in question is wildly successful.
She's saying that's nuts because doing so creates insanely perverse incentives -- it corrodes all sense of shared purpose, as employees within a division turn their energies away from creating successes that make everyone look good, and instead focus on behaviors that make themselves look good and their colleagues look bad.
And it's also nuts because you wind up losing some of your best employees. The "bottom 10%" in the division that's doing 1000% better than all the others? You probably don't want to be too quick to fire those folks. (Even leaving aside all the questions about just how accurately one can really reduce anybody's "performance" to a single percent rating.)
John Chu @ 12: What I don't understand is under which circumstances is stack ranking a good idea. Maybe business management theory makes assumptions that hold true for some industries but not the one I work in? Maybe it makes more sense if you're trying to optimize the widget manufacturing? (Or maybe when you assume that Alice and Bob are equivalent to spheres of water that mass the same as they do?)
There is a genuine problem that ranking is intended to solve. It just creates new problem(s) when it solves it.
One of the things a company wants to achieve in promotions and pay is internal equity. If Alice and Bob are doing work of similar complexity with the same degree of quality and productivity, then you want them to be similarly rewarded.
I know someone who was a first level manager at a company working on project ABC when project XYZ was cancelled. Developers from XYZ were added to her team. Prior to being cancelled, project XYZ was the hot thing with all the cool kids. Six months later, when she got the paperwork to do reviews, she was horrified to find that the people from XYZ were all at least one pay grade higher than the folks originally on ABC. They did not have higher academic degrees, more years of experience, or better performance -- apparently, there had been promotion inflation in XYZ (or maybe stagnation in ABC). If she gave them all similar reviews (their performance was similar), then the original-ABC folks would get raises, and the XYZ folks would either get no raise, or have their pay cut. That's because you had to do more to justify your pay if you were a higher pay grade. This was years ago, and I cannot remember what she decided to do. I do recall that she found a job at another company shortly after this.
By creating rank groups of people doing similar jobs, and then having managers from different areas agree on their performance, this prevents "pockets" of higher, or lower, standards and rewards. So you improve internal equity in pay, while discouraging teamwork.
Nearly a dozen other comments went up while I was writing #22, and some of them make the same points with considerably more economy.
To Hal's #13, certainly there are plenty of cases of publishing companies making decisions based on crude "shoot the losers, promote the winners" criteria. And sometimes that "works", for a while at least.
I seriously don't think Hal gets what Teresa is arguing. She isn't saying this kind of practice is bad because it's ruthless and mean and we should all be nice. (Although, mind you, I do think very few of the world's problems stem from people being too nice.) She's saying this kind of practice is bad because it sets up crazypants incentives toward war-of-all-against-all competitiveness inside organizations where, for the organization to succeed, people actually need to be motivated to cooperate.
Whatever the rights and wrongs of publishers and how they treat authors, it's an issue that says nothing to Teresa's argument, because a given publishing company's authors aren't in fact in the same situation. There isn't anything like the urgent need to motivate cooperation between authors.
Hal, there are a lot of different ways to sort jellybeans, and I'm not talking about most of them. Stack ranking is a fairly specific employee evaluation technique and management approach.
Cisco's been doing the same thing -- went from being a great place to work to a terrible place to work. They laid off all the tech writers in my technology group -- then hired some of us back as contractors (I'm happy because I'm part time, and out from under a terrible manager). It's insane. There's no institutional memory. There's no one doing actual work. There are only mid- and upper-level execs.
I hated being punished for being efficient. When I was secretarial, I finished all the work in my department, so somebody up there figured I should be given the work that slower secretaries hadn't managed to get done. That certainly taught me to slow down.
I was also punished (with boredom and lectures) for having all my work done and not being seen with my fingers constantly in motion. I tried playing chess with the computer, and got back a report that I'd been seen playing games. "But when we got the new system, you told me to play with it." "That's not what I meant."
So I started hanging out with a bad crowd. Adjuncts...
Stacked - or forced - ranking became popular after being widely publicized by Jack Welch at GE.
Keep in mind that all large institutions are populated by staffs with two distinct, opposite and conflicting goals: execution and support. The former tends to value results, a "git 'er done" attitude and expects talent to exercise judgement and managed risk taking. The latter values process, policy and institutional survival over free-will. It really shouldn't be a surprise at all that forced ranking emerged from the support side of enterprises - it's the perfect means for removing as much leeway as possible from the evaluation process.
FR is a blessing from on high for weak managers - rather than work with poor performers and/or admit you made a bad hire, you just wait until the next round of rankings and let "the process" take the decision out of your hand. It's an awesome tool for evil managers. I am not going to pollute my karma expounding on that here - either you're already thinking of ways to exploit this or you are not capable of doing so. For great people managers, forced ranking is the worst "process improvement designed to enhance long-term productivity and effectiveness throughout the entire corporation". You have no leeway in the rankings, it becomes impossible to build high-performance teams, you're now explicitly competing against your bad and evil peers, and there is absolutely no way to shield, deflect or soften the impact (like you would for most other directives).
(Welcome back, Lance... Been a long time.)
Lance, I think I'm a good people manager -- I wouldn't claim to be a great one -- and what I knew the minute I read about stack ranking was that having to do it would break my heart.
If you can't bring yourself to do a fast, merciful, and (insofar as possible) constructive firing when it's necessary, you don't deserve to be a manager. You will not look good. The fired employee will not be understanding, and his or her friends will not support your decision. I recommend doing it before lunch so they can all go off together, talk trash about you, and come back tipsy an hour and a half late.
Stacked ranking comes from Jack Welch? Well, that makes sense (and explains a lot about my experiences with GE Lighting as an outside contractor).
I can see this kind of thinking coming out of Conference Room Mind. I'm sure some group somewhere came up with, "Well, how do we make teams more successful and apply the continuous improvement process to HR?" (Probably not in the same sentence, but I'm sure both concepts were in there) The obvious solution is "competition breeds success" so we'll not only make team compete against each other, but have that competition internally. And we can create better teams by removing the dross at the bottom in a bubble sort kind of way.
For some reason stack ranking reminds me of the Shadows from Babylon 5... which, perhaps, bodes not well for those managers....
I'm now picturing John Cleese ordering two pantomime horses to fight each other to the death for the single available job in a merchant bank.
abi @ 5: Jerry Weinberg somewhere tells the story of a software company that was trying to figure out why 40% of their projects failed, and what to do about it. At some point they were talking about firing one developer, because no one could think of anything spectacular she'd done. Then someone pointed out that no project she ever been part of had failed...
When I've talked with my colleagues at MSFT about the stupidities in their management processes, none of them have complained to me about stack ranking. Instead, I've heard about what sounds to me like an absolutely insane three-branched reporting hierarchy that runs all the way up to the level just below Ballmer.
This actually goes a long way toward answering a question I've long had about MSFT: how the fnork do they consistently and reliably manage not to turn any of the crazily amazing work of Microsoft Research (MR) into great products? During the 90s and for much of the 00s, MR put together a stable of truly impressive post-doc C.S. researchers. What happened?
Now, for every single technical greatness that leaks out of MR and into a crippled sideshow feature of a deprecated product, there are another ten great achievements that sit unloved and unused in the MR portfolio while their scientists appear to be nose to the grindstone cranking out the next ten things that will be roundly ignored by MSFT product engineering.
Yes, stack ranking could certainly be making that worse, but I think you have to reach for something more fundamental to explain why product managers with a powerful incentive to compete with one another aren't aggressively turning the work MR into viable products. My colleagues at MSFT won't say this, but I suspect they would secretly agree the main problem is that Ballmer is a goddamn waste of oxygen at this point, and nobody knows how to overcome that problem short of assassinating him. How do you replace Ballmer, when the three-legged crawling horror of a management structure underneath him is a freaking game of Diplomacy?
General Electric was the first company I read about which did this, particularly when Jack Welch was CEO. Others tried it as well. (2007 CBS News story)
The comments which mentioned GE ahead of me hadn't been posted when I started looking for confirmation elsenet. That's my story, and . . .
Having looked through Teresa's links, two points stick out at me:
1) Even the guy who wrote a book about how stack ranking is totally awesome and more companies should do it argues that you should never have stack ranking as a permanent program. He thinks it's best used for 3-5 years as a way of clearing out the cruft of the organization, which is decidedly not that way Microsoft has been using it.
2) There's a fundamental theoretical problem with stack ranking which suggests to me that it can't ever work as intended. Basically, the problem is that it doesn't work when the group to be stack-ranked is smaller than about 30 or larger than about 10. (Which is to say, any group of any side.)
If the group is smaller than about 30, the statistics simply don't work out. Expecting small groups to shake out nicely into the stacking scheme is like rolling a six-sided die six times and expecting to roll each number between one and six exactly once. It just doesn't work that way. The result is that you actively penalize workers for being on high-performing teams, and encourage managers who care about their high-performing workers to keep a rotating cast of no-hopers on the payroll to receive those negative reviews.
But if the group is larger than about 10, then it's no longer a case of a single manager sitting down and ranking his workers, (And if it is, the manager is most likely operating from a position of profound ignorance as to what the people he's ranking actually do.) It's much more likely that a committee of managers has to sit down and fight it out over whose workers get the As and whose get the Fs. And that means that the biggest single determinant of a given employee's grade is not his job performance, but his manager's clout. And if that's not a recipe for political infighting and backbiting, I don't know what is.
@Chris W: "...it's best used for 3-5 years as a way of clearing out the cruft of the organization..."
I'm in my cups, so allow me to lift the veil on what I believe is a collective myopia in perceiving this as just another Dilbertarian idea and admire the absolute unadulterated evil genius of forced ranking.
Imagine if a CEO wanted to trim 10% of the staff using traditional layoffs - every year.
* Layoffs are for companies in a death spiral. Long-term share price and market perception change significantly based on this change in competitive position.
* CEO's are vilified for layoffs and become a convenient fall guy for the board.
* Layoffs are highly visible, generate enormously bad PR and really tick off local politicians and activists.
* Layoffs require expensive separation packages, vetting by Legal and HR, potential regulatory scrutiny, etc.
* Layoffs are far more damaging to corporate morale and employee psyche than forced ranking.
So, that's why we have forced rankings. They are a method for performing mass layoffs and/or downsizing without any of the external negative impacts. All damage is confined within the walls to the easily replaced meat puppets.
(PS Hi Serge!)
Companies have been doing forced ranking since Enron, at least. I remember when the utility that I used to work at brought it in. We were told that Enron was eating our lunch so we had to adopt their methods. Ironic, n'est pas?
Hope I'm welcome here in the living room.
I come at this topic from 2 sides. As an MBA and a victim of force ranking.
As an MBA it was brought up in the HR and labor economics classes as a way to do things. And was usually pointed out by the professor as a piss poor way of doing things. Mainly because you got the behavior you gave incentives for. And this incentivized ass kissing and shifting blame.
As a victim of it, I worked for SkyTel the paging company back in the day, hired before Bernie Ebbers and WorldCom gobbled it up, then gobbled MCI and screwed the pooch with the FCC. After that, MCI began to integrate everything uniformly. Which meant my department (pricing if you must know) with 4 people in it, including the manager, had to have the 20/70/10 rule. Which means all of us did OK, none did great and no one got fired.
After that I decided I had no reason what so ever to bust my ass for MCI for any reason. And look elsewhere for a job (helped along by looking at the company's financial assumptions for the next year (i.e.: nowhere near reality)).
This is the first time I've heard of this. The use of the word downfall reminds me of a certain movie scene with subtitles. I can see Balmer asking, after it's too late, "What would Gates do?"
But I've heard of something similar-but-not at IBM - the company would have several teams, each working on its own version of a product, and the team with the best product got to see it go to market (as well as perhaps being rewarded in other ways such as bonuses). Important differences are the people within a team were NOT competing with one another, and the losing team(s) were NOT fired. IBM has historically been reluctant to fire employees, resulting in loyalty between the company and employees that goes both ways.
So, tying it together:
A child, intensively trained to view life as an individual competition, goes on to create and run a company. A company culture develops, attracts like-minded leadership, and sustains itself.
A personnel culture which creates intra-team competition, sacrificing cohesion and retention. A management culture of infighting and divisional warfare, hobbling the company. And a headquarters culture of systemic illegality, and indifference to the greater good of the industry, and of society.
Stack ranking wasn't a management mistake. It's who Microsoft is.
All about winning, interpreted narrowly, without heed to common good.
abi@5: Microsoft has never valued facilitators or people who would spend their time preventing or resolving arguments. I've known two program managers who were really skilled at it; both of them were driven out of the company eventually. (There was also some sexism involved, so I'm not sure how much of what led to which.)
Blaming stack ranking for Microsoft's last decade is kind of misleading, because they've been doing it since at least the mid-1980s. What you CAN blame is the Infinite Piles of Money they used to get from MS-DOS, for papering over all kinds of stupidity until about 1998. Windows lost money for seven YEARS before they ever got anything back, "LAN Manager" for about ten (it got folded into Windows NT, which finally broke even around version 4.0), that Word and Excel for Windows spent three or four years before going anywhere, that there were all kinds of failed products out of Redmond during the 80s and 90s. Microsoft was never a smart software company; it had more margin for screwing up than its competitors.
Now that there are viable, mass-market alternatives to Microsoft's operating system, it doesn't have as much margin.
Stack ranking strikes me as the most efficient way to break apart a good team. It forces a level of competitiveness that will inevitably turn toxic.
My wife (caroline) mentioned it further up the thread, but I got to hear lots of stories from my parents as a teenager about what this kind of system does to the work environment. The company they worked for would repeatedly lay off the bottom (I think) 15% or so of low performers. Since they quickly laid off all the useless people, they soon had to move on to people who were actually useful just to hit some metric. The work environment very quickly turned extremely toxic, with everyone constantly fearing for their jobs. Thankfully, My parents were luckily in a position to opt out (one was able to take retirement and the other soon openly asked his manager to be in the next round of layoffs, since the compensation would be better than just quitting) and moved us across the country.
One genuine question I have is how well it works with certain modifications. At Valve**, they use stack ranking, but it's done by peers, not managers. I wonder if this changes the dynamic at all. Also, Valve explicitly cites teamwork as one of the components measured for stack ranking. I wonder if this changes the dynamic enough to erase the downsides of stack ranking.
**The game company responsible for Portal, among other classics. If you haven't seen it, look at their employee handbook, which describes a ridiculous democratic utopia with no set internal structure. It is, by far, the most entertaining internal company document I have ever read.
As I mentioned in my comment on stacked ranking in the current open thread, I don't believe this was the reason for Microsoft's great failures of nerve and/or innovation. I lived through rating and ranking at Intel in the early 1980's; it's one of the reasons I finally left Intel. But while it was bad for morale and did result in some rather evil management decisions, it was not the downfall of Intel in any sense, and I did not see major changes in the amount or quality of innovation as a result of it. The problems Intel had at the time were primarily ones of growth and communication between divisions. The major problem of innovation was something that has always been true at Intel: the chip designers were the rock stars, and they made all the important product decisions, even when the relevant requirements were not at the component or chip level, but were system requirements.
My take on Microsoft's innovation problem is that MS culture seriously flattered the egos of MS technical and marketing personnel, telling them that they were fundamentally better than their peers at other companies, and that therefore MS's decisions must always be right, and other companies' decisions must be wrong. This tended to encourage group think and discourage analysis of the actions of their competitors, or of changes in their product markets.
As for why MS didn't use any of the really good work of their own Labs, I've commented on that here before based on my experience at Tektronix Labs, and my discussions with colleagues from Xerox PARC, HP Labs, and IBM's T J Watson Labs. Briefly, there is a fundamental tension between corporate research divisons and product development divisions. Product development is a source of bottom line profit, but research is an expense. If an entire division is dedicated to research, development divisions have to pay some part of their profit to meet the research division's expenses, which makes the developers reluctant to encourage research they don't see an immediate use for. In addition, researchers typically have post-graduate degrees, most often doctorates, and are seen by many development engineers as academics, impractical theoreticians. So there's a lot of skepticism among developers about the usefulness of the work that comes out of the research.
The worst part of the system is that it actively hides problems, I think. (Take as given for a moment that the 20-70-10 system is wrong.) In particular, in a system which doesn't mandate that distribution, it's easy to see that, if group has all underperforming reports, maybe the problem is the group -- or maybe the problem is the manager.
Technology companies, especially in their early days but still today, have always needed to be a little cannibalistic, constantly competing with their old selves to find the next big thing. Everybody knows the stories of tech companies too wedded to their old business models to try out and then embrace the new, who missed the boat on major revolutions. (IBM, we're looking at you. Maybe this is an example of the problem of too much cooperation?)
And indeed Microsoft is one of the companies I cite as doing that well -- the 1996 launch of Internet Explorer was, at the time, a huge pivot for the company which kept them relevant at a time when it wasn't clear to everyone that this weird Internet thing was ever going to be big, and the successful launch of the XBox is another case of them doing something which panned out big at the potential cost of market share in other areas. I wonder if one of the things stack ranking is designed to prevent was that stagnation. (j h woodyatt @36's "why product managers with a powerful incentive to compete with one another aren't aggressively turning the work MR into viable products.")
Unfortunately the stories of the pivots Microsoft didn't make, to protect Windows and Office, and Microsoft's current stuck state, suggest that it's not a panacea, if indeed it helped at all.
How you balance that need to encourage competition within the company, to keep building revolutionary new things, with the need for cooperation within the company, to keep everybody pushing together for the collective's gain and not just for their own, is a problem I think we're going to keep coming back to.
I wonder if stack ranking is part of where the "10x developer" idea comes from. To explain: if you spend enough time around Hacker News or other places that startup-types hang out, you will eventually come across the idea that a really good developer is 10x, 100x, 1000x as productive as a bad developer, and the goal for your company is to attract as many of the 1000x developers as you can. Thus you will see job boards littered with postings for "rockstar developers", "PHP ninjas", etc. ad nauseum.
(One wonders how many innocent young startups have advertised for rockstar developers and been lucky enough to actually catch one, only to discover that rockstar developers also have the rockstar personality and ultimately the rockstar wandering eye, and their catch was ultimately enticed away by another innocent young startup, leaving a sadder but wiser startup in its wake. To this day, there are no brown M&M's in that startup's candy bowl...)
I think the idea that developers are either gods or they're shit is an immediately poisonous idea, and bad both for the profession and the industry, but it's got deep roots, and it plays into a lot of things that developers like to believe about themselves and what they do, even as they're scrambling to keep up with the people on Hacker News and secretly afraid that they're one of those dreaded "1x developers". Dunno what the counter is.
Kevin @48: Let me comment on this "10x better developer" thing from the perspective of having just spent a couple of weeks being one of those 10x better developers.
Because that's actually the way in which it works. My long-term colleague and I (and some other team members who are irrelevant to the story) have been working on a bit of program that neither of us work on a lot. It's arcane and dirty and grew organically with only a modicum of design and is nearly self-modifying. I set up a list of work items for us to do and some time estimates on them, and a very rough approximation is that my time estimates have been off by about 3x for both of us -- in opposite directions. So, to a rough approximation (and with the caveat that the story is oversimplified for the telling), I'm doing this stuff about 9x faster than he is.
So, yeah, I believe in 10x performance differences in developers -- on a particular limited project. It happens all the time.
But in this case it doesn't mean I'm a better developer than he is. I'm not. There's stuff where he'll be 10x faster than I am. We're both good programmers. But the particular sort of messy code that we've been working with lately happens to agree pretty well with me and I can do it pretty quickly, whereas it drives him batty and he gets stuck. Next week, he'll probably be the one running circles around me.
(The real key here is that, when people's skills vary that much -- and we're not anywhere near unique -- the winning move is not in finding a pile of independent rockstars who can maybe all do killer drum solos, but in finding people who can work together to make a good band. And, because the band metaphor only goes so far, you want people who can help the people around them be good.)
But it's so butch, so [Herbert ]Spencerian, and it makes managers fell like they are cruel-but-fair sharp operators...surely these benefits more than make up for its not doing what it says it were there to do?
The first company that hired me, when I started there, had been split a few years before into separate divisions that "competed", and by "competed", I mean "the division with the most clout dictated terms of work to the other divisions". (Apparently the split was done because it was the thing to do. This is generally a terrible way to do things. Exception: if everyone you see is running full speed in one direction. Run first, figure out later.)
It was a great system if you were in that division. I was not. Lavish Christmas party for the haves, December layoffs for the have-nots. It contributed to a poisonous environment that persisted long after the divisions were merged. (It was still around when I "left"; it may still be around now, but most other people I knew there have also fled the ship.)
I've not, to my knowledge, worked at a company that used stack ranking per se, but I am aware of at least one place where the manager basically adopted the 10% philosophy. It had the effect of pretty much gutting his group, partly because he wasn't able to tell which were truly the "bottom 10%". Of course it was a blessing for anyone designated as such; they were free to find work at a company with a manager who was capable of doing the job properly.
Though the conversation has moved on a bit...
Hal, the thing that you aren't getting is that authors are not employees of publishing companies. Authors are the product.
Years ago, I read a story about what happened when chickens were evaluated by stack ranking - the top egg layers were kept and the hens that produced the fewest eggs were sent to the fryer. This did not produce a super-successful farm; it produced a farm full of chickens that attacked each other and destroyed each others' eggs whenever possible. If I find the cite, I'll post it (perhaps it was in a book about Enron).
Andrew, I've read that story, too, but the best I've managed for a cite so far is that it probably isn't in an essay by Paul Graham.
Meanwhile, Hard Facts, Dangerous Half-Truths And Total Nonsense: Profiting From Evidence-Based Management has a lot about how hard it is to develop an incentive scheme that doesn't destroy cooperation.
I know someone who has spent most of the year selling double-paned windows to homeowners. (It's practically a spring ritual in the San Francisco Bay Area: is it March? Then annoying young men with clipboards will soon be knocking on your door to sell you home security systems or double-paned windows.) The windows he sells are, in fact, a good product, but the package (windows/installation/warranty) is stunningly overpriced. He gets a salary plus a commission, but what he doesn't have is job security: he can be fired any month he doesn't close his sales at whatever percentage seems good to his manager, who can also be fired any month the people he has working for him fail to max out their sales at whatever pre-determined rate the company requires.
The whole thing strikes me as morally pernicious, but it works in the short term: if you sell a lot, you earn a lot for however many months you can stand it. Salespeople who do well work 14 hour days until they burn out, but they leave with a lot of money; salespeople who do poorly are fired within two months, or even sooner. I had never heard the term stack ranking before I read Teresa's post, but this seems to me to be a clear example of it.
Yet another thing stacked ranking (and other ranking systems with pernicious reward/effort scales) will do is invoke the in-for-a-penny effect. If you're in a group where the stars are known and you're not them (or just where stardom is uncontrollable) then the sensible thing to do is calibrate your effort to just above redshirt status. Your stars will thank you for the lack of competition as your co-workers do the same.
I'll not knock evolution: blind competitive pressure has created some pretty amazing things over the past billion years. One of things it has consistently produced, however, is cooperation. Time and again, it's turned out that in a cutthroat take-no-prisoners dog-eat-dog existential battle of all against all, the most winningest strategy is working together. What these competition-inducing schemes to improve upon the inefficiency of group production constantly miss is that group production originates in the first instance by out-competing everything else. Cooperation is where competition leads. Trying to use blind evolution to improve upon cooperative systems is like noticing that great square wheel you made is getting rounded on the corners from wear and setting about sharpening them back up.
Lizzy L @55: Actually, that doesn't sound like stack ranking to me. It sounds like if everyone on the sales team hits their target, they're all fine, and if no one does, they're all fired. Stack ranking would mean "Well, you sold 17 doors today, but everyone else on your team sold 18, so you're fired anyway," and then on another team the person selling 2 doors would be getting a bonus because everyone else sold 1 or none. It's a completely different form of pernicious management techniques.
Fade Manley, as I understand it -- I didn't say this in my original post -- the competition among salespeople is fierce. Only people with the highest percentage sales numbers are retained, AND you have to keep upping your percentages to remain on the job. So if the first week you close 30% of your sales and are allowed to remain, you must close 40% the following week, and 50% the week after that. If it isn't stack ranking, seems to me it's close enough...
beth 52: Hmm. Vendors, I'd say. Suppliers of raw material that publishing companies process into their ultimate product. Or do you mean something different that I'm not getting?
A long time ago, I went to work for a good-sized company that had just been acquired by a large multinational that used stack rankings. The first year worked pretty well (particularly for me), but after that I began to see some significant downsides. If you happen to hit a bad patch--get stuck, or lose faith in certain project management shibboleths--then no matter how well you have performed earlier, you end up with the bad end of the stick. And once you've done that, then the perception of how you're doing is colored by how you've *been* doing; IOW, you won't get a good break and be able to recoup.
Further, once everyone has figured out how the system works, the spirit of, umm, individual competition overwhelms the team ethos, and more time is spent on indirect Dilbertian psychological sabotage than on actually kicking the whale down the beach.
After three years, I was more than ready for a change, and even the fast-mo disaster that was my next job seemed a relief.
Sea captains from the Napoleonic era who weren't Jack Aubrey used to pratice stack management--they'd have the last man off the mast flogged . . .
Lizzy, #59: What that sounds like to me is the same philosophy that's enshrined in NCLB. Underperforming schools are closed down; the ones that make the mark MUST improve their testing scores every single year, or they too are labeled "underperforming". The only difference is that, unlike public schools, there's a never-ending supply of new suckers salesmen.
One point of clarity that some commenters seem to be missing: "stack ranking" isn't just about ordering your employees by performance as part of the evaluation process, which is a completely reasonable thing to do as part of a more involved process. It's about doing that ordering and then applying a pre-specified filter for what percentage of your employees get rewarded or punished, regardless of the actual performance of the people in question.
Dropping authors due to low sales is not, of itself, stack ranking. Firing salespeople for not meeting a given quota is not stack ranking (because the firing threshold is defined in terms of actual performance metrics, not as a constant percentage of employees).
Anyway. From the stories I've heard from ex-MS folks, it gets worse - because the stack-ranking process is applied again to merge the results from multiple teams within an organization, and that is a process of politics among the managers within the organization. The specific incident that was related to me was a three-team organization where two of the managers agreed ahead of the ranking meeting to rank each other's employees highly, which left the third manager (who was apparently new to the org) with an entire team of people at the bottom of the org's ranking stack.
The two managers may have been doing the best thing they could for their respective teams, but any system that not only allows two people to collude and screw over an entire team of people but actually makes it optimal behavior is utterly reprehensible.
kevin @48: The 10x developer thing was in The Mythical Man-Month (he found a factor of 30x, that's why he organized "surgical teams" around them). Chapter 12 of Robert X. Cringely's "Accidental Empires" expanded on it. I wrote a column series about it a decade ago (linky: http://landley.net/writing/#3waves ).
Yes, 1000x programmers exist. Here's one: http://blog.smartbear.com/software-quality/bid/167059/Fabrice-Bellard-Portrait-of-a-Superproductive-Programmer
The downside of a 10x (or 100x, or 1000x) developer is they get writer's block and go off on tangents. (In software, Linus Torvalds put Linux development on hold to write a new source control system. His once second in command Alan Cox switched his widely read development blog to welsh, and took a year off out of the blue to get an MBA.)
The better they are, the more susceptible they are to sudden tangents. They have an artistic temperament and are _bothered_ by doing it wrong. And they drop what they're currently doing when they come up with a better (or simply more compelling) idea.
Stack ranking eats these guys alive.
lizzy @55 Nah, that's not stack ranking, because it's not a zero sum game. There's no interaction between salesbeings and no obvious way sabotaging somebody else makes your situation better.
Stack Ranking is musical chairs meets the prisoner's dilemma: somebody is guaranteed screwed, how do you game the system to minimize damage to yourself?
lizzy @55 Nah, that's not stack ranking, because it's not a zero sum game. There's no interaction between salesbeings and no obvious way sabotaging somebody else makes your situation better.
Stack Ranking is musical chairs meets the prisoner's dilemma: somebody is guaranteed screwed, how do you game the system to minimize damage to yourself?
heresiarch #57: And indeed, stack ranking produces cooperation against the company: paul #56 gives a typical example, Darth Paradox #64 gives another.
Definitely not what a sane company wants....
I worked (briefly) for Intel during a period that included two massive layoffs. The effects of the forced ranking where overwhelmed at that time by the layoffs. The second layoff (when I got cut) was conducted using a "skills evaluation" metric that was supposed to measure skills people had vs skills the company needed, and was purportedly to avoid any effect of seniority on the cuts. Yet, funny thing, in my (larger) group nearly all of the layoffs ended up being among the more recent hires, including me (one of the newest, since the layoff period also included a hiring freeze).
On the positive side, my current company (Autodesk) has just switched away from a forced ranking model. We're in the first year of this new system, but it is structured to focus more on individual accomplishments and less in a comparative/competitive mode, so I have hopes that it will work well.
While Microsoft was functional, stock options rewarded us for cooperation; we all knew that if the whole company did well we'd get our piece of it. When enough people said `X is really boring and unglamorous but it needs fixed', fixing it got props. There must have been more peer input to the employee rankings, as described at Valve in #46; and the company was desperate to hire, not to fire.
I used to say that being a Microsoftie in the US was a lot like being a USian in the rest of the world. It was easy to mistake inherited wealth for personal virtue, and to mistake caution and fear on the part of our neighbors for awe and respect. And then the shadow came...
I do think about how to reward cooperation within a competitive enterprise. It can't be stock *options*, because those only pay off in the boom. Vesting slowly in stock that pays dividends makes sense to me. Maybe that would be a decent pension scheme, for companies big enough to pay their own pensions? It seems possible that long-term and retired employees would eventually be a shareholder voice with a useful perspective. Are there US companies that pull this off? (Bob's Red Mill?)
I suspect my company has decided to use some form of stack ranking -- my supervisor mentioned last year that HR was forcing all departments to adhere to a fairly strict "bell-curve" type of ranking. Even teams like ours, with 5 employees.
I can say, at least in my own case, that my morale has been pretty damaged.
I work at Big Blue and it has been this way for years. As Lance @40 said, it is used for continuous small layoffs that are not made public, ie it's not a layoff, it's firing underperformers. Cycle in cheap new hires (US or overseas) and remove higher paid employees. Grades are across large organizations, so it's not just how well you did locally and are known, but also how important your small area is in the big picture. It's obvious what that does to morale and quality of product, but when the goal is easy "not-layoffs" to cut costs and raise stock value, that's acceptable to execs.
"Stank Racking."
Sorry, I just had to type that out to see what it looked like.
A copy of "Mythical Man Month" was left behind by a coworker moving offices. Worth reading? (versus reading this book about the Linux Kernel?)
Stefan Jones @73, I'd recommend reading The Mythical Man-Month. Take it with a grain of salt - I don't know that his proposed solution to the issues is necessarily the correct one. But his identification of some common stumbling blocks is sufficiently dead-on, and sufficiently generalizable to projects of many types, that I still quote the book occasionally even though it's been probably 30 years since I first read it, and I no longer do software development professionally.
Stefan Jones @73: I suspect that The Mythical Man-Month may still be relevant when the Linux kernel is a historical footnote (though I expect the latter to remain relevant for a long time).
I've never understood why so many companies consider annual or semi-annual performance reviews so important. I understand the need for feedback, even if as a neurotic introvert I find it extremely uncomfortable. But the times I've gotten negative feedback in a performance review, my second thought (after my first thought, "no, you're wrong, I'm perfect!") was always "if you didn't like what I did 6 months ago, why didn't you tell me then, so I'd stop doing it!?"
OtterB @ 74:
One of my major gripes with with the software industry is that many of the problems identified in The Mythical Man Month are still there, still largely unnoticed by the managers who are perpetrating them, decades later.
Actually, there are some new problems too, so I guess we can't fault the innovation of the industry :-( Ask me about Agile Methodology sometime if you've got a couple of hours to kill and don't mind seeing my head explode at the end. (shorter me: It's actually a good idea, but 99% of the people using it shouldn't be).
Mythical Man Month is applicable way beyond software. Some of the descriptions apply (as with Augustine's Laws) to almost any enterprise of a certain size. His prescriptions are old-fashioned, but their main use is in helping to recognize that certain kinds and sizes of project just aren't going to happen. "If it fails when you do that, don't do that."
The Mythical Man-Month describes wicked problems that are probably innate to software development.
Who Moved My Cheese? is a sign that wicked people are going to cause you problems (e.g., stack rank layoffs).
Brooks' The Mythical Man-Month is one of the classics. Strongly recommend reading it if you haven't already, even if you aren't in the computer field.
(If you are in the computer management field and haven't read it, drop whatever you're doing and read it right now!)
I had a boss once who summarized the "man-month" mindset as "the belief that nine women can have a baby in one month."
clew @78:
Back when WMMC came out, I was fascinated in a sort of trainwreck way that anyone could even imagine the title question to be illegitimate. Since then, of course, things have only gotten worse.
Hmm, maybe that's what we should call Mitt and his fellows: they're not job-creators, they're cheese-movers.
Mythical Man-Month, Peopleware, and The Psychology of Computer Programming are excellent reads for folks who work in the software industry, especially for people who manage developers.
In response to Jeremy Leader, comment #75: I think annual or other such regularly timed reviews are pretty useful as adjuncts to immediate feedback and as opportunities for thinking about and discussing subtler patterns. But annual reviews absolutely should not be replacements for immediate feedback. And the self-review aspect is pretty important, too. As a manager and as an individual contributor, I know from experience how much hard workers often concentrate on our day-to-day, month-to-month goals, and need reminders to look up and think about the longer term.
I'm liking how we're doing it right now at my workplace (the Wikimedia Foundation). I'm experiencing a reasonable combination of informal and formal communication with my boss, plus a yearly review process. And no forced stack ranking, thank goodness.
I'd be interested to know what permutations of "360 degree" review processes people have thought worked well.
The instant I read this stack-ranking nonsense, I was reminded of a cheesy TV-series I'd recently watched where the insane bad-guy killed whichever underling bore the fault for a failure, as if this example would keep the others in line.
It was so patently absurd in the TV-show, I felt it was being played as a farce, but here's almost the same thing in real life. You go out as a part of a team, and then you see trouble on the horizon, and you know that whoever catches the blame for it will be the one to get cut. Suddenly, instead of being a "let's all get away from the bear situation" it turns into "let's see if I can outrun you."
(I may have mixed a metaphor in there somewhere.)
"stole our cheese and claimed to be a cow."
Sumana Harihareswara @82
Your bit about reminders to take the long view makes me think of old cheerleading articles from the 70s and 80s about "japanese-style" management and the complete freakouts that US managers and subordinates had when their new bosses responded to questions about what to do about a particular situation with "act in the best long-term interest of the company."
The idea that people not at the top are still intelligent and capable of seeing big pictures has not merely been forgotten but become undoctrine.
There must be some sort of Game Of Thrones "Stark ranking" joke available here.
Sumana @ 82 "I'd be interested to know what permutations of "360 degree" review processes people have thought worked well.
Can't help with worked well, but the one and only time I have come across it, was when an actually crazy boss demanded that his demoralised team give him 'upward feedback'. We refused, at first, because what were we going to say? The dude was nuts. He eventually moved heaven and earth to get a facililtator to run a session where we all had to talk about his management performance. We were all as guarded as can be, but still her report sent him into exactly the kind of uncontrollable rage we had all expected it to.
The next meeting we had with him, he behaved in such a way that three of us resigned on the spot. There were only five in his team. My exit interview was a doozy.
When I was rehired as a contractor on better money *three weeks later*, he was gone, after his boss found the desk drawer full of unpaid accounts, and his email showed that he'd done nothing except harass his staff for months. He was sacked the week after we all resigned.
I still don't know what possessed him.
To the three titles Sumana Harihareswara listed in #82 I would add Controlling Software Projects by Tom DeMarco (co-author of Peopleware). First paperback I ever spent $52 on, and well worth it, IMHO.
DeMarco also wrote Slack: Getting Past Burnout, Busywork, and the Myth of Total Efficiency, a very concise look at a number of destructive styles of management.
heresiarch @57:
Time and again, it's turned out that in a cutthroat take-no-prisoners dog-eat-dog existential battle of all against all, the most winningest strategy is working together.
Which, taken into another context, is why The Pure Market consistently declines into monopoly, collusion and cronyism.
abi @90:
You can also think about markets (or at least certain kinds of markets) in terms of power relations, thus completing the circle. In a "commodity" market aka perfect competition, the buyers are in control, because there is always a reserve army of unemployed production capacity. So the price drops to the marginal cost of production. In a seller's market, where you have lock-in or imperfect competition or asymmetric information the price rises to the buyer's marginal avoided cost. Both of these are ultimately bad because of the combination of the winner's curse and the micawber principle.
What's needed, of course, is some kind of cooperation between buyers and sellers so that the price stabilizes somewhere in the middle, with benefits to both sides.
paul @ #81:
Your comment is entirely incompatible with my existing understanding of what Who Moved My Cheese? is about.
I'm not saying your comment is wrong, mind you; I've never actually read the book myself, only second-hand summaries, so it's entirely possible that it's my understanding that's at fault.
It's just that the part of my brain responsible for cognitive consonance has been distractingly engaged in a futile attempt at reconciliation ever since I read your comment, and I'm hoping that if I acknowledge the futility out loud it'll give over and let me get on with something else.
Xopher @80 --
No, that's "Nine *men* can make a baby in a month". Studies Have Shown* that men have higher productivity that women ....
* weasel word warning.
Is it my imagination or is this "stack ranking" nonsense just like "reality television"? "Vote them off the island!"
abi @ 90: "Which, taken into another context, is why The Pure Market consistently declines into monopoly, collusion and cronyism."
Also known as the socialism of the rich.
Is this URL worth a particle? The comments, particularly, are on-topic for stack ranking.
Jason, usually the URL in "And (optional) your URL here" is meant to be YOUR blog or website. We link in the comment body to whatever we'd like to share; sometimes these links are picked up as Particles, but usually Teresa just makes Particles out of whatever she thinks is cool.
Paul A @92 - I read who moved my cheese 11 years ago, when I was somewhat younger and more naieve, albeit becoming more politically aware. Even I could see the major flaw in the argument in the book, which was to ignore the question of where the cheese came from and where it went. The correct behaviour according to the author was, if your job/ market niche/ source of money dried up, was to run around searching for a new one.
If taken at complete face value and in certain limited circumstances, you do indeed need to find a new market for your product if the old one dries up. But if someone moved the cheese, by say bribing politicians to make your product illegal, or by introducing bad methods of management which result in you losing your job, what then? Answer - you get into politics.
Basically the story can be seen as putting the onus on you to control your own personal circumstances rather than dealing with the bigger issues which affect your life; you are passive in the face of opression instead of acting against it. This is reinforced by the story being set in a maze - why not break out of the maze to avoid playing the game in the first place? Of course if you take it at a sufficiently abstract level the maze is life itself and thus you cannot escape, but you can also easily see that at a lower level of abstraction the story encourages you to remain trapped within the maze of the workplace and the greater capitalist and market based system.
I feel remarkably dim for not having mentioned the other thing that's wrong with the analogy in comment #2. Darth Paradox spells it out in comment #64: stack ranking is a zero-sum game, one in which the lowest-performing players are dropped even if their performance is actually good.
A publishing company dropping authors who fail to earn money, or who lose money, isn't the same kind of zero-sum game. If we published fifty authors and even the lowest-selling ones met our ROI targets and made money, we wouldn't just drop the lowest-selling 10% for kicks. But that is how "stack ranking" works.
Patrick @99: It's a complex, irregularly shaped piece of fractal mistakenness, which makes it difficult to identify the thing's centerpoint. My current take on that hypothetical center (and yes, I'm aware that this is like comparing Mercator vs. Orthographic vs. Dymaxion vs. Winkel Tripel vs. Waterman Butterfly) is that it's based on one factual error and two questionable value judgements.
The factual error is thinking that we only consider sales, rather than a book's own peculiar charms and virtues. How Hal imagines we decide to publish first novels is a mystery. How he got this idea at all is also a mystery, since almost every insider who's ever written about book publishing has mentioned our persistent habit of publishing good books we know are almost certain to lose money.
(Three smaller factual errors I'll note in passing. First, a book's sales are influenced but not determined by the sales of the author's previous book. We're always judging our books' intrinsic qualities. Second, making a big deal out of numerically ranking books and authors by sales isn't so much a publishing obsession as a bookselling obsession. We just make use of it when we're writing copy. Third, that practice, no matter who does it, has zero resemblance to stack ranking. Granted, they both involve sorting, but so do pachinko machines, speed dating services, and the NBAA championship playoffs.)
Onward to the two questionable value judgements.
The first, I'll admit, is more inferred than stated: that there's something wrong or hurtful about keeping track of which authors' books sell how many copies. Like you, I'm a practical, down-to-earth liberal, and not really sure how these advanced libertarian positions work; but I'll take it on faith that Hal understands it.
The other value judgement (only a little bit inferred) is that it would be a good idea if we stopped paying attention to such things. Happily, I know exactly what the answer is to this one: We can't. If we don't make money, they'll eventually stop letting us buy more books.
Being unable to buy more books also has nothing to do with stack ranking, except insofar as they both suck.
Guthrie @98, that's very close to my reading of the "who moved the cheese" story. People who "move the cheese" by un-leveling the playing field or arbitrarily firing the mice bear no responsibility for doing it, and it's foolish and irrelevant to even consider that question. Being deprived of cheese is just something that happens to mice, so it's their responsibility to stop grousing, get out there, and find themselves a new cheese: happy ending!
The only real difference is that I've always assumed that any book that sells that many copies must be dispensing reassurance rather than bracing pep talks, and that there must be a reason no one in the story owns the cheese removal process.
My assumption is that the book's core audience is executives, and that its purpose is to reassure them that it's okay to do rotten things to their employees: "Sure, they'll seem distressed at first, but that's just a mouse thing -- they aren't as brave and realistic as you are about the inevitability of cheeses getting moved. Soon they'll pull themselves together, go out, find a new cheese, and be just as happy as they were before."
What bothered me about Who Moved My Cheese? is that it seemed to me that the mouse would starve to death.
From one angle I'm taking an allegory too literally, but from another I think it's more evidence that the author wasn't thinking about what they were saying.
Abi @5:
One significant problem with this approach is that it rarely values catalysts and facilitators well enough.True. I once saw someone sort out a problem of several years' standing that had been hobbling the workflow of three entire departments, and damaging the company's ability to market its products. It took them about ninety minutes from start to finish, including travel time. "Help things run smoothly, and keep bad things from happening" is an implicit part of everyone's job description, but some people are better at it than others.
All my smartest supervisors have been big on the idea that getting along with your co-workers is as much a part of your job as anything else you do.
Sarah @34:
I'm now picturing John Cleese ordering two pantomime horses to fight each other to the death for the single available job in a merchant bank.Given how many of his best Monty Python skits were about bad management practices, it's wonderfully logical that in 1972, Cleese started a company that makes funny management training videos. They have a good reputation. Their most famous video is Meetings, Bloody Meetings.
Lance Weber @40, that's a distressingly plausible scenario.
Kevin Riggle @48:
The worst part of the system is that it actively hides problems, I think. (Take as given for a moment that the 20-70-10 system is wrong.) In particular, in a system which doesn't mandate that distribution, it's easy to see that, if group has all underperforming reports, maybe the problem is the group -- or maybe the problem is the manager.Unless there are significant extenuating circumstances, it's either the manager or the manager's boss. Who else is responsible for seeing that the people match up with the work, and that their resources and circumstances make it possible for them to do it?
I've always promised myself that if I'm ever in upper management (unlikely), I'll keep an eye out for departments that have unnaturally high turnover rates. I've never seen one of those where the boss wasn't the problem.
(The boss Emma in Sydney described in comment #87 is exactly the kind that produces that effect.)
Nancy Lebovitz @54: Hard Facts, Dangerous Half-Truths And Total Nonsense: Profiting From Evidence-Based Management looks like an interesting book. I went and had a look at it on Amazon, and the top-ranked review had this to say:
This is an outstanding book written by two business and engineering Stanford professors. Analyzing modern management practices using surveys and studies, they debunk many of modern management practices. According to their studies, pay-for-performance does not work. Companies that had the widest range of pay scale between top and bottom performers also suffered the poorest financial results. So, pay-for-performance does not translate into superior stock performance. Similarly, forced ranking where employees performances are clustered in three different buckets (top 20%, middle 70%, and bottom 10%) where the weakest bucket is expectedly weeded out does not work either. Companies using this system have been plagued with an employee force with low morale, high turnover, and low productivity.I honestly believe upper executives like stack ranking because it matches the rationale ("success is a matter of getting the right superstar executives, and you have to pay top dollar to retain them") that's been used to grossly inflate their compensation. It's nice to know that someone's actually run the numbers.The authors debunk tens of other well established managerial practices. These practices are often so well established that no one seemed to question them until these two academic types came along. By doing so, they have done a great service to the business community by opening our eyes using the scientific method.
So, why have such practices that seemed to be part of corporate capitalism not work so well? According to the authors' analysis, it is because they all foster a winner-take-all mentality. They reinforce an individual star system. That works well in individual sports like alpine skiing, where it is one individual against the clock. The corporate business world is more like a team sport. Soccer comes to mind. One star within an otherwise demoralized team does not stand a chance against a motivated high-performance team. In the corporate world it gets even more complicated than that, because the team concept extends way beyond the walls of the corporation. Effective teamwork entails including suppliers and customers in product design and management decision. In such an environment, internally pushing a star system that is demoralizing to the 99% who come in distant seconds does not foster optimal team performance either internally or externally.
Heresiarch @57, I've quoted most of that on the front page.
Lee @63:
What that sounds like to me is the same philosophy that's enshrined in [No Child Left Behind]. Underperforming schools are closed down; the ones that make the mark MUST improve their testing scores every single year, or they too are labeled "underperforming".A lot of education writers are saying the same thing -- and pointing out how many of these initiatives originate with Bill and Melinda Gates.
I'll add that the other sinister influence is Kaplan Testing, which is owned by the same people who own the Washington Post. The newspaper business is collapsing, but Kaplan is lucrative, and the WaPo pushes policies like NCLB that deliver vast amounts of education money into Kaplan's hands.
Darth Paradox @64, I saw a striking if oversimplified explanation yesterday for stack ranking.
Say the company is looking to reduce its staff. At review time, HR asks managers to accurately rank their employees' virtues and habits. The more truthful and/or naive managers are the only ones who actually do this. More savvy managers claim that all the employees who report to them are outstandingly perfect.
Result: the naive managers are the ones who lose staff positions, and their remaining employees are pushed to the bottom of the list for raises and bonuses.
HR observes this happening. The next year, they institute stack ranking.
Clew @70, I don't know of any companies that work that way. Meanwhile, true pensions are getting rarer and rarer. They've been replaced by 401(k) plans, in which the competition model is applied to our retirement funds, as demonstrated in the most recent couple of crashes.
Bruce Cohen @76, if we ever get the opportunity, I'll keep buying the drinks as long as you keep explaining Agile Methodology and similar heresies.
Lightning @94:
Is it my imagination or is this "stack ranking" nonsense just like "reality television"? "Vote them off the island!"IMO, it's just like formally decimating a Roman Legion, except they pretend that all the blame belongs to the unfortunate 10%.
Jason Aronowitz @96: I don't know yet, but I'm interested in finding out.
Nancy Lebovitz @54 and TNH @103
Hard Facts, Dangerous Half-Truths And Total Nonsense: Profiting From Evidence-Based Management does look interesting.
I noticed, when I looked at it on Amazon, that one of the authors, Robert I. Sutton, also has a book entitled The No Asshole Rule: Building a Civilized Workplace and Surviving One That Isn't, which is also appealing.
Patrick Nielsen Hayden @ 99: "If we published fifty authors and even the lowest-selling ones met our ROI targets and made money, we wouldn't just drop the lowest-selling 10% for kicks."
I suspect, however, that you'd find your ROI targets gradually creeping upwards--as would your competitors, even if their books were still selling just as well as before.
I pick this nit because I sense an undercurrent in this conversation's critique of stack ranking, something along the lines of "all this circular firing squad nonsense can be avoided by using a set of nice objective criteria like sales metrics or individualized benchmarks." But even assuming a nicely objective and accurate metric presents itself, it still doesn't solve the central problem of determining who is good at their job.
Say Alice has a job making widgets. She makes 10,000 widgets in a day. Her productivity is easy to measure, uninfluenced by subjective opinion and she's even working solo, avoiding the complexity of group interactions. So is she a good worker? That question is impossible to answer without knowing how many widgets Bao made in a day, and Charlie and Dagmar and every one else in the widget trade. Efficiency is relative. Stack ranking is stupid (among other reasons) because it tries to apply that macro-scale effect to micro-samples--key in the above is comparing to everyone else doing work in widgets. But all criteria are ultimately comparative.
Teamwork as a criterion...
does this mean I have to prove I am better at working with you than you are at working with me?
The bind moggles...
Clew @78: Wicked problems! Great concept!
When I'm teaching writing, I have a semi-standard rap on the commonest ways novels go wrong, which of these problems are fixable, and what it takes to fix them. Then I add the other class of problem: the misconceived book. There are no useful rules you can derive for fixing those, because every one of them is misconceived in a different way, and very few of them are fixable.
I now understand this better: misconceived novels are a wicked problem. Jinxed books may be one as well.
Heresiarch @105: Objective criteria are a charming idea, but most of the benefit comes from the attempt to be objective. Problems are caused by thinking you've succeeded.
I snagged one of my favorite authors ever because his previous publisher decided to cut loose every author whose sales fell below a certain mark. My own calculation was different: Well-beloved author, consistent quality, selling steadily for decades, has in hand one finished and one nearly-finished novel in his best-known series, besides which he's the worldcon GoH next year ... And you know, I can work with that.
Made a profit on it too, bwah-ha-ha-hah.
Someday let me tell you how objective criteria ruined the CIA.
Teresa (#108): I'm glad you did, especially if that series comes out in nice DRM-free ebooks when the Tor ebook store goes live; it's one of my favorites, too....
(It wasn't difficult to figure out who the author was, given that many clues.)
I'd love to hear you tell how objective criteria ruined the CIA, Jim. Can I buy you a drink?
Living here in Seattle I've known a number of MS employees of varying duration and rank within the company. When I first heard of stack ranking my first thought was that it would encourage sabotage as equally as productivity. Screwing a team member was as good as doing a better job yourself. And much easier. Why did this obviously flawed system persist? When the company's attitude is "we're smarter than everyone" there's no reason for self-criticism. In fact, self criticism can be seen as traitorous in that it challenges the ideology of being smarter than everyone else.
112
That brings to mind a few other companies who were sure they were the Smartest People Around. And they all went broke (or got bailed out of being broke).
PNH@99: we wouldn't just drop the lowest-selling 10% for kicks. But that is how "stack ranking" works.
Interestingly, that's not the case. At least at Microsoft, the fixed quota applies to the 3.0 'working as expected, no bonus' ranking, not the '2.5, see me after school' one.
Another thing people seem to be missing the point on is the fact they use numbers for the output result doesn't really signify much different than if they had used letters, words, pictures or whatever. True, it does have a level of attempted consistency and impersonality you wouldn't get in a smaller company, but that is kind of inevitable for a massive corporation like MS. And anything inevitable can't be a novel problem.
The thing is, despite not being, in reality, as bad as the cartoon version that gets painted, it does appear to be every inch as destructive of value.
That's a commonplace for many of the very worst decisions made by large organisations with otherwise functioning internal decision making processes. They are both mildly immoral, and superficially similar to another option that is even more so. Those two characteristics are generally enough to prevent any useful discussion as to whether or not they actually work. Everyone is too busy running morality dramas in their head, including _both_ those who see themselves as resisting temptation, and succumbing to it.
Evidence becomes simply a weapon used by the opposing bad guys (naive idiots or greedy individualists, respectively). Taking it seriously becomes a moral failing.
Something being morally slightly bad, though better than an alternative, provides so many opportunities for interesting and dramatic narratives that the prosaic reality that it is a deeply stupid and inneffective idea simply can't compete for mindspace.
See also: why waterboarding more or less literally _always works_ in fiction, when employed by the protagonist or their sidekick. There is no drama in abstaining from something you have no need to do, or from breaking a rule you have no need to break.
Tom #111
Sure. I'm not going to tell that story in public.
Teresa @108
Could that also be indicative of the proper time-scale being longer than some managers might think, and of ROI depending to some degree on risk. You took a longer-term view, and saw a reliable source of income, both in production of new novels and sales. A solid 10% a year helps make sure the business can pay its bills. 20% last year and 0% this year might be the same average, but you have to plan for that.
A good many stock scams depend on people not realising that a high ROI comes from high risk.
I remember reading an economist saying that a lot of the changes and "innovations" in the financial industry are driven solely by finding new ways to fool investors and often the financial company itself into thinking they'd found the impossible financial Holy Grail: a permanent way of increasing returns without risk. A way to break the previous set of risk models to incorporate undetected risks.
Teresa #101 - I would extend your target market for the book. People who need reassurance that they are doing the right thing, i.e. running around searching for new cheese, rather than stopping and wondering why there is a limited supply of cheese in the first place. Thus it dovetails perfectly with the myth (I believe it is most pernicious in the USA, but also exists elsewhere) that if you work hard enough you can get where you want to be/ lots of money and success.
This is patently untrue, but has a strong cultural influence.
guthrie @118 and Teresa @101 and others.
In fairness, I will say that while I was working for a career counseling organization, the counselors found the book useful for people who needed to be encouraged to think about actively managing their careers instead of waiting for someone to tell them what to do next. That need is becoming less prevalent, I suppose, as younger workers know this from the get-go. But when that book first came out, there were a fair number of people still holding to an outdated notion that you went to work for a big company, you took the work assignments they gave you and the training they thought you needed, and you didn't rock the boat, and in return they would take care of you. It was never true for everyone, but it used to be true for more than it is now.
I worked for Microsoft for seven years.
Anonymi, @44, nails it.
But it was worse than that.
I'm pretty sure I first heard of the concept of stack ranking in Flatliners. There's a scene where our intrepid heroes are running around doing panicky plot things during a final exam, while in the background their instructor is snarling about their grades: one of you will get an A, one of you will get a B, one of you will fail, etc. The point of the scene was to demonstrate how vicious and cutthroat medical school was; it didn't matter how good you were objectively, you also had to be better than everyone else, and you were all competing against each other.
That anyone thinks this would be a great strategy for encouraging a successful company is beyond belief.
Nancy Lebovitz @ 102... What bothered me about Who Moved My Cheese? is that it seemed to me that the mouse would starve to death.
Not if the mouse has a tendency to sing "Here I come to save the day!"
Wolf Baginski expected to be scared. The job he had, you would be crazy if you were not. And Lt. Maureen Brown, the hyperactive mouse who ran the Naval Syndicate's intelligence branch in the Spontoon Islands, had put him in some pretty bad spots. That was her job. She was Naval Syndicate, he was Army Union, they fought... Well, no, they didn't fight, they were on the same side and they were professionals. They didn't waste lives and talents. He trusted her not to waste him.
She took a deep breath (which would have distracted some men), and said, "An Italian motor ship, built to carry refrigerated goods, docked on Tuesday evening. Wednesday morning, she was beached on the little delta on Main Island that was deposited by the outflow from Crater Lake."
"Crew?"
"All missing. Nobody went ashore after they docked, no trace on the ship. Still places to check on board, but nothing suggesting violence."
Wolf nodded. "Main Island, strangers would get noticed. but they could hide for a while in the jungle."
"The militia is on the job. You know how good they are."
"Pretty good," admitted Wolf. "And that's their home territory. I wouldn't want to be doing escape and evasion for real if they were chasing me." He shrugged. "Italian, that's a little odd. We're nowhere near Abyssinia."
"The holds are empty too, hatch covers pushed open from the inside and tarps torn free, and they really were chilled down. She was loaded when she docked, down to her plimsoll line."
Wolf nodded slowly. He was beginning to think the Dynamo was scared, and he knew enough about the Spontoon Islands to think she had some good reasons. "Pretty obvious what I have to do, I guess."
"First step, anyway. You're a qualified combat diver. Go and help the Spontoonies check that ship." She handed Wolf a folded sheet of paper. "Those are your written orders, and if you need 'em we're all in big trouble. Richelieu Rules. Verbally, find out who, or what, moved the Mio Formaggio, find the cargo, and make sure it isn't a threat."
Wolf nodded. "And if it is a threat?"
"Try not to frighten the tourists too much."
Mythago #121 - and an environment surely unlikely to produce doctors with a good bedside manner and an ability to admit when they have made mistakes and correct themselves in future.
But even assuming a nicely objective and accurate metric presents itself, it still doesn't solve the central problem of determining who is good at their job.
But surely one of the problems with stack ranking is that it is *by design* incapable of accepting the answer "everyone in this room", even if it is true and well-supported by evidence.
P.S. This kind of thing is why I'm suspicious of so many school "reforms": they seem to consist largely of stack-ranking teachers, and then claiming that will benefit students and schools and solve all problems.
I would not be surprised to hear he stole it from someone else, but as far as I know, "stack ranking" was invented and promulgated by Jack Welch of GE. (He is credited with this in the subject article about MS.) He called it the LEP (Least Effective Person) program. We called it the Loser Program or Leper Program. Out of every ten people, two were A's and got good raises; seven were B's and got cost-of-living raises; and one was a C (the Loser) and got fired. Only if you did that, a typical of office of eleven people would have ten people the next year and nine the year after that, to do the same amount of work, so what most managers did was rotate the "Loser" from year to year, pretending that last year's Loser had turned over a new leaf. Welch caught onto that after a while though. Welch was all about layoffs. I called the LEP Program his Permanent Floating Layoff.
The other thing managers could do is trade jobs and fire the last person the previous manager had hired - since they were the least experienced people (but it doesn't look good to fire the person you yourself have just hired).
A couple of vignettes:
The most experienced, best engineer we had, coming out of his annual review meeting to find several people lined up waiting at his desk for mentoring: "Don't ask me any more questions - I'M JUST A B PLAYER!"
The winner of "Best Co-op of the year 2000", hired the next year, fired the year after by a new bungee-boss (*).
Every time I see Welch's book "Straight From The Gut" I think, yeah, right, but always from the wrong end.
* Dilbert reference - the early Dilbert cartons seemed obviously based on GE/Welch source material, e.g. Six Sigma = "qualicide".
Jim Macdonald @ 109:
I'd argue that the CIA was ruined from the start because it was started by cowboys and continued by middle managers. But more bad management theory and pernicious incentives could always make it worse.
Bruce Cohen STM @128: Not all the people involved at the start of the CIA were cowboys, and not all the continuity came from middle managers. Those may have had too much power, but it's more complex than that. (See Sherman Kent's Strategic Intelligence for American World Policy for a contemporary look at the philosophy of those who were involved in starting it -- still a good book about what intelligence is for.)
Wasn't Alice Sheldon, aka James Tiptree, one of the CIA's original people?
JimV: my impression was that Adams got a lot of his early material from his job as a business analyst at Pacific Bell. In the late 80s or early 90s I visited PacBell headquarters in San Ramon where he worked, they had many thousands of people in cubicles in one huge building.
He's described his work background and its contributions to the strip.
For some years I managed a highly performing software team. When the order came to decimate my unit I had no poor performers, but was forced to name two. I was unhappy, having built that team ever so carefully. After the layoff season I found both my sacrifices in lead roles on other teams who had not been as good at recruiting.
At least the next year I was able to negotiate transfers explicitly instead of going through the dec,action silliness. I can't say that I've found much to like about most software team management.
Serge@130,
I don't think Alice Sheldon was original, but Paul Linebarger, aka Cordwainer Smith, aka Felix Forrest, was part of the OSS.
JimV, if you don't mind my asking, why would Six Sigma be considered qualicide?
I'm asking because I have the option of getting a green then maybe a black belt, and I would like to know the downsides.
Jeremy Leader@131 - I worked for Bell Labs from the late 70s to early 90s, and various other parts of AT&T since then. When Dilbert first came out, it was mostly centered on the kinds of business idiocy we had at The Phone Company, and gradually became more generalized, because we certainly didn't have a monopoly on bureaucratic idiocy. I did various projects at Bank of America during the mid-90s, and their cubicle walls were starting to get some Dilbert cartoons mixed in with the Gary Larson, and Dilbert was starting to look fairly comfortable there. (This was several buyouts ago, when BofA was its original California incarnation, and was big, bureaucratic, and amazingly slow, as opposed to Wells Fargo, which after several mergers was big, fast, and flaky.)
Bill Stewart: there was a venture capitalist some years ago who was interviewed and the topic came down to what he looked for in a tech company to invest in. He said that a scattering of Dilbert cartoons throughout the place was fine, but if he found large concentrations of them he'd put on the brakes and start digging. This stuck in my mind because there was a wall of them at US West Bellevue that was 8' x 12' and growing when I was there.
The old 11 plus exam worked on a stack ranking system. Or as it is called in education norm standardised (that is where the top 10% get an A the next 10% a B &c.). The LEA would have a certain number of places at Grammar Schools say 10% and the pass mark would be adjusted so 10% passed and 90% failed. If you failed you went to a Secondary Modern.
Usually at a Secondary Modern you could only take CSEs and had therefore had essentially no chance of pursuing education beyond 16 as a top grade pass at CSE was only equivalent to a C at GCE O level. The even more evil part of the system was that there was a large difference in the number of places available to different pupils. The disparity was based on class, region and sex. The North had far fewer places than the South there were more places for boys than girls. In 1963 Wales had 33% Grammar school places the East 22%.
Later on Secondary Moderns were more likely to offer O levels to their better pupils which fixed some of the problems. The system was then gradually abolished, and the reference standardised GCSE brought in to replace the norm standardised CSE & GCE O level. There had been intended to be some transfer of pupils between Secondary Moderns and Grammar Schools, however in practice this almost never happened, you were branded a failure at 11. There had been meant to be a third category the Secondary Technical, but they never really got started.
Shorter Soru #114: "For obscure technical reasons I won't bother to explain, you're totally wrong. Here's how it really is: Exactly what you said, except I'm saying it, so I get the credit for the insight instead."
OtterB @104: The No-Asshole Rule
That book sounds interesting. I do have to wonder if one of the failure modes of applying that rule is a company with a few passive-aggressive assholes in key positions who misinterpret assertiveness as assholish behavior, or simply label as an "asshole" anyone they find threatening.
I've bought a copy, so I may shortly discover that the authors anticipate me, but my experience of middle management is that it is not merely sufficient to not be actively sadistic, which is a big caveat that the sound-bite version of the rule fails to capture.
Kevin Riggle @139: "It's more of a guideline, really".
I think any rule, mindlessly applied (see also "zero tolerance") can lead to bad results, especially in the presence of assholes attempting to game the rules.
I think the "no assholes" rule is intended to be a minimum standard, to help resist the temptation of hiring someone who is really excellent, but a bit of an asshole. I don't think "not an asshole" is intended as sufficient criteria for anyone, especially not a manager!
The one thing I saw missing in The No Asshole Rule was that there was no mention of how a boss can tell if there's someone who's friendly (perhaps even very flattering) to the boss while being an asshole to subordinates.
Nancy C. Mittens @ 134
I worked for a GE-owned company that had attempted to use Six Sigma methodology.
My observation was that Six Sigma is a very good tool for its purpose, which is basically process control of small, well-defined, frequently-repeated processes (very common in manufacturing). When applied to a complex, poorly-defined, infrequent process (compiling annual regulatory statements), it bears the same resemblance to quality improvement that adding up random numbers from road signs[1] does to mathematical analysis.
1) A favored recreaction of small children in my house--look, that sign has a $1; and that one has a 55; that makes 56! The fact that one is a price and one a speed limit means that the 56 has no informational content.
Also not having read it -- does he address the fact that we're all assholes occasionally? So, maybe, it should be "No chronic assholes"?
And I'm only using the term because it's the title, since I would prefer not to demean that sphincter.
Nancy Lebovitz @141: How does one tell if there's someone who's friendly (perhaps even very flattering) to the boss while being an asshole to subordinates?
The organizations I've been a part of leading, it's usually been a peer relationship, and I found out someone was being an asshole because I got told about it, but that's hard to count on, and I don't know how much went unreported. Then again the organizations were all-volunteer, so people who theoretically reported to an asshole often just failed to show up, which is a much more desperate option when a paycheck is involved.
Kevin, I have no idea how one would tell-- that's why I was hoping the book would have an answer.
Nancy Lebovitz @145: My apologies, that was more me musing out loud than a direct question of you.
Thanks Sam! I don't know if it will be any good for my current job, but there may be some use in it.
The whole article about what went wrong at Microsoft is available.
The fascinating thing is that they've been doing destructive firing for years while apparently remaining wildly overstaffed.
Tom Whitmore @143 I just checked the book out of my library. It does address the question that we're all assholes occasionally. He differentiates between being a temporary asshole -- we all slip up or misunderstand or just have a bad day now and then -- and being a chronic one, where there's a pattern, essentially, of making people feel worse for having interacted with you.
It's not a deep book, but it was interesting. Much of it seems obvious to us, I think, because of the discussions about what makes for healthy and thriving conversations on this site.
Thanks for the report, OtterB. Good to know.
Stack ranking doesn't always need to be so draconian. Just because a leader rates his/her staff against a set of metrics or behaviors and then sorts the team based on these results doesn't mean they need to cut from the bottom... Sometimes it simply provides additional perspective and insight into where your team needs work - both from the top as well as the bottom... here's a free Excel template I created to help leaders do this very thing... http://www.mlynn.org/2012/06/zen-and-the-art-of-stack-ranking-with-excel/#.UNXrKZPjn0o
#151 is a first time post with a link to a self-promoting website; however, is not bot-generated spam, by any means.
Re #151: Technically, it's spam, but it's also on-topic and not automated. Mods, your call.
That post was flagged by the filters, but released because it was hand-made and on-topic.
We gnomes try to err on the side of mercy.
I know I'm mind-bogglingly late to this conversation, but I must thank the participants (except the spammer).
To me the problem with WMMC? is the basic underlying premise: workers are identified with parasitic creatures devouring something created by Higher Beings. Anybody who believes that is either a Koch brother or a devout and unusually stupid Randroid.
Comments containing more than seven URLs will be held for approval. If you want to comment on a thread that's been closed, please post to the most recent "Open Thread" discussion.
You can subscribe (via RSS) to this particular comment thread. (If this option is baffling, here's a quick introduction.)
HTML Tags:
<strong>Strong</strong> = Strong
<em>Emphasized</em> = Emphasized
<a href="http://www.url.com">Linked text</a> = Linked text
Spelling reference:
Tolkien. Minuscule. Gandhi. Millennium. Delany. Embarrassment. Publishers Weekly. Occurrence. Asimov. Weird. Connoisseur. Accommodate. Hierarchy. Deity. Etiquette. Pharaoh. Teresa. Its. Macdonald. Nielsen Hayden. It's. Fluorosphere. Barack. More here.
