5 ms·

I think there are also two different "flavors" of stack ranking usually at play. There is the one where the company wants to reward top performers. I think thi

by qqtt 4y ago

I think there are also two different "flavors" of stack ranking usually at play. There is the one where the company wants to reward top performers. I think this one is most common, and I have never worked at a company that doesn't so some form of bonus budgeting that tries to capture a group of employees who are performing well and give them a bonus. The budget itself dictates how many employees can fit into that group, so ultimately what happens is a natural bell curve of the biggest rewards going to the perceived top performers, as agreed upon by managers usually at the organization level. The other more insidious stack ranking involves a punitive element, whereby a group must necessarily be singled out for performance improvement reasons. This is the type of thing where managers, no matter what the team members actually accomplished, have to put someone on a performance improvement plan or start performance managing them to "get them up the bar" because someone naturally has to fall into the "bottom performer" bucket of stack ranking. I think the former is common and pretty unavoidable unless a company has an unusually large bonus budget pool available. The latter approach is what causes problems with companies, because all of a sudden team mates become competitors and cultures can go downhill rapidly. Think Amazon, Microsoft under Ballmer, and GE under Walsh. It's kind of interesting how many companies who are doing the former bonus budget pooling eventually get seduced into doing the latter and destroying their culture.

TeMPOraL 4y ago
I've seen this play out at my university in two different ways. The more obvious way is very obvious: we had a professor who told us he'll be grading us by first tallying up the answers, and then applying a bell curve to the scores, so that X% of the students get an A, Y% get a B, etc., and some Z% of the students will fail. As you can imagine, everyone was up in arms about how obviously unfair this is. The less obvious case, but close to what you describe, was scolarships. For almost every "track"[0], scolarships were awarded for top X% of students. In the "track" I was in, and another one in our faculty, the scolarships were instead given to anyone who crossed a certain absolute grade threshold, in (IIRC three) tiers. The higher the tier you reached, the more money you got, independent of how other students fared (probably subject to budget constraints, though). Now, I can't prove it, but there was one other notable difference between our "track" and that other one with same scolarship model, and every other "track" at the university I was aware of: ours were famously cooperative, everyone else was fiercely competitive. I don't think it's a coincidence. My university experience was that of everyone helping each other out, because it was always a win-win both socially sometimes financially. Everywhere else, this was unthinkable - helping other students created a risk of them getting ahead of you and taking your scolarship. I'm dismayed to see the same thing repeating in companies too. -- [0] - I can't seem to find the correct word to describe the set of students across the years who pursue a degree in a particular subject (e.g. "Computer Science") on a specific faculty. I.e. the vertical slice.
- abdullahkhalids 4y ago
  'Batch' is the term commonly used for the set of students who start a program at the same time. For instance, the CS MS batch of 2023, who are expected to finish their MS in CS in 2023 (So they all should have stared in 2021 assuming a two-year program).
  - oso2k 4y ago
    Do you mean cohort or class? MBA programs in the US will use those terms.
    - urthor 4y ago
      Those are far more common.
    - drblastoff 4y ago
      I heard “batch” used often in grad school by my non-American classmates, but “class” seems far more common in the U.S.
    - radicalbyte 4y ago
      I've heard cohort, batch, class and year used for both primary, secondary and higher education.
  - TeMPOraL 4y ago
    I thought it was just "class", as in "CS class of 2023"? The term I'm looking for, though, is the set of batches abstracted over time. That is, all students who take CS at particular faculty of particular university, regardless of the year they started. A sequence of "batches". Such students will share curriculum and specific teaching staff. This is in contrast to another such group that e.g. does Applied CS at the same faculty, or Theoretical Physics at different faculty. EDIT: perhaps a picture is in order. Imagine an university tallying up how many students started a given class on a given year, per class, across entire university. They could create a spreadsheet that looks like: | University of Example | |----------+---------------+-----------------------+-----------------| | | Division of | Division of | Division of | | | Telecoms & EE | Robotics & Bio | Social sciences | |----------+-------+-------+-------+-------+-------+-------|---------| | Class of | Appld | Tele | Biomd | Appld | Auto | Social| ... | | ↓ | CS | comm | eng | CS | matics| media | ... | | 2022 | ### | ### | ### | ### | ### | ### | ... | | 2021 | ### | ### | ### | ### | ### | ### | ... | | 2020 | ### | ### | ### | ### | ### | ### | ... | | 2019 | ### | ### | ### | ### | ### | ### | ... | | ... | ... | ... | ... | ... | ... | ... | ... | The term I'm looking for is the one you would use to call a single column of ### here - e.g. "all classes (years) of Applied CS in Division of Robotics and Bioinformatics".
sph 4y ago
They tend to behave exactly the same though. If you rank only good performers, it's a small step before you use that against bad performers: "you've been with us 5 years and never managed to get on the top performers list. I am afraid you are not a good fit for our culture here"
- bluefirebrand 4y ago
  It's very easy to see how people might make this logical jump, but honestly if your bosses manage you out for never being a top performer they are doing you a favor: you don't have to work for their dumb asses anymore. Good managers recognize the value of role players. Not everyone can be a top performer. It's an absurd expectation to have for any organization.
kqr 4y ago
> I think the former is common and pretty unavoidable unless a company has an unusually large bonus budget pool available. The problem with this is not the size of the bonus pool. The immediate problem is the thresholding. Due to the sharp threshold, the bonus will practically never stand in proportion to performance. An employee that is judged to perform at a level of 6.1/7 gets nothing, but the one at 6.3/7 gets a lot. This makes the scheme anti-collaborative. The less obvious problem is the ridiculousness of the idea that you can assign accurate individual performance numbers to people. These things are insanely noisy to the point of being meaningless. It ends up effectively a lottery. That would be fine if you called it a lottery, but by pretending it's not you create perverse incentives. There are only two ways to do bonuses well: - either forego individual bonuses and reward the entire company equally (or maybe scaled by salary) based on company performance, or - hand out individual bonuses only to true (statistically verified) outperformers -- in this case you might end up not handing out any bonuses most years.
- 2b3a51 4y ago
  "These things are insanely noisy to the point of being meaningless." Goldstein and Spiegelhalter did a paper on league tables for organisations (e.g. schools and hospitals) along the lines that noise in the data made the results questionable.[1] Perhaps the idea transfers to individuals within an organisation that uses stack ranking? [1] https://www.bristol.ac.uk/media-library/sites/cmm/migrated/documents/statistical-issues-for-league-tables1.pdf https://www.bristol.ac.uk/media-library/sites/cmm/migrated/d...
raxxorraxor 4y ago
I believe evaluating software development only works at the bottom line of tasks with formal metrics like issues, bugs or tickets solved. Also in general only possible in larger teams and probably more for maintenance tasks than active development. There are no boni for top performers in my company at all. If business is good everyone gets a bonus. The only exception here is sales. Here people are more competitive. I guess I would get martyred if I even suggested it for other departments. And I believe it would be pretty bad for morale and teamwork overall. It is good that people don't use energy to present themselves in a good light and instead solve the problems we have. I still believe there is a lot of incentive for employees to see the company performing well. As you said, I have the impression that this becomes a background issue if employees compete with each other. There was a time I would love to have worked at Blizzard, even while being aware of the volatility and risk of the industry. I wouldn't want that today regardless of compensation. Still, I don't believe it is either or. Both systems you suggest probably have an overall net negative effect for development. There are other forms for employees to participate on company success. Administration for employee boni is wasteful overhead for the most part. If it is true what the article reports, I believe the manager did the correct thing. On that metric he might be a high performer among his peers.
snapcaster 4y ago
minor nitpick: I think you meant "welch" for GE

by qqtt 4y ago

TeMPOraL 4y ago

I've seen this play out at my university in two different ways. The more obvious way is very obvious: we had a professor who told us he'll be grading us by first tallying up the answers, and then applying a bell curve to the scores, so that X% of the students get an A, Y% get a B, etc., and some Z% of the students will fail. As you can imagine, everyone was up in arms about how obviously unfair this is. The less obvious case, but close to what you describe, was scolarships. For almost every "track"[0], scolarships were awarded for top X% of students. In the "track" I was in, and another one in our faculty, the scolarships were instead given to anyone who crossed a certain absolute grade threshold, in (IIRC three) tiers. The higher the tier you reached, the more money you got, independent of how other students fared (probably subject to budget constraints, though). Now, I can't prove it, but there was one other notable difference between our "track" and that other one with same scolarship model, and every other "track" at the university I was aware of: ours were famously cooperative, everyone else was fiercely competitive. I don't think it's a coincidence. My university experience was that of everyone helping each other out, because it was always a win-win both socially sometimes financially. Everywhere else, this was unthinkable - helping other students created a risk of them getting ahead of you and taking your scolarship. I'm dismayed to see the same thing repeating in companies too. -- [0] - I can't seem to find the correct word to describe the set of students across the years who pursue a degree in a particular subject (e.g. "Computer Science") on a specific faculty. I.e. the vertical slice.

abdullahkhalids 4y ago
'Batch' is the term commonly used for the set of students who start a program at the same time. For instance, the CS MS batch of 2023, who are expected to finish their MS in CS in 2023 (So they all should have stared in 2021 assuming a two-year program).
- oso2k 4y ago
  Do you mean cohort or class? MBA programs in the US will use those terms.
  - urthor 4y ago
    Those are far more common.
  - drblastoff 4y ago
    I heard “batch” used often in grad school by my non-American classmates, but “class” seems far more common in the U.S.
  - radicalbyte 4y ago
    I've heard cohort, batch, class and year used for both primary, secondary and higher education.
- TeMPOraL 4y ago
  I thought it was just "class", as in "CS class of 2023"? The term I'm looking for, though, is the set of batches abstracted over time. That is, all students who take CS at particular faculty of particular university, regardless of the year they started. A sequence of "batches". Such students will share curriculum and specific teaching staff. This is in contrast to another such group that e.g. does Applied CS at the same faculty, or Theoretical Physics at different faculty. EDIT: perhaps a picture is in order. Imagine an university tallying up how many students started a given class on a given year, per class, across entire university. They could create a spreadsheet that looks like: | University of Example | |----------+---------------+-----------------------+-----------------| | | Division of | Division of | Division of | | | Telecoms & EE | Robotics & Bio | Social sciences | |----------+-------+-------+-------+-------+-------+-------|---------| | Class of | Appld | Tele | Biomd | Appld | Auto | Social| ... | | ↓ | CS | comm | eng | CS | matics| media | ... | | 2022 | ### | ### | ### | ### | ### | ### | ... | | 2021 | ### | ### | ### | ### | ### | ### | ... | | 2020 | ### | ### | ### | ### | ### | ### | ... | | 2019 | ### | ### | ### | ### | ### | ### | ... | | ... | ... | ... | ... | ... | ... | ... | ... | The term I'm looking for is the one you would use to call a single column of ### here - e.g. "all classes (years) of Applied CS in Division of Robotics and Bioinformatics".

sph 4y ago

They tend to behave exactly the same though. If you rank only good performers, it's a small step before you use that against bad performers: "you've been with us 5 years and never managed to get on the top performers list. I am afraid you are not a good fit for our culture here"

bluefirebrand 4y ago
It's very easy to see how people might make this logical jump, but honestly if your bosses manage you out for never being a top performer they are doing you a favor: you don't have to work for their dumb asses anymore. Good managers recognize the value of role players. Not everyone can be a top performer. It's an absurd expectation to have for any organization.

kqr 4y ago

> I think the former is common and pretty unavoidable unless a company has an unusually large bonus budget pool available. The problem with this is not the size of the bonus pool. The immediate problem is the thresholding. Due to the sharp threshold, the bonus will practically never stand in proportion to performance. An employee that is judged to perform at a level of 6.1/7 gets nothing, but the one at 6.3/7 gets a lot. This makes the scheme anti-collaborative. The less obvious problem is the ridiculousness of the idea that you can assign accurate individual performance numbers to people. These things are insanely noisy to the point of being meaningless. It ends up effectively a lottery. That would be fine if you called it a lottery, but by pretending it's not you create perverse incentives. There are only two ways to do bonuses well: - either forego individual bonuses and reward the entire company equally (or maybe scaled by salary) based on company performance, or - hand out individual bonuses only to true (statistically verified) outperformers -- in this case you might end up not handing out any bonuses most years.

2b3a51 4y ago
"These things are insanely noisy to the point of being meaningless." Goldstein and Spiegelhalter did a paper on league tables for organisations (e.g. schools and hospitals) along the lines that noise in the data made the results questionable.[1] Perhaps the idea transfers to individuals within an organisation that uses stack ranking? [1] https://www.bristol.ac.uk/media-library/sites/cmm/migrated/documents/statistical-issues-for-league-tables1.pdf https://www.bristol.ac.uk/media-library/sites/cmm/migrated/d...

raxxorraxor 4y ago

I believe evaluating software development only works at the bottom line of tasks with formal metrics like issues, bugs or tickets solved. Also in general only possible in larger teams and probably more for maintenance tasks than active development. There are no boni for top performers in my company at all. If business is good everyone gets a bonus. The only exception here is sales. Here people are more competitive. I guess I would get martyred if I even suggested it for other departments. And I believe it would be pretty bad for morale and teamwork overall. It is good that people don't use energy to present themselves in a good light and instead solve the problems we have. I still believe there is a lot of incentive for employees to see the company performing well. As you said, I have the impression that this becomes a background issue if employees compete with each other. There was a time I would love to have worked at Blizzard, even while being aware of the volatility and risk of the industry. I wouldn't want that today regardless of compensation. Still, I don't believe it is either or. Both systems you suggest probably have an overall net negative effect for development. There are other forms for employees to participate on company success. Administration for employee boni is wasteful overhead for the most part. If it is true what the article reports, I believe the manager did the correct thing. On that metric he might be a high performer among his peers.

snapcaster 4y ago

minor nitpick: I think you meant "welch" for GE