Microsoft does away with stack ranking
zdnet.com
zdnet.com
I've had employees on probation for lack of productivity judged as excellent by their peers because of their ability to socialize their "abilities". I had the best developer on my team (a 10x developer if ever there was one) get slammed on technical ability rankings because of his prickly personality.
My VP and I used have quite the chuckle at all my "strengths" identified by my team and peers ... because they liked me and I was effective they attributed positive characteristics to me that I did not have.
You're basically saying that those who worked daily in close interaction with the 10x developer and had to work with the code that developer produced, had lower regard for his technical ability than a manager who presumably is further removed?
You can't ignore the social aspect. A team with social frictions isn't going to work as well as a team that enjoys working together. Obviously just being social isn't enough but a strong team needs a combination of the social and technical aspects. Also 360 should mean going beyond the team, so the manager should be evaluated by his peers, his managers and his team and technical people on the team should be evaluated by members of other teams they work with.
In the situations I indicated I was also a developer on the team so I knew how productive and skilled the 10x prickly developer was. His peers ranked him high on technical skills but only as high as more amicable and lessor skilled peers.
I'm not saying I ignored the social aspect, in fact that is why I called him "prickly". I'm just saying that, while his teamwork related areas were properly low, his technical areas were improperly low because of this halo effect.
I swiped the Roman Evaluation Method from Luke Hohmann. During interviews, a No from one is a No from all. Self selecting teams is one of those agile advices.
If your prickly superstar was chosen collectively, he would have been given a lot of latitude. Eyes wide open kinda thing.
The only sticky interpersonal problems I've had is when some higher ranking genius assigns people to my teams, mostly unwanted and unwelcome.
And sometimes just plan bad soup. People good individually, bad together. Like when the boss assigned an outspoken Lebonese and an outspoken Israeli to my team. They got along as well as peanut butter and lug nuts.
Next thing that happens is people start choosing members based on how much comfortable they are (which doesn't translate into technical skills and delivering stuff) or start talking to each other, trying to convince people.. this sucks!
While it might work for small companies with not much variation in technical skills and sociability between employees.. for anything else it's kinda hard.
Plus, that initial list of 2-3 employees that start talking and picking others was selected by someone, wasn't it? And the list that they can pick from was also selected, no?
I think if you need a team of 10 and you have 30 possibly good candidates within your company (and they are all available to move to your team).. sure, why not. You as a manager already pre-selects a bunch of people and then let your team members filter it even further.
But putting that responsibility from the beginning solely on the team members.. I guess that's risky. But anyway, how often does that happen?
If we are talking about selecting 1-2 people for a already formed team than fine.. of course other team members should do it (but that's mostly the norm in a lot of companies.. the manager/tech lead runs the resumes through people he/she trusts, ask them to participate in interviews.. no big difference there).
How? What makes you so sure you weren't biased?
How? He wrote a ton of great code in a short period of time that worked amazingly well. How else would you tell somebody was productive and skilled?
But, think whatever you want!
At the finale, 5 people were doing nothing but fixing bugs in the "stars" previous projects. That's when senior management finally clued in to the problem and fired both the manager and the "super star".
The guy was attributing statements to me, from the conversation we were having in that moment, that i had never said. He would even say "what you said, or what you were going to say..."
I found it extremely bizarre. I am a pretty precise person, and I wanted to correct him but i realized it is pointless. the lesson i take away was powerful - if people "like" you there is almost no level they won't go to make it work out.
one thing we can do about this is cry about it. the other thing we can do is to figure out how to "hack" likability. i think my mother taught me most of the basics already.
of course - much easier to do this for one hour in an interview, than for an entire year at a job. but there is something to be said for smiling, a few kind words, and then ... graceful evasion of everything else when you want to do real work.
Actually, the key to life is sincerity. Once you can fake that, you've got it made.
You seem to know a lot about psychology. What do you call someone who doesn't recognize jokes?
People's ideals generally move into conformity with their actions, not the other way round. That's one of the myriad reasons being a spy sucks. If anyone reading was following the police infiltration of left wing groups scandal in the UK, with added sex and dead babies identities the unit involved had a policy of only using married men after one guy never came back. Probably gone native.
edit: and I don't ask this to be insulting, but judging performance is a very difficult task and I've never seen a reliable methodology for doing it. Thus, I'm naturally extremely skeptical when someone claims that they can do it better than others. Especially so when it includes broad generalizations about a group of people.
People who think judging performance is hard have a broken development methodology or no technical chops.
That's also why I included the comment on my own rankings; I have some pretty serious weaknesses that were ranked as areas of strengths by my peers and reports.
And this is exactly why I asked the question. Here you have judged me, since I said judging performance was hard, on both my development methodology and technical chops. Yet you know nothing about me at all.
The problem of evaluation is an important one, and one I'm always interested in reading more on. Unfortunately the strategy of taking one's self as a gold standard isn't exactly what I was looking for. It does seem obvious that being an expert in your domain is a necessary trait to be able to evaluate others, but it's not at all obvious that it is a sufficient trait.
The idea of multi-source feedback seems intuitively good, despite the conflicting research on it. That more input will expose biases or inaccuracies is a decent hypothesis. On the other hand, judgment from a single person seems highly error prone. Whatever faults that one person has will never be exposed and reviews seem almost guaranteed to have bias.
In the end, I'm left with no idea why you so quickly discard the reports of your peers.
I've never had a problem judging performance when I was doing my job right. Any time I've been unsure it's because I was either not managing the team properly or did not have enough technical knowledge to evaluate the estimates and claims (e.g. this took longer because of X and Y).
I know you can manage performance in a way that allows evaluation even if you don't have the technical chops to evaluate it technically, because I have seen technically weak peers do a good job of it.
I didn't discard the reports of my peers. I noticed a trend of the peers who liked me rating me higher in areas of weakness (as perceived by both myself and my boss) than my skills deserved.
Data that contains some bias is not useless if you can adjust for the bias.
I'm especially skeptical of managers who evaluate themselves as having above average technical ability after they have asserted how poorly most people evaluate their peers. The necessary conclusion of such an assertion is that people therefore haven't a rational basis on which to judge their own skills relative to others.
It's not what you don't know that hurts you. It's what you know that just ain't so.
I suggest you reflect on this. It seems to me that you've convinced yourself that you are good at certain things but there isn't really any evidence supporting your beliefs.
When I was younger, I actually used to think like you - that perf evaluation is easy and a good engineer (e.g. me) would be able to evaluate others easily. Over time, I've realized this is an incredibly complex and nuanced issue because everyone's strengths and weaknesses are so very multifaceted. I am more or less convinced that people who think they have all the right answers simply have a lot of blindspots.
Performance analysis is incredibly complex and nuanced. Judging if somebody is performing poorly because they are unskilled or not managed properly or having marital problems is hard. Improving performance is even more difficult because it requires the analysis and a remediation plan.
And I don't know your scale of 'good' programmer either: What do you mean by that? Are you saying you are as good as Linus Torvalds or are you saying you are better than your current peers? I am assuming it is the latter - which probably means zilch in a site like this. Many folks here receive that recognition and over time they come to a conclusion that it doesn't mean much. Which is why you get a lot of naysayers. They aren't doubting you, they are just doubting the general feeling as been there, done that.
Look at this thread, there are several who are very explicitly doubting me in no uncertain terms. Even doubting the idea that I'm good compared to my current peers.
If someone's social behavior jeopardizes the success of projects they work on, then perhaps it doesn't matter how technically competent they are.
It's very possible to have a bad stack ranking but excellent peer feedback.
With this system, someone who delivers very well, but undermines other people or projects they work on would likely receive a high stack ranking, but poor peer feedback resulting in private discussions. People in charge of managing said employee may ignore the problem, attempt to place the employee under people who will help them address social problems, or quietly find a way to move them to a different org. It's quite likely none of this shows up in stack ranking.
As a counterexample to yours, what if someone is very friendly, witty, and amicable, but is severely lacking in technical skills? Perhas they are not even remotely pulling their weight in projects, resulting in friendly coworkers essentially doing their work for them and covering for them? In such a case you'll want to know that their technical skills are below what they should be.
The reason you do 360's is that you have to take every individual review with a grain of salt. When you have 7-10 reviews the biases start to average out.
One thing that is true, is often times the manager has more information about the various reasons for certain choices and goings on in the business, and thus is in a better position to evaluate people.
The question then is just, how do you make sure that manager do not discriminate based on soft indicators (personalities match / clashes / favouritism etc.).
But I don't think stack ranking is our problem. I think our problem is that we value this notion of the brilliant and excellent individual contributor instead of valuing employees that value teamwork and team problem solving. I suspect this is an institutional problem, having been started by exactly that sort of person.
It's this that causes the backstabbing and the mess that people attribute to "stack ranking". Until we put value in changing this culture, I don't see too much changing.
(Disclaimer: my team at Microsoft is sort of odd - we're not in Redmond, and I don't see a lot of this backstabbing that is oft-reported. I do see our managers overvaluing the brilliant IC and I do see my organization undervaluing teamwork, but at least we're not doing that stupid game playing like joining weak teams to get great reviews.)
Overlaid on those high-stakes struggles for control and prominence are some correspondingly huge egos, whence come dogmas like "Everything Windows."
The Enterprise part of the company suffers less from this because the growth there is more organic.
I see this move overall as a positive one for Microsoft - without removing the perverse rewards to screw over your teammates and other teams, the actual cultural fix can't take place.
I think the culture will shift soon - Microsoft, for more than a decade, had no real competitors. They made more money each year whether they executed (see Win2k) or not (see Vista). Consequently, their biggest threat was actually internal. Stacked ranking enforced this, and prevented internal response.
Now that situation is changing (or has changed). Microsoft is looking at actual competition in the form of Apple and Google as mobile and search erode the desktop. On the datacenter side, Amazon and other cloud provider are threatening Microsoft's server profits.
When the heat gets strong enough, there will be a unifying force within the company (or it'll just crumble - personally don't think that's the case).
Currently, I go into a bucket - let's say I'm a two, for instance - and all the twos get the same compensation. Can I backstab somebody so that I get a two and they get a three? Maybe.
But if we remove the buckets and we have some continuity, where now if my coworker outperformed me and we were both twos... now this coworker just straight up outperformed me. I'm more inclined to want to stab that person in the back so that I'm better, since being almost as good isn't valuable anymore.
Now, it's possible that we'll end up with stack ranking all over again. The previous compensation before the current stack ranking system put you in a bucket from 5-1 where 5 was good and 1 was bad. So... it turns out that we don't change this very much even when we do change things.
Far more importantly, I have spent my 25 year career in non-stack ranked companies, and I have never seen the behavior you describe. Yet just about everyone at stack ranked companies report back stabbing behavior. I think empiricism wins over speculation.
But this is still stack ranking. What you are describing is still an ordinal system where employees had to be ranked in order from 'best' to 'worst'.
As I understand it, Microsoft are proposing to remove any sort of relative ranking between team members and instead reward for actual productivity.
So yes, employees will indeed be sorted from 'best' to 'worst' by some metric. This is pretty much true of every performance-based evaluation system, no?
Removing the "stack ranking" system refers to removing the 20%/70%/10% buckets from 1-5 that we currently sort into, not the fact that we're doing a merit-based reward.
Thus my point remains: if our review system is a problem (and certainly a great many people seem to think that it is) then this seems like a tweak on the system, not an overhaul. What we're being evaluated on and who is doing the evaluating is much more important than whether we're explicitly "stack ranked" into buckets or not.
People are constantly doing stuff to make them "visible" instead of doing what is best for the project. How many features do people fight nail and tooth for, that should be cut, just because if they aren't in the product they won't have anything to show for their work when review time comes around?
People are terrified of making decisions, because if they make a bad one, it's going to look poorly when they're ranked. I don't know if you've noticed this, but our managers tend to give "guidance" rather than outright telling us what to do. That way when something goes wrong, it isn't their fault. They didn't say we had to do it the wrong way and it's our fault for not bringing up that it was wrong sooner.
The first year I was here, I didn't think anything of the stack ranking. It was just one of those evils of working at a huge company. Then I started to notice that people were just doing weird stuff. Calling pointless meetings, making big stinks over small things, and just doing weird things just to create more work for themselves. I realized all of that was just so people could have more bullet points on their list of accomplishments at the end of the year.
Sure, killing the curve won't solve everything and I think there is a lot of truth to what you're saying about not valuing teamwork. I just think that killing the stack ranking is a BIG step in the right direction.
This reminds me very much of the article the other day, "You're only getting the nice feedback": https://news.ycombinator.com/item?id=6712289
Sure, that's plenty true, but it's one thing to have a small incidence of herpes and another thing entirely to have a pandemic of HIV or syphilis.
"Some countries initially felt they were isolated from AIDS, but now they realize there is no such thing. There is no border, no boundary. We've learned that walls endanger because they encourage a false sense of security. Even if you could impose the perfect program for screening international travellers, infected people will get in, and some of your own citizens will come back infected. In the meantime, people won't follow safe sex practices, because they figure they're protected inside their walls. (...)
One of the major barriers we face in trying to get countries to deal effectively with AIDS is the tremendous gap between social myth and social reality. I think closing this gap is a useful step. It's important to deal with things as they are, and not as somebody would like them to be. (...)
I have asked a lot of government officials and experts, 'At what age do young men and women in your society begin to have sexual intercourse?' This is not prurient curiosity on my part. I'm trying to figure out when you might start certain kinds of educational programs. The expert thinks a minute. He may take on a reflective look as he considers his own adolescence, and he makes a decision. Is he going to tell me when he had sexual intercourse, or when he should have had sexual intercourse? The answers are not at all scientific, and frankly, people often don't know what's going on in their own society in terms of sexual practices and drug use."
- Jonathan Mann, director of the Global Program on AIDS, from the book: "Reinventing the Future: Conversations with the World's Leading Scientists" by Thomas Bass, 1994
Great book. But the (somewhat paraphrased) quote fits rather nicely if you replace screening with "hiring process" and "career advancement" with "sexual practices".
How do we want our organizations to work? What should the life-cycle of an employee be? Can we state that frankly in a way that's both good for the company and good for the employees? Where do managers go as they grow older? Up or out? If up, where does the CEO go?
So why embrace a process that seems to celebrate daring? A move-fast-and-break-things approach to projects? You'd think MS would be an analysis-paralysis company, because every MS product needs to integrate with hundreds of other MS products and be supported on oodles of platforms for like a decade.
But instead? We see this crazy wasteful churn of APIs and packages and MSDN Magazine-driven fashions.
I mean, how long will MS be stuck supporting Silverlight? Or Linq2SQL? Or ClickOnce installers? Or the zillion different SKUs of SQL Server and Access?
Also, the new powershell ISE is slick (reminds me of Python), but at the same time the near-total lack of integration between Powershell and Visual Studio is pretty surprising. I generally find Powershell a bit disappointing... it feels kind of messy next to C# or Python or plain command-line. I'd probably rather just have a C# REPL with good intellisense. I'm sure one exists.
Imagine this experiment: you have a great, productive team who has one member they all hate (and therefore assume isn't doing much for the team.) You take that person off the team... and the team's productivity plummets.
Any time the best way for me to get ahead is by using "The Art of War" on my peers you have set up your organization for eventual failure.
Cooperation and team work are also meritorious, but they are not rewarded or encouraged in this type of merit-based system.
The idea that "there's a fixed budget for raises" or "that money goes into a different pot" is simply a way of refusing while transferring blame to an inanimate object.
It honestly surprises me that MS gets away with having such poor working conditions and team work [1]. I guess the end of stack ranking shows that things are changing at least.
I know lots of people in IT, but somehow no one who works at MS [2]. Its seen as uncool or worse unmoral in my circles [3]. And these people, including me, didn't even know how bad the working conditions are at MS! I always thought of MS to be like one of those 'defense' companies [4]. You either don't really want to work there, because even if the technical challenge and the available resources are great, you don't agree with their mission or you fear that your friends and peers would look down on you. But to make up for that, they have really good work conditions [5]. Thats at least what I hear from people who worked on military projects at Raytheon, Boeing and EADS.
Hope you don't take this comparison as an insult. I think MS makes great things! OneNote, VisualStudio, Kinect, Age of Empires are all great products. And personally, I would have much, much (!!!) less reservations against working for them, than a so called defense company [6].
--- [1] And, I can't help myself pointing out, that MS's version control system is called 'Team Foundation'. New speak?
[2] The exception is MS Research, which is awesome, but thats not really MS.
[3] Most of us were born in the 80's (and many including me in Germany) and grew up hating MS. Even if you wouldn't use computers on your own, everyone including parents and teachers would tell you how bad MS, their products and their business practice were.
[4] Other examples are 'big pharma', tabaco, finances, esp. investment banking. I know people working in all of those industries. They all report excellent work conditions and compensations. Some are ok with the company actions it, some are proud to work there. But all of them look very self aware when they get asked "What do you work?" The counterexample are jobs like social work, medical care taking, NGOs, academic research and yes, usually engineering as well. Many people want to work their, either because they believe in it, or because it brings them social prestige. As a result the high demand allows for poorer working conditions and less pay.
[5] Salaries at MS are still very good, as far as I know.
[6] Though, DARPA projects are very attractive. I can't deny that.
I work in dev in Redmond. All I can say is, for me, it really is a wonderful fun place to work and not at all like the outside impressions.
(I have a friend who works at BofA. They wanted him to take on a new project that he wasn't interested in. As a reward for agreeing, they went back to his old performance reviews and changed them from "Meets expectations" to "Exceeds expectations" and then promoted him for excellent performance. LOL.)
Clearly there is a curve that when applied to an entire population shows that there are under-performers. So far, so good, however, the problem is that stack ranking isn't applied to the entire population since it's impossible to normalize the performance of each individual. Instead, companies that stack rank do it on a team-by-team basis, which is a statistically inaccurate sample of the population where there can be a lot of skew.
Since it's up to the manager to rank their employees, this causes a lot of problems. This is particularly a problem if the manager sucks at hiring, or over hires for their team, and there is a lot of chaff. There is a massive incentive to do this, because the more employees a manager has, the more corresponding perceived worth in the organization they'll have. Plus, if it comes time to cull, it's really easy to shield the favourite employees, even if they have performed poorly, without disrupting the organization.
On the flip side, if you're really, really careful about who you bring on board, your team is in a weaker position because there is no one to terminate. This means you get rid of good performers based on bullshit metrics. I saw this happen at a largish virtualization company where one of the best employees on our team was let go because he had some minor HR-ish type issue. This caused a big percentage of the team to be demoralized and ultimately leave.
Is this not in effect the same as a curve? Giving higher bonuses to one person means giving lower bonuses to another. Giving everyone the same bonus discourages high performers from staying (and also working on the same team). People who receive little to no bonuses take the hint to leave, just as if they had received a 4 or a 5. Considering only 10% of employees receive a 4, and 5% receive a 5, I'm thinking that effectively, there has been no change.
A healthy organization will always "manage out" people who are not fit for the job. At Microsoft, folks are given the hint early enough that they can explore other internal options before leaving the company entirely.
Though, to be honest, if someone just can't cut it, they will leave the company.
Except that it doesn't work. No internal hiring manager will ever allow the interview loop/transfer for the internal candidate with the score 4 or 5 (I think those are the lowest scores in the current model:-). In 99.9% of cases those unlucky who got 5 get laid off by the next review period.
Maybe if they fire all middle and upper management.
If two people are both performing at about the same level, they can get the exact same bonus. In a ranked system one employee must get more than the other, even though their actual output is pretty close to identical.
Not to mention when you make the ratings tied to rewards, as opposed to punishments, you're removing a lot of politics. Stack ranking discourages high performers from working together, as their performance rating is penalized by their proximity to smart people. Because stack ranking is inherently punitive, this also makes these employees less mobile. So not only are you stuck in a team full of super-ninjas who make you look bad, but the stack rank you got makes you toxic when it comes to internal transfers.
A fixed bonus pool won't fix the "everyone near me is too smart" problem, but it at least would give people a way out to another team. An otherwise good employee who rolled the dice wrong when picking teams does not gain a permanent black mark for it.
I seem to think it's much easier to convince them your awesome team producing awesome products (and taking in awesome money) should get extra money to spend than it is to convince people you shouldn't have to rank any of your employees lowly when they all have to.
I do work for you, I am owed compensation based on terms we've negotiated before I started working for you. From time to time these terms will come up for renegotiation.
There is no need for compensation in excess of what we've negotiated. Sure, I am not one to turn down free money, but understand that this does not affect our normal compensation discussions.
My work is worth $X on the market. I expect that you will pay me $X (or come up with alternative forms of compensation that are mutually agreeable). I will not sign a deal that adds up to $X where part of the compensation is discretionary. My work is worth what it's worth. If it is not up to your satisfaction, you are free to terminate this relationship or renegotiate my pay. You may not unilaterally decide to pay me less than what we agreed upon, and I will not submit to a system that allows this.
Which is to say, you will be unsuccessful if you try to persuade me to lower my base salary on the promise of "bonuses".
I feel that "bonuses" are just ways to hoodwink people who haven't been around the block. Every time I have come across bonuses of this variety, it has been an excuse to lower your base salary. I have never once received the full possible bonus, even though I have been promised multiple times prior to joining the company that "everyone" gets the full amount, and that only very poor performers don't. I have never received a poor performance review.
If you do not pay me $X, whether in salary or in bonuses or in suitcases of bacon, I will leave. I do not give one flying hoot about stack ranks, bonus pools, or any such nonsense. I am not willing to inject uncertainty into my income without a proportional upside for myself.
Discretionary bonuses are full of shit.
Whether it actually does this well is certainly an important question, as well as how it can be abused and how frequently that happens, but it seems remiss to rail against them without noting their intended purpose - which, in the abstract, seems entirely legitimate.
If I am not performing to expectations, fire me or renegotiate my salary. There are a great many ways to tune my compensation to my actual determined worth.
A discretionary bonus in this case is simply "you should take a below-market salary until we're convinced you're the real deal, and then maybe you will get the salary difference back". I'm not convinced anyone who isn't desperate will take this deal.
And having experienced the discretionary bonuses at two very large and well-known software companies, I am absolutely, completely, vehemently disinclined to ever enter into such an arrangement again.
For discretionary bonuses to make even the most remote amount of sense, it needs to offer the employee an upside for taking on the uncertainty.
Many companies give you a market rate and a bonus (sign on / stock RSU/ cash bonus etc) on top of it to incentivize extra work / ownership.
My case: my base salary is pretty competitive considering the market/location. I also get a stock award on top of that based on my future potential / current work etc. For me, the stock component has been a reasonable contributor to the final net take home.
Essentially: you work hard, you can potentially make 2x-3x your base and this multiplier only goes up as you go up levels.
I am not against discretionary bonuses like the above (obviously :). Come for the base, toil for the bonus.
Both of these are high overhead, which means harder to tune. Depending on circumstance, that may work to the advantage of the employer or employee, or to the advantage or disadvantage of both.
In principle it needn't require desperation, just confidence that your market salary should be above what you can demonstrate it should be during the interview process.
Again, I'm not speaking to actual practicality - I know in a limited fashion how it has and hasn't worked for me, where discretionary bonuses have occasionally been present but have generally not been a large part of my compensation; I don't have any particular insight into their function / misfunction broadly.
"For discretionary bonuses to make even the most remote amount of sense, it needs to offer the employee an upside for taking on the uncertainty."
Obviously.
Anyone saying "Base + max bonus gives the compensation you expect" is trying to take advantage. It should be "base + expected bonus is marginally higher than the fixed rate you would deserve if you are as good as you think".
It's both a carrot and a stick for the company over the employee.
The forced curve before meant that every team needed to assign x% as high performers and x% as low performers.
Now, how you assign the high and low performers, from a limited pot of 'rewards', when every manager is going to want most of his team to be considered high performers -- well, that's what the 'forced curve' was intended to deal with. But I don't doubt it also led to increased competition and politics and decreased teamwork, as the one well-known critique of it suggested.
If it's the former, you're right and the details of how individual bonuses are assigned don't matter much.
The latter however would could up real opportunities: a team that gets together and delivers beyond expectations could all be generously rewarded instead of having the weird fight over who was the individual star performer that gets all the credit.
I think what they've done here is to allow each manager in the chain to decide how to handle their organization's evaluation and compensation. You could be VP of a 500-person division and instruct your directors and managers to rank everyone in your org in a single pool; or you could delegate to your directors and tell them to do what they think is best for their groups within their own individual budget constraints, which you assigned.
But the truth is that when you assemble such a large group, you end up comparing apples to oranges. Jane works on a doomed software project and valiantly manages to salvage some of it for other efforts, but management ultimately looks at her project as a failure and has trouble disassociating her work from the result. Meanwhile Bob works on a fairly straightforward hardware project that gets completed on time and within budget, but not necessarily as a result of his actions, and looks like a "team player" who gets rewarded as such.
Assigning a single scalar value ("top $x%") to each of these people necessitates ignoring a huge number of important differences both personal and contextual. Small groups with similar functions allow you to make more meaningful comparisons but only at the cost of ruining the statistical basis for making those comparisons in the first place.
The whole problem with forcing a predetermined model on your organization is that performance appraisals should be an empirical process where the observations peers and management make determine the model that is ultimately adopted (and which will guide future hiring/firing decisions). Forcing a predetermined model onto your organization gives you no data and leads to a self-fulfilling prophecy at best and corruption and politicking at worst.
And that, in turn, is created by management's inability to trust in their own abilities. This is proved to be true in case of Ballmer and overall Microsoft's performance while under his leadership.
These changes are surely a positive sign for Microsoft.
I believe that this is also the effect of working for a public company with a stock that hasn't moved in price in 10 years.
On the other hand, the lower 5% people only cost about 2% of companies spending. It is not even worth to worry about it. You need to interview equivalent to 50% of employee count to replace that 5%, plus whole set of HR bandwidth to handle it. Microsoft has been losing a lot of value to fire lower 5% while keep a lower 5% CEO at the top for so many years.
Interesting. I would refer to those as countries with "weaker" or "perverse" employment laws...
Take for example working hours. A country with strong labor laws will usually prohibit employers from making employees work more than x hours a week, and if they do that they have to pay for overtime. That will make some people happy and productive because they will feel safe having protection from the system. But some people will abuse it and slack off.
On the other hand, if you don't have control over that you might end up in a situation where people are only considered good employees and productive if they work crazy hours all the time. Instead of being abused by employees, the system will be abused by the people at the top.
And I guess that covers underperformance. Who defines what bad/average/good performance is? Is the guy who does really good work considered to be underperforming just because he works 9-5 and wants to spend quality time with his family instead of making the office his second home? That kind of thing has to be prevented by law.
- Difficulty knowing who the best employees are.
- Employees surprised by layoffs. "I thought I was doing well."
- Managers slacking on giving constructive feedback.
Performance management is such a tough subject. Both extremes (ruthless stacking, and no-curve) have issues and I don't know a better solution.
The risk going forward is that without being forced to a distribution, wishy-washy managers will trend towards the middle. High achievers will be under-compensated, and under-achievers will be overcompensated. The budget hasn't changed, so the same pool of money has to be divided among the same set of people, managers now have more freedom as to how.
The compensation budget is the key part here. I don't know how this is going to account for:
- Failed products with smart developers and their retention. - Avg products with very smart developers - Less 'visible' projects. Those that management don't realize is necessary until you have no one working on it. (This happens a LOT in big companies).
All this can be solved by pumping a ton of money into this as they are alluding elsewhere though.
Overall the fact that they are taking feedback seriously and are working towards a solution is pretty positive though.
http://blog.learningbyshipping.com/2013/11/09/realities-of-p...
But, I think that leadership/true emotional control means that, even if someone is treating you wrongly or criticizing you, you can dissect the behavior, and then appeal to reason.
For example, I appreciate, and even invite, ruthless criticism ("go ahead and insult me if necessary") when getting feedback (on my code, performance, etc.)
The importance of having friends who can say, "you're an idiot - you don't understand X, Y, Z" is high.
Sometimes I dislike when, at previous jobs, I've had to mask the truth simply to avoid hurting someone's feelings. Because, if you hurt someone's feelings, that person will close up to you.
Just because now 30% of the staff can fall into the Exceed Expectations bucket doesn't necessarily mean that the budget increases. The money gets divided differently, that's all.
It does seem like a positive change though if they can improve the culture of the workplace and minimize the amount of backstabbing, cronyism and political nonsense.
One should look at where this revenue was coming from and compare GE to other companies in the sector (i.e., other makers of nuclear reactors and jet engines) during this time before drawing conclusions about personnel management theories (of all things).
In terms of hard science, unfortunately blind randomized controlled trials are impossible, so it's not possible to definitively isolate causal relationships and confounding factors. "It seemed to work for GE in the 1980s" is the best that's on offer.